Restructuring electricity markets has enabled wholesalers to exercise market power. Using a common method to measure competition, several studies have found substantial inefficiencies. This method overstates actual welfare loss by ignoring production constraints that result in non-convex costs. I develop an alternative method that accounts for these constraints and apply it to the Pennsylvania, New Jersey, and Maryland market. For the summer following restructuring, the common method implies that market imperfections resulted in considerable welfare loss, with actual production costs exceeding the competitive model's estimates by 13%–21%. In contrast, my method finds that actual costs were only between 3% and 8% above the competitive levels.