| Abstract: | Research on differentiated products markets often uses structural demand/supply models to identify firms' marginal costs as product‐level cost data are unavailable. Using unique demand and cost data from cable TV, I evaluate a differentiated products model's ability to identify marginal costs. I find firms systematically price below profit‐maximizing levels, leading to biases in the model's marginal cost estimates. I study the implications for merger simulations and find that these biases compromise estimates of merger‐related cost efficiencies, yet do not prevent these models from generating useful predictions of the price and nonprice effects of mergers. |
