Estimating the Congestion Benefits of Batteries When Network Connections Are Unobserved

I estimate the grid impacts of 126.3 MW of battery storage capacity operating in California during 2010 to 2017. I find that 1 MW of energy storage decreases evening peak prices by $1.11/MWh at the pricing node where the storage is installed while off-peak prices do not increase. The unobserved network structure is revealed and the price effects are estimated at other locations across the grid using a double-pooled LASSO-based estimator that reflects the grid’s constrained optimization price-determination process. Results imply a reduction in the cost of serving load of $123,700 per year per MW, an amount similar in magnitude to private revenue from arbitrage. Overall, I find early energy storage mandates in California are partially justified by reductions in the cost of serving load, and that effects are strongest at times in which increased penetration of zero-carbon solar generation impose higher costs on the grid. These patterns likely hold in any grid setting with inelastic peak supply relative to off-peak supply. JEL Classification: Q41, Energy: Demand and Supply; Prices; Q48, Energy: Government Policy; L14, Transactional Relationships; Contracts and Reputation; Networks; H23, Taxation and Subsidies: Externalities; Redistributive Effects; Environmental Taxes and Subsidies