We econometrically estimate the effect of transaction costs on the cost-effectiveness of the tradable permit market created during the US lead phasedown. We develop a methodology to identify transaction costs in the absence of price data. We find that refineries generally trade efficiently. We also, however, find evidence that transaction costs affect trading. We find evidence that refineries are less likely to trade in cases where theory suggests they will face high transaction costs. The data were collected from 30 major oil companies and include the trading partners and quantities traded for all permit trades carried out by each of 87 refineries over 8 quarters.