Electricity generation and emissions reduction decisions under uncertainty : a general equilibrium analysis
Author(s)Morris, Jennifer F. (Jennifer Faye)
Massachusetts Institute of Technology. Engineering Systems Division.
Mort D. Webster.
MetadataShow full item record
The electric power sector, which accounts for approximately 40% of U.S. carbon dioxide emissions, will be a critical component of any policy the U.S. government pursues to confront climate change. In the context of uncertainty in future policy limiting emissions and future technology costs, society faces the following question: What should the electricity mix we build in the next decade look like? We can continue to focus on conventional generation or invest in low-carbon technologies. There is no obvious answer without explicitly considering the risks created by uncertainty. This research investigates socially optimal near-term electricity investment decisions under uncertainty in future policy and technology costs. It employs a novel framework that models decision-making under uncertainty with learning in an economy-wide setting that can measure social welfare impacts. Specifically, a computable general equilibrium (CGE) model is formulated as a two-stage stochastic dynamic program focused on decisions in the electric power sector. The new model is applied to investigate a number of factors affecting optimal near-term electricity investments: (1) policy uncertainty, (2) expansion rate limits on low-carbon generation, (3) low-carbon technology cost uncertainty, (4) technological learning (i.e., near-term investment lowers the expected future technology cost), and (5) the inclusion of a safety valve in future policy which allows the emissions cap to be exceeded, but at a cost. In modeling decision-making under uncertainty, an optimal electricity investment hedging strategy is identified. Given the experimental design, the optimal hedging strategy reduces the expected policy costs by over 50% compared to a strategy derived using the expected value for the uncertain parameter; and by 12-400% compared to strategies developed under a perfect foresight or myopic framework. This research also shows that uncertainty has a cost, beyond the cost of meeting a policy. In the experimental design used here, uncertainty in the future policy increases the expected cost of policy by over 45%. If political consensus can be reached and the climate science uncertainties resolved, setting clear, long-term policies can minimize expected policy costs. In addition, this work contributes to the learning-by-doing literature by presenting a stochastic formulation of technological learning in which near-term investments in a technology affect the probability distribution of the future cost of that technology. Results using this formulation demonstrate that learning rates lower than those found in the literature can lead to significant additional near-term investment in low-carbon technology in order to lower the expected future cost of the technology in case a stringent policy is adopted.Ultimately, this dissertation demonstrates that near-term investments in low-carbon technologies should be greater than what would be justified to meet near-term goals alone. Near-term low-carbon investments can lower the expected cost of future policy by developing a less carbon-intensive electricity mix, spreading the burden of emissions reductions over time, helping to overcome technology expansion rate constraints, and reducing the expected future cost of low-carbon technologies-all of which provide future flexibility in meeting a policy. The additional near-term cost of low-carbon investments is justified by the future flexibility that such investments create. The value of this flexibility is only explicitly considered in the context of decision-making under uncertainty.
Thesis (Ph. D.)--Massachusetts Institute of Technology, Engineering Systems Division, 2013.This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections.Cataloged from student-submitted PDF version of thesis.Includes bibliographical references (pages 169-183).
DepartmentMassachusetts Institute of Technology. Engineering Systems Division.; Massachusetts Institute of Technology. Engineering Systems Division
Massachusetts Institute of Technology
Engineering Systems Division.