This paper studies the cost effectiveness of climate policy if there are technology externalities. For this purpose, we develop a forward-looking CGE model that captures empirical links between CO2 emissions associated with energy use, directed technical change and the economy. We find the cost-effective climate policy to include a combination of R&D subsidies and CO2 emission constraints, although R&D subsidies raise the shadow value of the CO2 constraint (i.e. CO2 price) because of a strong rebound effect from stimulating innovation. Furthermore, we find that CO2 constraints differentiated toward CO2-intensive sectors are more cost effective than constraints that generate uniform CO2 prices among sectors. Differentiated CO2 prices, through technical change and concomitant technology externalities, encourage growth in the non-CO2 intensive sectors and discourage growth in CO2-intensive sectors. Thus, it is cost effective to let the latter bear relatively more of the abatement burden. This result is robust to whether emission constraints, R&D subsidies or combinations of both are used to reduce CO2 emissions.