Pareto improvements are theoretical concept. neonological argues here, that those are impossible in reality. I think this is pretty hard to argue one or other way, because side effects of our activities aren't always obvious. Things like tumbleweed in northern america or any accidentally imported plant/animal is good example about costs, that arrive later and for other people. Though I think there is possible to construct some clear Pareto improvement cases, like early improvements of internal combustion engine, where I don't see possibility of negative effects, only additional power out of same amount of fuel.