The emerging stylized story about administrative unit proliferation in the developing world is that it is often a result of political machinations by national and local elites intent on creating new units for marginalized groups and for the ruler to buy votes; and that such proliferation only serves to re-centralize actual power — see for example these really cool papers by Grossman and Lewis (on the specific case of Uganda), Mai Hasssan (on the use of new districts to buy votes in Kenya) and Kimuli Kasara (also on how heightened electoral competition after 1992 accelerated the process of administrative unit proliferation in Kenya).
But there is also a slightly different, and in some ways complementary, story.
Regarding the creation of new provinces in Vietnam, Edmund Malesky notes:
The timing of provincial separations after Party Congresses, the dominance of Non-state Provinces despite little change in national output, and the decisive political outcome of this dominance at the 2001 Party Congress bolster the argument that reformers had an explicit electoral strategy in calling for the splitting of provinces in 1996. By creating new Non-state Provinces, modernizers believed they could influence the outcomes of future CCOM debates about
grand strategies and smaller NA debates about implementation of these new policies. While rhetorically it was easier to argue for new provinces based on efficiency, it would seem they were studying maps of
district economic composition and creating new reform-oriented
provinces out of SOE-dominated areas.
The key difference between administrative unit proliferation in Vietnam and Uganda (and Kenya before 2010) is the electoral connection (an aspect that, in my view, is missing in the current literature). Because the provinces had votes (in party congresses and plenums), the creation of new Vietnamese provinces had significant implications for the de facto distribution of power in both Hanoi and the periphery (and in Malesky’s story, made reforms possible). Provincial splits in Vietman were therefore not just about patronage and marginalized groups, but also about securing a win for the reformist bloc at the centre.
This might not be the case in countries where new units can be created without altering the balance of power in the party congress or parliament — either because such action does not create new electoral districts; or the president gets to nominate or can credibly influence the election of the representatives of the new districts. For this reason, I would predict that Kenya, Nigeria, and South Africa (whose subnational units are electorally significant and have a fair amount of fiscal autonomy) are unlikely to create new primary subnational units willy-nilly.