In some fields, most papers are referred to as working papers even when they are basically preprints. And often they are explicitly in archives for "working papers," or published in a series of "working papers" (often if they were presented at a university for feedback).
After determining that there is not a newer or published version, I usually cite these papers, regardless of age, as a working paper if it is referred to as such. However, depending on the citation system, there may not be an official way to cite something as a "working paper", and so in that case I often default to citing it as an "unpublished paper." (A similar problem, or individual preferences, may be why different authors choose to cite it in different ways.)
I do not think time should change that classification, because "working paper" is a phrase indicating the general status of the work to those in the field. At the extreme, I might still refer to a "working paper" by someone who retired or died, even though it is certainly not still a work-in-progress. For instance, here is the official BibTex file for the first working paper in the NBER archive, a 1973 work by Finis Welch. The citation has type "Working Paper."