A reassessment of the hybrid approach to the construction of regional input-output tables
This thesis lays the foundation for the creation of an economic database for the counties of Devon and Cornwall in the form of a regional input-output table. The thesis reconsiders the popular hybrid approach to the construction of such tables. In particular, the nonsurvey-to-survey ordering of procedure is questioned. The thesis attempts to restore a more logical preference-order which begins with first-best (survey) estimation methods and extends to second-best (survey-based-nonsurvey) methods. The third-best methods of estimation (pure nonsurvey i.e. location quotient) are excluded from the process altogether. The thesis is largely concerned with the development of the second-best method. The second-best method is derived from an empirical analysis of the nature of nonsurvey estimation error. The analysis is able to reject the Stevens et al. (1983) hypothesis that differences in regional and national production functions are insignificant. Nevertheless, the strategy of developing 'trade-only' nonsurvey estimation methods is found to be valid since, whilst the error associated with regional trade misspecification can be reduced within a broad method of estimation, the error attributable to the misspecification of regional production functions remains largely intractable to such an approach. Survey resources must therefore be devoted to the specification of these functions. The second-best methodology extends the Stevens et al. (1983) by deriving equations that specify the RAS algorithm and local expenditure propensities for households from empirical data for Scotland. These equations have general application within the new hybrid methodology. By restoring a more logical preference-order of approach to estimating hybrid regional input-output tables, emphasis is placed on the analytical strength afforded by a good data set, and not on the analytical 'strength' of magic-box mathematics. This should encourage the regional input-output table to be implemented as an evolving local economic database, which will improve the general quality of regional analysis and, in the long-run, offer cost-savings in data collection and collation.