Towards developing a hybrid method for input-output table compilation and identifying a fundamental economic structure

This dissertation focuses on two topics that are important to regional economists. The first is the development of a hybrid method for input-output table compilation which distinguishes between economic relationships that need to be surveyed and those that can be obtained by other means. The second is the identification of those parts of an economy that can be considered basic or fundamental to an economy. Both these concepts are developed by using nine regional South African input-output tables, aggregated to a 6-sector level and open with respect to households. Specific properties of these tables are analyzed viz. importance, stability and predictability. By using the three properties and certain decision rules a projection matrix is developed, which distinguishes between the economic relationships that need to be surveyed, those that can be predicted, and those for which base data can be used. The accuracy of the method is tested on three different aggregation levels (6-, 19- and 32-sectors), initially with households excluded and then with households included in the intermediate matrix. The tests show that the projection matrices provide sufficiently accurate estimates of regional coefficients and that the accuracy improves with a decrease in aggregation level. With households included the accuracy is somewhat negatively affected, although not significantly. The fundamental economic structure is defined as those economic relationships that are important building blocks for the economy and whose magnitudes reflect the level of economic activity within a region . It is shown that the fundamental economic structure consists mostly of services and certain manufacturing activities and has an urban orientation. Similar results are obtained at the 6-, 19- and 32-sector levels