Australian Bureau of Statistics
5216.0 - Australian National Accounts: Concepts, Sources and Methods, 2000
Previous ISSUE Released at 11:30 AM (CANBERRA TIME) 15/11/2000
|Page tools: Print Page Print All RSS Search this Product|
Basic structure of the input-output and associated tables
9.19 Preceding sections of this chapter referred to goods and services accounts, S-U tables and input-output tables. Technically these accounts or tables are variants on a theme. They each describe the supply and disposition of the products, or outputs of industries, of an entire economic system for a particular period. The differences between these various types of tables relate to differences in valuation and structure which do not need further elaboration here. Readers interested in more detail regarding these differences should refer to Chapter 15 of SNA93.
9.1 INDUSTRY-BY-INDUSTRY MATRIX
9.23 The basic structure of an industry-by-industry table with direct allocation of imports is shown in table 9.1. The notation of rows and columns is that used in Australian National Accounts: Input-Output Tables (Cat. no. 5209.0). Flows between domestic industries are shown in Quadrant 1 (rows and columns 0101 to 9601). This is usually referred to as the inter-industry quadrant. Each column in this quadrant shows the intermediate inputs into an industry in the form of goods and services produced by other industries, and each row shows those parts of an industry's output which have been absorbed by other industries. For example, the cell at the intersection of row i and column j shows how much output of industry i has been absorbed by industry j for current production. Disposition of output to categories of final demand is shown in Quadrant 2, comprising rows 0101 to 9601 and columns Q1 to Q7. Quadrant 3 (rows P1 to P5 and columns 0101 to 9601) shows entries usually referred to as primary inputs: compensation of employees; gross operating surplus and gross mixed income; imports; and various types of taxes on production. These inputs differ from the intermediate inputs in that they are not part of the output of current domestic production.
9.24 Quadrants 1 and 2 together show the total usage of the goods and services supplied by each industry. Total usage equals total supply because Quadrant 2 includes changes in inventories (which may be positive or negative). Quadrants 1 and 3 together show the inputs used to produce the total supply (outputs) of each industry. The sum of the inputs equals total supply (outputs) because the primary inputs in Quadrant 3 include gross operating surplus and gross mixed income (which may be positive or negative conceptually).
9.25 In some tables, the figures shown for total supply from each industry include not only Australian output but also similar products which are imported; these tables are said to have an indirect allocation of imports. When the tables are arranged in this way, the amounts of inputs into one industry supplied by each of the other industries reflect technological relationships between all inputs into the industry, whether or not they are domestically produced, and the output of that industry. The assumption of a functional relationship between an industry's inputs and its output is implicit in many uses of input-output tables, and this is an important consideration in the design of the tables.
9.26 In the preceding description of the basic structure of input-output tables, a row or column in Quadrant 1 was said to refer to an industry. However, in some tables a row or column may represent a product or a group of products rather than an industry. For this reason, rows and columns in Quadrant 1 are called sectors. This part of the chapter discusses the possible content and number of sectors, and the problems and analytical implications of the sectoring used.
9.29 The stability of coefficients is affected by the interaction of two factors: aggregation into a single sector of products with different input structures, and changes in the sector's product mix over time. This is very important when the input-output coefficients are only available from infrequent surveys, and it is therefore necessary, when compiling input-output tables, to assume that the coefficients observed in one year apply in neighbouring years, at least as a starting point. For example, if the textile industry is defined as a single sector, the coefficients for yarn inputs will be different for textiles produced from different yarns and will change if the proportions of these textiles change. Again, in an industry such as motor vehicles, trucks may contain a significant amount of timber whereas cars do not. If an increase in fixed capital formation leads to a higher demand for trucks than in the survey year, the projected requirements for timber will be understated because the input coefficient in the survey year relates to timber used to produce trucks and cars in the proportions for that year. Such problems will arise in industries producing a range of products, particularly when each product has a different input structure.
9.30 Even in large input-output tables there is substantial aggregation, which leads to a departure from these ideals and affects the homogeneity of sectors. There are two ways in which the aggregation can be made. One is a grouping of industries, leading to an industry-by-industry table. The other is a grouping by products, leading to a product-by-product table. These two approaches have different implications for homogeneity and therefore for the subsequent analytical uses of the tables. No system of sectoring can completely eliminate the aggregation problem, but an appropriate sectoring can keep it within acceptable limits. The choice of sectors depends partly on the practical problems of compilation.
