I like papers like these because they have lots of data, and once you have lots of data, you can disagree about what the data means.
The main problem I see with the paper is that it fails to connect the GDP sector numbers with changes in the Chinese economy from 1993-2004. During the late 1990’s, bankrupt Chinese SOE’s were rapidly closing and shedding tens of urban millions of workers, and the Chinese economy was struggling to employ those workers so there wasn’t any increase in employment.
Kuijs mentions some issues with data quality and talks about the how the floating population and the underestimate in services doesn’t really affect his main conclusions. There is one more factor that should be considered. Between 1990 and 1998, China dismantled the last remnants of central planning and moved to a market oriented pricing system. One persistent problem with centrally planned systems is that they tend to wildly underestimate the price of inputs and wildly overestimate the value of outputs. Once you move to more rational prices, this causes an apparent increase in capital intensity, but that just reveals the hidden capital intensity of the system under state planning.
Then again maybe not. The nice thing about papers with lots of data is that you can take these as hypothesis and examine how it influences the data, which is what I’ll be doing in the next week.