Affiliations: Austrian Central Statistical Office, Division 7: National Accounts, P.O. Box 7000, A-1033 Vienna, Austria
Abstract: Conventional National Accounting (NA) concepts on the territorial delineation of an economy are not a field of particular controversy, due to their usually only marginal impact on the overall outcome. Therefore it might seem a natural solution to transpose the same rules and conventions valid on the national level of NA onto the subnational levels of territory. Through this analogy, however, it is ignored that with increasing degrees of regional detail at the same time difficulties are trading in, increasing more than proportionately. In view of this trade-off and limitations seldom reflected theoretically, a re-consideration of the basics of Regional Accounting (RA) seems to be necessary. This is particularly true in a period when important revisions of the related basic systems are approaching the stage of practical implementation. This is all the more true for RA based thereupon which are going to become particularly significant for practical policies. The following is a theoretical examination of this issue, starting from basics which are common to economic statistics, as well as NA and RA. These concepts, above all, refer to statistical units, classifications, peculiarities of observation and the limitations of regional economics, recognized in terms of empirical description rather than arm-chair assumptions and projection. A fundamental, clear categorical distinction between “causality” (which is accessible to observation) and “finality” (which is hardly so, or not at all) is advanced, which is to be applied as a primary characterisation of data specifically useful in the present context. On that basis several layers of observation, reliability and regionality are systematically distinguished so that in the end a pattern of RA develops which is quite different from nowadays approaches. While not providing the full range of data usually expected from RA, this system is still suitable for interpretation and analysis and perhaps even more so thanks to its avoidance of prejudice, arbitrariness and anticipation. Indeed, the unquestionable “hard core” of such a RA system turns out to be of relatively limited scope, complemented by segments of decreasing significance for the present purposes. Notwithstanding these distinctions, the data framework as used nowadays will not be lost: appropriately redesigned and reordered it might still be useful to accommodate the above distinctions. However, nobody would see himself longer caught in a situation either to believe the whole story or to refuse it at all. The text is completed by a brief (though comprehensive) discussion of concepts useful in this statistical context, as well as by a summary of relevant theories.
