For this study four input-output tables were constructed for the Northern Territory for the 1980/81 financial year. The means of construction were by the simple and modified RAS techniques and by the GRIT technique (a preliminary and final version). The essential difference between the preliminary and final tables (or the simple and modified tables in the case of the RAS technique) is that the "preliminary" tables were constructed by purely mechanical means while the "final" tables benefitted from the incorporation of various kinds of "superior" data (i.e. survey data, government statistics, etc.).
Two sets of comparisons were made between the direct coefficients matrices of the various tables. First, comparisons were made between the preliminary and final tables to assess the significance, or otherwise, of the superior data. Second, comparisons were made between the final GRIT table and the modified RAS table to determine if there was any significant difference between the tables estimated by these two techniques. The method of comparison closely followed the Butterfield-Mules routine.
The results showed that the superior data collected for this study were significant and effective. This conclusion adds weight to the already solid argument that purely mechanical techniques, by themselves, cannot produce satisfactory or reliable regional input-output tables.
The results, however, were inconclusive with respect to the difference between the tables estimated by the modified RAS and final GRIT methodologies. It was recognized that examination of just one region is insufficient to draw definitive conclusions and evaluations of the type made here need to be done for as many and as wide a variety of regions as possible.