Where has all the dynamism gone? : Productivity growth in China’s manufacturing sector, 1998-2013
Brandt, Loren; Van Biesebroeck, Johannes; Wang, Luhang; Zhang, Yifan (30.11.2023)
Numero
6/2023Julkaisija
Bank of Finland
2023
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi-fe20231130150500Tiivistelmä
China’s manufacturing sector has been a key source of the economy’s dynamism. Analysis after 2007 however is hampered by problems in the key data source for empirical analysis, the National Bureau of Statistics’ (NBS) annual survey of industrial firms. Issues include missing information on value added and intermediate inputs, and concerns of over-reporting. The annual survey of firms conducted by China’s State Taxation Administration (STA) provides a reliable, alternative source of firm-level data for years from 2007 to 2013. Since the sample is not representative and the precise sampling scheme is not known, the data cannot be used directly to draw inferences on China’s manufacturing sector. By comparing the joint distribution of key variables for which both surveys provide reasonably reliable information, we recover the sampling scheme of the STA survey and use it to simulate samples for 2007 to 2013 that are comparable to the NBS sample in earlier years. Our estimates reveal a marked slowdown in revenue-based total factor productivity growth that cuts across all industries, ownership types, and regions. The loss of dynamism in the private sector, and the reduced contribution of firm entry to aggregate productivity growth are especially prominent.
Julkaisuhuomautus
NON-TECHNICAL SUMMARY
FOCUS
China’s manufacturing sector has been a key source of dynamism. Analysis after 2007 is hampered by problems in the survey of industrial firms by China’s National Bureau of Statistics (NBS), including missing information on value added and intermediates and concerns of over-reporting. We leverage alternative firm-level data collected by China’s State Tax Administration (STA) in which the misreporting problems are less pronounced and extend earlier productivity estimates.
CONTRIBUTION
The STA survey provides a reliable, alternative data source, however the sample is not representative and the sampling scheme is not known. Therefore the STA sample cannot be used directly to draw inferences on China’s manufacturing sector. By comparing the joint distribution of variables reported reliably in the NBS and the STA surveys, we recover the STA sampling scheme and simulate samples for 2007 to 2013 that are comparable to the NBS sample in earlier years. We then estimate productivity and calculate aggregate statistics on simulated samples to provide a broader view of China’s manufacturing TFP growth over the whole period from 1998 to 2013.
FINDINGS
We find over-reporting problems in the NBS micro data after 2007 parallel those identified at the macro level and become more serious over time. Investigation of the simulated samples reveals an annual TFP growth rate of 1.4 percent between 2007-2013, about a third of the growth rate estimate between 1998-2007 based on the NBS data. The decline is observed across all industries, regions, and ownership types. Most importantly, the growth along the extensive margin, which made substantial contribution to China’s manufacturing TFP growth in the earlier period, has dissipated.
FOCUS
China’s manufacturing sector has been a key source of dynamism. Analysis after 2007 is hampered by problems in the survey of industrial firms by China’s National Bureau of Statistics (NBS), including missing information on value added and intermediates and concerns of over-reporting. We leverage alternative firm-level data collected by China’s State Tax Administration (STA) in which the misreporting problems are less pronounced and extend earlier productivity estimates.
CONTRIBUTION
The STA survey provides a reliable, alternative data source, however the sample is not representative and the sampling scheme is not known. Therefore the STA sample cannot be used directly to draw inferences on China’s manufacturing sector. By comparing the joint distribution of variables reported reliably in the NBS and the STA surveys, we recover the STA sampling scheme and simulate samples for 2007 to 2013 that are comparable to the NBS sample in earlier years. We then estimate productivity and calculate aggregate statistics on simulated samples to provide a broader view of China’s manufacturing TFP growth over the whole period from 1998 to 2013.
FINDINGS
We find over-reporting problems in the NBS micro data after 2007 parallel those identified at the macro level and become more serious over time. Investigation of the simulated samples reveals an annual TFP growth rate of 1.4 percent between 2007-2013, about a third of the growth rate estimate between 1998-2007 based on the NBS data. The decline is observed across all industries, regions, and ownership types. Most importantly, the growth along the extensive margin, which made substantial contribution to China’s manufacturing TFP growth in the earlier period, has dissipated.