Exchanging Regional Input—Output Coefficients: A Reply and Further Comments

Abstract
Some issues raised by Thumann in response to an earlier article by Hewings are addressed in the context of discovering efficient ways of exchanging input–output coefficients. Several extensions to the RAS procedure are reviewed; in particular, attention is focused on the issue of fixing certain coefficients prior to the application of biproportional adjustment algorithms. An appendix contains a discussion of measures which minimize the distance between two matrices.