9.31 It would appear, at first sight, that it might be best to define sectors as fairly narrow product groupings. This would go close to satisfying the first two criteria of homogeneity, but there are disadvantages in defining sectors this way. The resulting tables may be too detailed for many uses. They may take too long to compile. Necessary data may not be available in sufficient detail. In particular, when an establishment produces products classified to different sectors, there are difficulties in obtaining separate details of the inputs into the different product sectors. In an economy such as Australia there may be the overriding disadvantage that the resulting estimates could not be published anyway, because they would be confidential. Finally, it is likely that the third homogeneity criterion would not be satisfied. If, for example, electric cables made of aluminium were in a different sector from cables made of copper, there could well be substitution between the products of the two sectors and some conclusions drawn from the tables could be of doubtful validity.
9.32 If industries are chosen as sectors, homogeneity may be impaired by the wide range of products produced by the establishments in some industries. Products are described as primary to an industry when they are produced mainly by that industry. The secondary outputs of an industry are those outputs which are primary to another industry. Where the range of primary products for a given industry is wide, the output of that industry may contain products which may have very different input structures. In addition, they may contain not only products primary to the industry but also secondary products (primary to other industries) and the corresponding inputs. If, for example, the basic iron and steel industry also produces non-ferrous castings, the column for this industry will show inputs of non-ferrous metals and the row will include sales to industries using non-ferrous castings. Such a presentation may be unsatisfactory to users who are interested in the production and disposition of basic iron and steel products only. More importantly, if either of these problems exists, the second criterion of homogeneity will not be satisfied and the input coefficients will vary in response to changes in product mix. Unless the production of secondary products represents a fixed proportion of the industry's output, the requirements calculated from the table may be misleading.
9.33 Often there is no necessary connection between the production of primary and secondary products, and it cannot be expected that the proportions will remain constant. However, where the secondary products are joint products or by-products (such as production of sulphuric acid in basic metal smelting) the proportions will normally remain constant and there will be no problem on that account. Nevertheless, a problem may arise in this case because a change in demand for these products is more likely to affect the output of specialist producers of those products than that of the industries which produce them as by-products.
9.34 The extent of secondary production (products primary to another industry) depends on the range of products produced by individual establishments and on whether the establishments are grouped into a large number of narrowly defined industries or a smaller number of broadly defined industries. With narrowly defined industries, a large proportion of some products will be produced by industries to which the products are not primary. This tends to conflict with all the homogeneity requirements and, most seriously, it conflicts with the non-substitution requirement. Where significant proportions of a product are produced by a number of industries there can be easy substitution between that product produced by one industry and the same product produced by another industry. There is then a very weak link between the demand for that product and the output of a single industry. Thus, given basic statistics of establishments classified to narrowly defined industries, combining some of these industries will improve homogeneity in one important respect. There is a limit, though, because the improvement may be offset by a more heterogeneous product mix. Also, provision should be made, if possible, for users wishing to undertake detailed product or industrial analyses.
9.35 As well as conceptual considerations, the choice of sectors is influenced by the nature of the statistical data available. Thus, detailed information on sales or output is normally available for products, but information on costs or inputs may not be available. The total value of inputs used by an establishment or enterprise is often the only information available. As it is necessary to relate inputs to outputs, the statistical data normally available have to be arranged to bring out this relationship.
9.36 Product-by-product tables are theoretically more appropriate than industry-by-industry tables for some analytical purposes, but the differences between these types of tables arise only because of secondary production and depend on the extent of this production. If there is a significant amount of secondary production, the use of industry-by-industry tables for product analyses will produce less accurate results. However, if the extent of secondary production is small to start with, or if it has been possible to minimise it by appropriate sectoring and redefinition of industries (i.e. transferring some activity attributed to an industry to another, more appropriate, industry), the results obtained by using an industry-by-industry table may differ only slightly from the results which would have been obtained if a product-by-product table had been used instead.
