A Taxonomy of Extractions

Much of the work that has been done on economic linkage measures that are derived from input-output data has been or can be cast in the framework of partitioned matrices. In this paper, we explore the ramifications of alternative hypothetical extractions within that framework. We examine all possible extractions and speculate on the plausibility of the economic stories that might underpin them, as well as whether or not they measure any interesting kind of linkage. We demonstrate that a number of alternative extractions produce identical results for certain measures of sectoral importance. In numerical illustrations, we find that an even larger number of the extractions produce identical rankings of sectors