This section presents a comparison of manufacturing employment data from the 2000 census of China with the annual enterprise data for the same year and attempts to explain the discrepancies between the two sources. The comparisons here show that the regular administrative reporting system misses many millions of workers, not only in the manufacturing sector, but also in many other sectors of the economy. In addition, a discussion of the census results highlights an apparent tendency on the part of rural households to report household workers as agricultural workers, even if they work in manufacturing part time or for part of the year. Therefore, although the census achieved more complete reporting than did the official annual compilations from enterprises, the census, too, appears to have undercounted manufacturing employment, especially outside the cities and towns.
The 2000 census of China discovered more manufacturing workers than were reported from annual administrative data. Both the 1990 and 2000 censuses asked respondents information about the employment of all persons aged 15 years and older. In the 2000 census, the data were gathered in a long form filled out by about 10 percent of civilian households in every locality and chosen to be representative of the population as a whole. Figures cited in the rest of this report are extrapolated to the entire counted civilian population aged 15 years and older. 26
Employment data from annual enterprise reporting and from China’s 2000 census do not agree with each other. For example, table 3 shows the estimated numbers of employees in each major sector of China’s economy at or near the end of the year 2000 from the two major data sources. On November 1, 2000, the census recorded a total employed population of approximately 709.7 million workers. Two months later, administrative compilations of data from enterprises, economic units, and self-employed individuals recorded a total of 629.8 million workers, 80 million fewer than the census. (See table 3.)
What are the sources of the discrepancies between these two sets of data? We can see from table 3 that the census recorded 123 million more workers in agriculture than did annual administrative data. One reason for this large difference is that the census asked about employment only in the last week of October 2000, the week just prior to the date the census was taken. The census surely detected individuals who work in agriculture during peak planting and harvest seasons, but not the rest of the time, and these workers were counted as employed in agriculture during the peak autumn harvest season.
The way employment questions are asked in China’s censuses and the instructions for filling out the census forms apparently bias rural household respondents in favor of reporting all household members as agricultural workers, even if some adults in the family actually work in nonagricultural sectors of the economy most of the time. Therefore, the decennial censuses may overreport employment in agriculture and underreport actual employment in many industrial and service sectors of the economy. In particular, the censuses of 1990 and 2000 probably underreported the total number of manufacturing employees in China. 27