https://www.npr.org/sections/parallels/2017/04/16/523632920/...
This can be seen first hand with Beijing+Tianjin versus Hebei.
Those 2 cities were under direct State Council control so it was easier to push economic reforms versus a largely rural province like Hebei, plus they were close enough that it made sense for anyone half ambitious in Hebei to leave for the big city than to stay back.
This is what happened in NY State as well with NYC and IL with Chicago. It's called primacy effect in economics.
The issue is, research into stuff like this and converting it into actionable policy is becoming difficult, as mid and lower level civil servants don't want to deviate from a literal definition of policies or their mandate as administrations have become much more vindictive due to the economic slowdown. Research like this ends up stepping on people's toes.
http://www.kjjb.org/EN/abstract/abstract18771.shtml
http://www.progressingeography.com/EN/10.18306/dlkxjz.2019.0...
https://www.sciencedirect.com/science/article/pii/S187705092...
https://www.sciencedirect.com/science/article/abs/pii/S02648...
I left Beijing, Shanghai, and Tianjin out due to the fact that they are governed differently than provinces, as they are under direct administration of the State Council and they are almost 100% urban. This makes them a bad comparison with Thailand as a whole. At a micro level, it might be interesting to compare BKK versus Beijing/Shanghai/Tianjin but Thailand allows free movement and domiciling, so it's not a great comparison.
> the question is if the poor parts have any low hanging fruit left to mimic the success of the earlier-developed and now high HDI areas, or have those fruit withered by now
It's a tough question and depends province to province. The provinces around Manchuria are probably all tapped out, but there might be some opportunity left in Guizhou, Yunnan, Sichuan, and Guangxi, but local and mid-level governance in China (like other middle income countries) sucks.
I have written a lot about this in a previous life, but I'd rather not write a whole Econometrics white paper on HN