Abstract ：
Under linear price and cost functions, the optimal environmental pol- icy is determined for duopolies with product di¤erentiation. In the case of non-point source pollutants, the standard policies cannot be applied since the government has limited information about the individual emis- sion, only the total size of the pollution is known. The …rms’ decisions are concerned with their outputs and abatement technologies, while the government chooses the uniform pollution tax rate. The optimal decisions are determined in a two-stage process. At the second stage, the …rms de- termine their outputs, taking the technologies and the tax as given. At the …rst stage, the …rms select abatement technologies with given the tax rate, and the government selects optimal tax rate with given choices of the …rms. Under asymmetric information, the government constructs the welfare function with uncertainty on the …rms’ outputs and determines the optimal tax rate by maximizing the welfare expectation and mini- mizing the welfare variance. Since the best reply of the government has a complicated form, the Nash equilibrium is numerically and graphically solved. It is shown that the optimal ambient charge tax is less than the Pigouvian tax. It is also shown that the ambient charge tax e¤ectively controls the total concentration of NPS pollution.

Keywrods: Emvironmental policy, Ambient charge, NPS pollution, Two-stage game, Cournot duopoly, Multi-objective optimization