Input-Output Table and Analysis

In compiling an input-output table the first requirement is to have all domestic production units classified by industry. In the latest set of tables for the United Kingdom no less than some hundred different industries have been distinguished for this purpose. The second requirement in constructing an input-output matrix is to have the intermediate product purchases of each industry classified by industry of origin. It is this requirement that presents the input-output statistician with a problem. His main source of information is the Census of Production. There he finds establishments grouped by industry on the basis of the goods or services they produce. Since each of these establishments consists typically of ‘the whole of the premises under the same ownership or management at a particular address (e.g. a factory or mine)’, they are likely to be rather homogeneous with respect to the goods or services they produce. However, occasionally one or more of an establishment’s products are not characteristic of the industrial sector to which that establishment has been allocated in the Census of Production. Thus certain commodities are produced in more than one industrial sector. Generally the input-output statistician has no means of knowing from which of these various sectors the different users have made their purchases of these commodities. The Census of Production fails to supply this information. One way of getting round this problem is to assume that all purchasers of such commodities distribute their purchases in the same way, percentagewise, over the various industrial sectors producing these commodities. This pro rata assumption is extensively used in the construction of the industry x industry input-output tables for the United Kingdom (table D of the 1979 tables). It is not a very realistic assumption, but it has the merit of simplicity.