I extend multi-unit auction estimation techniques to a setting in which firms can express cost complementarities over time. In the context of electricity markets, I show how the auction structure and bidding data can be used to estimate these complementarities, which in these markets arise due to startup costs. I find that startup costs are substantial and that taking them into account helps better explain firm bidding strategies and production patterns. As in other dynamic settings, I find that startup costs limit the ability of firms to change production over time, exacerbating fluctuations in market prices. These fluctuations can induce estimates of market power that ignore dynamic costs to overstate markup volatility, with predicted markups that can be even negative in periods of low demand. I show how accounting for startup costs can provide a natural correction for these markup biases.

We measure the pass-through of emissions costs to electricity prices. We perform both reduced-form and structural estimations based on optimal bidding in this market. Using rich micro-level data, we estimate the channels affecting pass-through in a flexible manner, with minimal functional form assumptions. Contrary to many studies in the general pass-through literature, we find that emissions costs are almost fully passed through to electricity prices. Since electricity is traded through high-frequency auctions for highly inelastic demand, firms have weak incentives to adjust markups after the cost shock. Furthermore, the costs of price adjustment are small.