An investigation into methods for producing hybrid regional input-output tables
Hybrid techniques are now the accepted means for producing regional input-output models. Nonetheless, the literature on hybrid model construction techniques has been sketchy at best in telling regional analysts where or how they might be able to obtain the "superior" data required to transform nonsurvey models into hybrid equivalents. Hence, it is the purpose of this dissertation to develop a strategy to produce a hybrid regional input-output model. To make the strategy practical, we limit the data foundation upon which it is built to that of the nonsurvey model of the same region. Through a thorough survey of the regional input-output literature, we note that several major issues need to be dealt with in producing a survey strategy for producing hybrid regional input-output models: (1) the level of aggregation preferred for the model, (2) the accuracy of the nonsurvey regionalization technique, (3) how to identify the minimum set of questions that should be asked of any sector, and (4) how to identify the sectors that should be surveyed. These are the topics of the four chapters that follow. We find that maintaining as disaggregate a table as possible helps to keep error to a minimum. In addition characteristics of the region and the sector being disturbed when using a regional input-output model may exacerbate the basic aggregation error. We also find that there are viable alternatives to the traditional regional purchase coefficient. In fact, there is evidence that leads us to believe that one of the alternatives performs considerably better in impact analysis. To identify questions and sectors to survey, we found it necessary to build an algorithm. The algorithm is largely based on Guy West's 1982 work in Australian Economic Papers on the sensitivity of the Leontief inverse to cell perturbations. We use a modified form of his sensitivity measure to identify sectors for which the Leontief inverse is most sensitive to proportional misestimations in imported inputs. We also use it to identify cells in the matrix corresponding to those import-sensitive sectors to which the Leontief inverse is most sensitive. RAS is used to reconcile the data from different data sources. Tests of the algorithm and its components are encouraging.
|Year of publication:||
|Authors:||Lahr, Michael Lincoln|
|Type of publication:||Other|
Dissertations available from ProQuest
Saved in favorites
Similar items by subject
Find similar items by using search terms and synonyms from our Thesaurus for Economics (STW).