The Cost of Decentralization? Welfare Comparisons between Double Auctions and Automated Market Makers in Transactive Energy Markets

This paper quantifies the welfare implications of adopting an Automated Market Maker (AMM) in place of the simultaneous time-delimited double auction used in the Transactive Energy Service System (TESS). We develop an agent-based simulation of a peer-to-peer electricity market populated by heterogeneous demand, solar, utility, and battery agents, and evaluate market outcomes under varying technical and operational parameters. Battery strategies are modeled both through value function iteration (VFI) dynamic programming and an alternative informed trader heuristic that exploits anticipated intertemporal price differentials. Regression analysis on simulation output reveals that, across a broad range of scenarios, the AMM delivers statistically significant gains in total social welfare relative to the double auction. While the VFI battery strategy often generates negative surplus under the AMM, these losses are offset by gains to other agents. The informed trader battery eliminates the negative battery surplus without diminishing total welfare, indicating that strategic adaptation can improve equity of surplus distribution while maintaining efficiency. These results suggest that the cost of decentralization—defined as welfare loss from substituting an AMM for a double auction—can be negative in certain contexts, implying a potential net benefit. However, we find that part of the AMM's advantage stems from its implicit capacity flexibility, which may not be feasible in practical deployment. We conclude by outlining future research to assess AMM performance under realistic capacity constraints and more diverse trading behaviors.