Real Time Pricing and Electricity Markets

Most US consumers are charged a near-constant retail price for electricity, despite substantial hourly variation in the wholesale market price. This paper evaluates a randomized experiment that exposed households to hourly real time pricing (RTP) and applies the demand estimates to counterfactual simulations in a structural model of the Pennsylvania-Jersey-Maryland electricity market. The model includes a di¤erent approach to the problem of multiple equilibria in multi-unit auctions: I non-parametrically estimate unobservables that rationalize past bidding behavior and use learning algorithms to move from the observed equilibrium counterfactual bid functions. This routine is nested as the second stage of a static entry model that captures an important institution called the Capacity Market, which acts in equilibrium as a minimum constraint on system capacity and transfers the shadow price to capacity owners. There are four central results. First, the experiment’s net e¤ect on consumer behavior was energy conservation during peak hours, not substitution from peak to o¤-peak. Second, I detail plausible conditions under which large scale RTP could actually increase wholesale electricity prices in peak hours, contrary to predictions from short run models, while decreasing Capacity Market prices and total entry. Third, although the increased demand elasticity from RTP reduces producers’market power, in practice this would be a second-order channel of e¢ ciency gains. Fourth, I …nd that the welfare gains from residential RTP are likely, but not certain, to outweigh the costs of the required hourly electricity meters. JEL Codes: D24, D43, D44, L10, L51, L94, Q40. Keywords: Electricity pricing, randomized …eld experiments, multi-unit auctions, learning in games, Capacity Markets. — — — — — — — — — — — — — — — — — — — — — — — — — — — — I thank, without implicating, Alberto Abadie, Susan Athey, Eric Budish, Drew Fudenberg, Michael Greenstone, Bill Hogan, Steven Joyce, Chris Knittel, Erin Mansur, Erich Muehlegger, Larry Katz, Sendhil Mullainathan, Paul Niehaus, Chris Nosko, Ariel Pakes, Dave Rapson, Rob Stavins, and seminar participants at Duke, Georgetown, Harvard, New York University, Notre Dame, Resources for the Future, Stanford, Tufts, and UC Davis. Joe Bowring, Howard Haas, Ellen Krawiec, and Matt Thompson of Monitoring Analytics allowed my access to sensitive bidding and cost data and provided valuable insights on the PJM market. I thank Sam Newell, Harvey Reed, Alex Rudkevich, Paul Sotkiewicz, and Assef Zobian for helpful conversations on the details of restructured electricity markets. Marjorie Isaacson, Larry Kotewa, and Anthony Star of the Center for Neighborhood Technology facilitated my understanding of the household electricity demand data. Financial support is acknowledged from the Harvard University Center for the Environment and from Harvard’s Mossavar-Rahmani Center for Business and Government.