This research presents a price endogenous mathematical programming model that incorporates the independent, optimizing behavior of individual participants to estimate the possible efficiency loss of a newly developed permit trading market for nitrogen oxides (NOx) control in southern Taiwan. The result shows that when control equipment decisions are indivisible, an efficiency loss may arise due to over-investment. The efficiency loss found here is not because of a bilateral trading process and/or insufficient information for finding trading partners, but it is due to not having full control ability of the installed equipment.