I think the neutral names are of limited value. They require everyone to understand the intended meaning and it is highly error prone that different collection manager will read the specifications differently. In Germany we have states, then a next level (administrative district, "Regierungsbezirk") hardly anyone outside of the administration would know which one the own town belongs to. Then we have the "Kreis" which is roughly equivalent to a county and fairly well recognized by most people. This has the slight catch that small cities belong to a county, whereas big cities are "county-free cities" (kreisfreie Stadt). As a result, one manager would map "Regierungsbezirk" to second level, others the "Kreis", others will put the "circuit-free city" in second level, others will leave it empty and put the county-free city in third, to have all cities in the same field...
One of the big advantages of computers is, in my eyes, that they can easily handle (at least) simple synonymy. With the advent of the Semantic Web this promise has become even more realistic.
Providing a set of semantically well defined and well labeled names, some of which are applicable only to some countries, and leave it to software and a configuration thereof to figure out which mapping is most appropriate for which use case, would be my preference over the mapping pre-defined and just numbered by level in the hierarchy.
Gregor