A Quantitative Analysis of the Effect of Market Design and Policy Uncertainty on Investment in Electricity Generation: A Reinforcement Learning Approach
Abstract
Evidence exists that electric market design and policy uncertainty significantly impact long-run electric generation investment. This research quantifies this relationship and provides policy makers with insights into the long-run implications of several proposed policies. It utilizes an innovative modeling technique to address the problem of modeling sequential investment under uncertainty. The first essay introduces a modeling framework that utilizes reinforcement learning (RL)-a recently developed technique for solving stochastic control problems-to model optimal long-run generation investment from both social welfare maximizing and monopolistic perspectives. This essay demonstrates that this technique can produce more realistic models of investment under uncertainty than other stochastic control methods because explicit definition of state transition probabilities is not required. The second essay utilizes the framework presented to determine the effect of capacity subsidies and price caps on investment and prices. Results show that capacity subsidies act to increase overall investment while reducing spot market price volatility. However, this policy increases total electricity prices once capacity charges are considered. The third essay uses the RL framework to investigate the manner in which policy uncertainty, relating to the enactment or repeal of investment tax credits (ITCs) and production tax credits (PTCs), impacts investment in wind power. These differing responses to uncertain tax policy result from the fundamental characteristics of the policies. Those policies that reward firms based on the year of a specific investment will produce near-term investment results that are opposite in direction to the intended result of the proposed change. Also, since substitution opportunities exist between wind and classical technology investments, the investment postponing and enhancing effects of ITC expectation are stronger than those found in previous researcc
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 17, 2000
- Accession Number
- ADA379983
Entities
People
- Jeffrey H. Grobman
Organizations
- Air Force Institute of Technology