9.37 The ABS prefers to compile industry-by-industry tables for a number of reasons. First, detailed information on inputs is not normally available for products. Therefore, the estimates of inputs in a product-by-product table must be based on assumptions and approximations with a consequent loss in accuracy. Second, experience in some overseas countries shows that product-by-product tables prepared entirely using the industry technology assumption (i.e. that a product has the same input structure wherever produced) can lead to anomalous or even unacceptable results. These anomalies could be avoided by using mixed assumptions as SNA93 recommends, but this approach is very expensive. Finally, it appears that most of the analytical applications of input-output statistics in Australia can be satisfied using industry-by-industry tables. Thus, analysis of the effects of changes in factor costs, productivity, incidence of taxes on production and imports, and primary input content of demand can be met by tables of this type.
9.38 Regardless of whether products or industries are used to define the sectors, the initial assembly of data is the same. It is necessary to record the product flows in the economy in a way suitable for input-output analysis. A system of building blocks is used, each of which shows, for a product (or, more commonly, a combination of products):
9.39 Recording supplies by industry of origin does not present any difficulty apart from the necessity to classify imports in the same way as locally produced products. The destination of products is more difficult to determine. The first requirement is information on the usage by each industry and final demand category, both in total and for the constituent products. Although the using industries can supply information on the nature of their inputs, the descriptions may be broad and may differ from descriptions used by the suppliers of the same products. Where the information is not available in sufficient detail it must be estimated.
9.40 Once these building blocks are ready they are arranged into four basic tables. The first of these is the supply table. It shows output of domestic industries and imports in the columns and output of products primary to these industries in the rows. Characteristically, the largest entries are on the main diagonal because an industry mainly produces products primary to it. For a large proportion of the cells in the supply table the estimate of the value of output is nil. In order to save space and assist readability, only cells with non-zero values are presented in the published supply table. This table provides insights into the way the production of products by industries is organised. The columns of the table show, for each industry, the products it produces (or the 'industry product mix', as it is sometimes called) and the extent to which each industry specialises in the production of products primary to it as well as the product composition of imports.
9.41 The use table has product groups and primary inputs in its rows, and industries and final demand categories in its columns. The rows of this table record the total supply of products, whether locally produced or imported, and show how these products are used by industries as intermediate inputs to current production and by final demand categories. Further down, the rows designated by prefix 'P' show the primary inputs which have been purchased by industries and by final demand. Reading down the columns one can find the composition of inputs (intermediate and primary) into each industry and the composition of each final demand category. Therefore, all flows of goods and services in the economy are covered.
9.42 The third basic table is the imports table. It shows in the columns the industries to which the imported products would have been primary had they been produced in Australia, and in the rows the usage of these products by industries and final demand categories. This dissection is shown only for competing imports, i.e. those products which are both produced domestically and imported, so that substitution between the two sources of supply is possible. It is not shown for complementary imports which, by definition, are of a kind not produced in Australia; nor for re-exports, which are goods imported into Australia and then exported without having been used or transformed in any way. These are recorded in separate columns rather than in the columns of industries to which they would have been primary if they had been produced in Australia. The imports table has not been included in the I-O publication, but is available on request (see Appendix D of Cat. no. 5209.0).
9.43 The fourth basic table is the margins table, which shows the difference between the basic price and purchasers' price of all flows in the use table. The margins table is the sum of separate tables for each type of margin (e.g. taxes on products (net), wholesale, retail). Table 3 in Cat. no. 5209.0 provides a summary margins table. The component margins tables are not included in Cat. no. 5209.0, but are available on request.
9.44 These four basic tables are simply a record of the estimated flows which occur in the process of production. However, the use table is not symmetric, which makes it unsuitable for some analytical purposes. It can be made symmetric by reorganising it so that both rows and columns refer either to industries or to products. In the first case, rows of the use table have to be adjusted to show industries purchasing industry output rather than products. In the second case, columns of the use table have to be adjusted to show inputs relevant to the production of products. These adjustments lead to symmetric flow tables which are either industry-by-industry or product-by-product tables. Only industry-by-industry tables are published by the ABS.
This page first published 15 November 2000, last updated 29 June 2012
Unless otherwise noted, content on this website is licensed under a Creative Commons Attribution 2.5 Australia Licence together with any terms, conditions and exclusions as set out in the website Copyright notice. For permission to do anything beyond the scope of this licence and copyright terms contact us.