Previous studies suggest that emissions taxes are more efficient at stimulating the development of improved pollution abatement technologies than other policy instruments, such as (non-auctioned) tradable emissions permits. We present results from a competitive model that cast some doubt on the empirical importance of this assertion. For example, we find that efficiency in the market for "environmental R&D" under tradable permits is typically less than 6 percent lower than that under an emissions tax for innovations that reduce pollution abatement costs by 10 percent or less. However the discrepancy is more significant in the case of more major innovations. We also find that the presence of R&D spillovers per se does not necessarily imply large inefficiency in the R&D market. For example, efficiency in the R&D market under a Pigouvian emissions tax is generally more than 90 percent of that in the first best outcome if the private benefit from innovation exceeds 50 percent of the social benefit. Thus the R&D spillover effect must substantially limit the private benefit from R&D in our analysis for there be a potentially "large" efficiency gain from additional policies -- such as research subsidies -- to stimulate innovation.