First, even for those dimensions of environmental quality where growth seems to have been associated with improving conditions, there is no reason to believe that the process has been an automatic one. In principle, environmental quality might improve automatically when countries develop if they substitute cleaner technologies for dirtier ones, or if there is a very pronounced effect on pollution of the typical patterns of structural transformation. Our methodology does not allow us to reject these hypotheses, or even to investigate the means by which income changes influence environmental outcomes.