RDP 2010-08: Sources of Chinese Demand for Resource Commodities Appendix B: Input-output Analysis

To compare the relative roles of manufacturing and construction as consumers of resources over time, we employ the 17 industry input-output (I-O) tables published by the NBS for the years 1995, 1997, 2000, 2002 and 2005, and collapse a 42 sector table published by the NBS (a version of the 135 sector table) for 2007 into a comparable 17 industry table. Table B1 shows how we have reclassified industries to contract the 42 sector I-O table.

These I-O tables do not distinguish between imported and domestically produced intermediate goods. To estimate the share of resources used as inputs to construction and manufacturing, we assume that the intensity in the use of imported inputs is the same for production for exports and production for domestic sales. This assumption has been used in the literature to construct separate intermediate use tables for imports and domestic production (see Dean et al 2008 for a discussion).

Construction is defined as a separate industry in the I-O tables, and we define manufacturing as ‘machinery and equipment’, ‘chemicals’, ‘building materials and non-metal mineral products’, ‘textile, sewing, leather and fur products’, and ‘other manufacturing’ industries. Metal products manufacturing includes both finished and unfinished metal products (including raw steel), and is thus closely related to mining production and resource imports. Consequently, we exclude this sector from ‘manufacturing’ to obtain a better understanding of the relationship between resource imports and manufacturing production.

We look at the direct uses of products from three resource-related industries: ‘mining and quarrying’, ‘metal products’, and ‘coal and petroleum’. For these three industries, their direct use is defined as the share of domestic supply used as intermediate inputs in the construction and manufacturing sectors. We define the share of domestic supply of (resource-related) industry j used by industry i as α_ij :

where: Y, M, and X and are gross output, imports and exports (of industry j); (Y +M −X) is domestic supply; and y_ij is the gross output of industry j consumed by industry i.

To estimate the direct use of resource-related industries' products by the manufacturing export sector, we multiply the ratio of exports to gross output for each manufacturing industry k (where k = 1…5) by its share of resource-related industry j's domestic supply. We then sum over k to obtain the total direct use of resource-related products by the manufacturing export sector. Denoting the direct use of products supplied by resource-related industry j to the manufacturing export sector as , and the export-to-gross output ratio for manufacturing industry k as xr_k we have:

To look at indirect effects of manufacturing and construction demand on the output of ‘metal products’, we calculate the effect of a one unit increase in manufacturing and construction final demand on ‘metal products’ gross output. This is done by computing the Leontief inverse of the direct input coefficient matrix from China's I-O tables over the 1995–2007 period.

First, denote the final demand of industry j as c_j. Second, denote the share of the gross output of industry j used by industry i as δ_ij. An industry j's gross output (that is, intermediate use plus final demand) can be written as:

Denoting the direct input coefficient matrix as A, and the vectors of final demand and gross output as C and :

we can summarise the input-output table as follows:

where (I – D)⁻¹ is the Leontief inverse. Element ij of (I – D)⁻¹ , say T_ij, is the impact of a one unit increase in c_i (the final demand of industry i) on Y_j (the gross output of industry j). T_ij is the total effect coefficient of an increase in final demand in industry i on the gross output of industry j. In effect, the coefficient T_ij estimates the output of ‘metal products’ required both directly and indirectly to produce an extra unit of ‘manufacturing’ or ‘construction’ final demand. See Miller and Blair (2009, Chapter 2) for further discussion of this exercise.

To calculate the effect of a one unit increase in manufacturing final demand, we weight each of the five manufacturing sub-industries' total effect coefficients by the manufacturing sub-industries' respective shares of total manufacturing final demand. We do not compute a separate effect for a rise in manufacturing export demand, since the NBS I-O tables do not allow us to distinguish between a change in final demand for manufactures that are exported and a change in final demand for manufactures that are consumed domestically.