In non-agglutinative languages we have certain word classes that would be considered productive, and we have word classes that wouldn't. Morphological inflection, on the other hand, is from my knowledge (please correct me otherwise) not ever considered productive. Is this the case in agglutinative languages as well, i.e. is the production only occurring on a root-level? Productive morphology would otherwise sound like a fascinating topic!
I don't think your definition matches the traditional one, which is more about the degree to which a morpheme can freely and grammatically combine with various words (within a given word-class) -- for example, the -er morpheme (as in singer, talker, digger, ...) is extremely productive with Vs, while the -man morpheme (as in workman, postman, journeyman, ...) is also productive, with Ns, but much less so, and the -smith morpheme (as in blacksmith, swordsmith and wordsmith) is rather unproductive, with Ns.
note also that inflectional morphology can be productive or not -- English -ed and -ing are rather productive, but -en is actually rather not productive (run a little 'wug' test on a native speaker).