Tacit Collusion in Real-Time U.S. Electricity Auctions

Abstract

The theory of infinitely repeated games finds that players will likely learn to collude, given sufficient incentive to do so, enabling them to capture greater payoffs than those attainable in the static setting. The theory posits that cooperation can arise in this framework if individuals believe that their current actions will affect opponents’ future strategies. That is, cooperation can be sustained in infinite games if opportunistic behavior triggers a punishment mechanism that discourages “cheating.” This paper models the restructured electricity spot market as an infinitely repeated, uniform last-price auction. It does so by starting with the framework of von der Fehr and Harbord (1992, 1993), who solved for an electricity market’s static Nash equilibrium. Next, it derives the payoffs attainable under the static and the proposed dynamic Nash (or collusive) equilibrium, assuming that deviations from the dynamic Nash equilibrium are met with a “grim trigger” strategy (i.e., playing the static Nash equilibrium from that point onwards). The paper finds that the proposed equilibrium constitutes the sole pure-strategy equilibrium of the dynamic game, contradicting von der Fehr and Harbord’s (1992, 1993) conclusion that no pure-strategy equilibrium exists in their framework. The paper explains that the implied threat of future punishment allows the players to reach a self-enforcing collusive equilibrium with greater payoffs than those of the static Nash equilibrium.