in which privately optimal R&D deviates from the first-best level depends on the kind of technical change in pollution abatement. Counterintuitively polluters might overinvest into R&D in pollution control if compensation falls short of harm, and may underinvest if compensation exceeds harm.