INFORMS Open Forum

Next ENRE online event 20 May: Erin Baker & Franklyn Kanyako on decision making under deep uncertainty

  • 1.  Next ENRE online event 20 May: Erin Baker & Franklyn Kanyako on decision making under deep uncertainty

    Posted 05-13-2021 06:39
    Dear colleagues,

    Please join us for the next ENRE online event on Thursday the 20th of May starting at 15:00 BST (UTC+1); see below for more details.

    Please feel free to forward this invitation to anyone who might be interested. For more information and recordings of past seminars, see
    https://blogs.ed.ac.uk/enre/

    We hope to see many of you at the event,
    Miguel Anjos, Harry van der Weijde, and Lars Schewe

    =====================================================

    20 May 2021 15:00-16:30 BST (UTC+1) (on Zoom) | Erin Baker (University of Massachusetts, Amherst) & Franklyn Kanyako (University of Massachusetts, Amherst & Tufts University)
    Link: https://ed-ac-uk.zoom.us/j/88097951730 (Meeting ID: 880 9795 1730, Passcode: enre1234)

    Multiple beliefs, dominance, and dynamic consistency (Erin Baker)
    Abstract: Inspired by challenges in designing climate change policy, we address the problem of decision making under "deep uncertainty." We investigate the characteristics of Belief Dominance, a prescriptive operationalization of a concept that has appeared in the literature under a number of names. Belief dominance supports decision-making under multiple characterizations of uncertainty by ruling out strategies that are dominated across a set of beliefs. We present a proof-of-concept application aimed at informing decisions over investments into clean energy technology R&D portfolios in the context of climate change and illustrate how this framework helps identify robust individual investments. We go on to explore the dynamic consistency of this dominance concept. We show that Belief Dominance is less dynamically consistent than subjective expected utility, but more dynamically consistent than non-expected-utility decision rules, such as minmax. We illustrate these concepts using a numerical example addressing climate policy under ambiguity.

    Low carbon energy R&D portfolios that are robust to multiple beliefs and multiple models (Franklyn Kanyako)
    Abstract: We identify low carbon R&D portfolios that are robust to two distinct types of deep uncertainty: parametric and structural uncertainty. By deep uncertainty, we mean the case where experts' disagree over (1) the probability distributions for important parameters, in this case the costs and efficiencies of energy technologies in response to R&D (parametric uncertainty); and (2) over causal chains, represented by the many different models used to understand, analyze, and assess climate change causes and effects (structural). We expand on Belief Dominance, identifying all portfolios of R&D investment that are non-dominated across a plausible set of probability distributions and models. We focus on the case of a $125/tCO2 tax on emissions. We find common ground among the expert beliefs and the models, indicating that a high investment in Bioelectricity and Solar are robust to all the beliefs and models, given the climate policy. We show that the disagreement among models is most significant for R&D investments in nuclear.

    ------------------------------
    Miguel F. Anjos, Ph.D., P.Eng., FHEA, SMIEEE, FEUROPT, FCAE
    President, INFORMS Section on Energy, Natural Resources, and the Environment (ENRE)
    Chair of Operational Research, School of Mathematics, University of Edinburgh, U.K.
    Inria International Chair
    Schöller Senior Fellow 2020
    http://www.miguelanjos.com/
    ------------------------------