Anyway, here's an interesting result:
[D]omestic R&D intensity has no statistically significant impact on MFP [multifactor productivity] growth.
This result comes from analysis of Belgian data. I would hypothesize that domestic R&D activity does indeed have negligble effect in small open economies in terms of productivity - they can basically free-ride on the rest of the world, but this ignores immediate two points, there are probably more.
One, a country may have a unique specialized industrial cluster. The Canadian example would be the tar sands. I do not know for sure, but I suspect that since these are such a uniquely Canadian story, their productivity has grown more or less grown in step with domestic research advances.
Second, there are of course positive externalities from all scientific endeavour. Even if domestic research isn't a substantial fraction of world research, having local expertise would presumably encourage the creation of local high-technology enterprise, for example.