Redesigning automated market power mitigation in electricity markets

Electricity markets are prone to the abuse of market power. Several US markets employ algorithms to monitor and mitigate market power abuse in real time. The performance of automated mitigation procedures is contingent on precise estimates of firms' marginal production costs. Currently, marginal cost are inferred from the past offers of a plant. We present new estimation approaches and compare them to the currently applied benchmark method. We test the performance of all the approaches on auction data from the Iberian power market. The results show that our novel approaches outperform the benchmark approach significantly, reducing the mean absolute estimation error from 11.53 €/MWh to 2.77 €/MWh for our most precise alternative approach. Applying this result to a market mitigation simulation we find sizeable overall welfare gains and welfare transfers from supplier to buyer surplus. Our research contributes to accurate monitoring of market power and improved automated mitigation. Although we focus on power markets, our findings are applicable to monitoring of renewable energy tenders or market power surveillance in rail and air traffic.