RDP 2010-08: Sources of Chinese Demand for Resource Commodities Appendix A: Fixed Asset Investment Data by Sector

Obtaining estimates of investment by sector for China is not straightforward. We use fixed asset investment (FAI) data, as disaggregated data on gross fixed capital formation are unavailable. FAI includes purchases of land, and existing structures and equipment (which should be excluded from GDP); this means these data are generally biased from a national accounting perspective. We proceed on the assumption that industry shares of gross fixed capital formation are similar to those for FAI. More seriously, a major change in classifications and in the scope of measured expenditure and output in 2004–2005, following the 2004 economic census, means that internally consistent series of investment by industry are not available over the longer term. From 1996 onwards, monthly data on total urban FAI by industry is available. Prior to that, there is only incomplete industry detail on investment in capital construction and innovation by industry, which is defined as FAI less investment in real estate development, urban private housing projects and rural FAI (NBS 2004).^[18] Owing to large differences between the FAI and the capital construction and innovation data, and the lack of industry detail for real estate and ‘other’ investment, we focus on developments since the mid 1990s.

While the changes in classifications around 2004 mean that long-term series for individual industries cannot be reliably constructed, we can aggregate industries into broader classifications to get a rough idea of the relative roles of large sectoral groupings. Barnett and Brooks (2006) decompose urban FAI since 2004 into manufacturing, infrastructure and real estate (which together account for 85 per cent of FAI according to their classification). Owing to the break in the data, and in the interests of obtaining a longer time series, we do not exactly replicate the decomposition of Barnett and Brooks for manufacturing and infrastructure investment, although our results for the post-2004 period are similar to theirs.

Our definition of manufacturing is the same as that of Barnett and Brooks from 2004 onwards, since it is given as a complete category in the FAI by industry data. Prior to 2004, we define manufacturing as ‘secondary industry’ less ‘energy’ and ‘construction’. Barnett and Brooks define ‘infrastructure’ investment as the sum of FAI in electricity, gas & water; transport, storage & post; water conservancy & environmental management; education; health, social security & welfare; and public administration & social organisations. From 2004, we follow the definition of Barnett and Brooks, except that we omit public administration & social organisations and include culture, sport & entertainment. Given the higher level of aggregation in the pre-2004 data, before 2004 we define infrastructure as the sum of ‘industry: energy’, transport, storage & telecommunications; culture, education & health care; and ‘other’ (since infrastructure-related categories that did not exist prior to 2004 such as water conservancy & environmental management were included in this category). Including investment in the ‘construction’ industry itself would make little difference to the calculation as it is small (around 1 per cent of total FAI), but we omit it as it is not clear that it constitutes ‘infrastructure’ investment as such. Since a (discontinued) urban real estate investment category is available prior to the 2004 reclassification, we use this series to extend the real estate FAI series back to 1996.^[19]

Footnotes

Capital construction investment refers to new or ‘extension’ construction projects with a total investment of CNY500,000 or greater. Innovation investment refers to technological improvements of existing facilities or renewal of fixed assets. [18]

Barnett and Brooks (2006, p 29) use a ‘wider definition of real estate investment than in the FAI survey and estimated … to include investment in residential buildings beyond that undertaken by real estate developers’. We do not adopt this strategy owing to uncertainty regarding how residential building investment is allocated across industries. [19]