Counting Errors: How Misunderstanding of China's Local Population and Employment Counts Affects Economic Analysis

Abstract

A confusion over local population counts and employment counts in China persists throughout the economic literature. The existing studies within many branches of economics related to China commonly use the count of people with local hukou registration, the GDP per capita using the hukou count as the population denominator, and incomplete employment counts reported in various Chinese statistical yearbooks as if they were appropriate measures of local population, local employment and local per capita income. This misunderstanding of China’s local population and employment counts may lead researchers astray and could contribute to wrong conclusions and mistaken policy implications. An analogous error would be to study the economic geography of, say, the United States using local population and employment counts that were only for citizens rather than for all residents; the resulting statistical pictures are likely to be biased since cities such as New York or Los Angeles have a much higher share of non-citizens than do other areas. Yet there is no similar criticism of the large and growing literature that relies on sub-national population counts for China, despite the equally egregious errors that are likely to have resulted from using the wrong population and employment counts. Five economic issues that are pertinent to modern China are studied in this thesis using the most reliable estimates of local resident population and employment from China’s 2010 population census. These issues are regional inequality, optimal city size, the city size distribution, the nature, location and size of urbanization economies, and the determinants of urban land area expansion. Each issue is studied in a single, self-contained, paper that in some cases has already appeared in a peer-reviewed journal, and in other cases is under review at such journals. In general, the findings based on the most reliable local resident population counts and employment counts from the 2010 census extensively challenge the existing evidence in the literature. In some cases the policy implications from this new research are the opposite of what was previously suggested from research that was potentially distorted by the counting errors described here.