General Remarks

The China Household Income Project (CHIP) data sets consist of urban, rural, and, for 2002 and 2007, rural-urban migrant samples. The sizes of these samples are not proportional to their shares in the Chinese national population. Also, their regional distributions differ from those in the population. Consequently, weights are needed in order to make the samples nationally representative.

In this Appendix we discuss the calculation of sample weights that can be used for analysis of the 2002 and 2007 CHIP data. We calculate these weights using data provided by the National Bureau of Statistics (NBS) from the 2000 census and the 2005 1 percent population sample survey, hereafter called the “2005 mini census.” The census and mini census are the most complete available accountings of China's population. Our sample weights are designed to reflect population shares in the census and the mini census.

We begin with a discussion of the CHIP sampling design and its implications for the calculation of weights (Section II). The calculation of weights requires data on population shares by geographic location and by urban, rural, and migrant classification, which we obtain using data from the 2000 census and the 2005 mini census. Section III discusses the census and mini-census data that we use for this purpose. In order to construct and apply the weights consistently, we must classify the location of residence for all individuals and households and make sure that there is no double counting. The classification of location is discussed in Section IV. The last section of this Appendix raises some suggestions for implementation of the weights in the analysis of the data.