Using firm-level based TFP indicators (as opposed to employment-based proxies) we estimate the effects of alternative sources of dynamic externalities at the local geographic level. Contrary to previous empirical work, we find that industrial specialization and scale indicators positively affect TFP growth at the city-industry level, while we do not find evidence that either the degree of local competition or productive variety impact on subsequent productivity growth. Employment-based regressions yield nearly the opposite results, in line with previous empirical work. We show that such regressions could suffer from serious identification problems when interpreted as evidence of dynamic externalities. This calls into question the conclusions of the existing literature on dynamic agglomeration economies.