UC Berkeley

This dissertation studies two topics in Demand Response (DR) in electricity markets, with some discussion of retail electricity pricing more broadly. In each of these investigations we posit a model of a consumer, or population of consumers, optimizing their consumption decisions for their private benefit. The first investigation considers the profit maximization problem of a DR aggregator, and the second studies the welfare impacts of existing and hypothetical retail tariffs and DR programs, with a combination of theoretical analysis and simulation experiments.

Part I provides a comprehensive introduction to the dissertation.

Part II of the dissertation formulates and analyzes the profit maximization problem of an aggregator that owns the production rights to a Variable Energy Resource's (VER) output, and also signs contracts with a population of DR participants for the right to curtail them in real time, according to a contractually specified probability distribution. The aggregator is situated in a market environment in which it bids a day-ahead offer into the wholesale market, and is penalized for deviations of its realized net production--renewable energy bundled with DR--from that offer. We consider the optimization of the aggregator's end-to-end problem: designing the menu of DR service contracts using contract theory, bidding into the wholesale market, and dispatching DR consistently with the contractual agreements. In our setting, DR participants have private information about their valuation for energy; and wholesale market prices, VER output, and participant demand are all stochastic, and possibly correlated.

In Part III, we study the welfare effects of various dynamic electricity pricing schemes, including Real-Time pricing, Time-of-Use pricing, Critical Peak Pricing, and Critical Peak Rebates (referred to simply as "Demand Response"), by simulating the behavior of rational consumers under a set of historical scenarios drawn from the greater San Francisco Bay Area.

Using realistic dynamic consumption models, we gain novel insights into the effects of intertemporal substitution on individual and social surplus. Defining the concept of a baseline-taking equilibrium, we are able to estimate the welfare impact of the perverse incentive to inflate the Demand Response baseline, under the assumption of perfect foresight. To summarize some of these findings: in a standard consumer model that does not allow for intertemporal substitution, the average magnitude of retail markups accounts for a much greater fraction of economic inefficiency than does the absence of real-time pricing, and DR programs have a negligible impact on economic efficiency. But with the introduction and improvement of load-shifting technology, real-time and other dynamic pricing programs become more important relative to the average magnitude of markups. In our models that incorporate consumption substitution, real-time pricing results in efficiency gains on the order of 10% or more of consumer expenditure (and a larger fraction of generation and capacity costs), whereas DR programs produce efficiency gains between one sixth and one half as large. The perverse incentive to inflate the DR baseline is greatly suppressed by the high retail energy prices that currently prevail, and would be further attenuated in a model that accounted for uncertainty. Existing retail tariffs, including those with Time-of-Use and demand charge components, have efficiency effects that depend strongly on the underlying model and parameters. Several existing tariffs give consumers incentives to substitute consumption in ways that are not necessarily welfare-improving, with the result that investment in load-shifting technology can have positive or negative effects, depending on the specifics of the tariff and consumption model.