Application of Generalised Inverse to Regionalisation of National Input-Output Tables

Abstract

This paper presents a methodology for deriving regional input-output (I-O) coefficients from national I-O tables using the Generalised Inverse of matrices, addressing the lack of regional survey data. Existing non-survey techniques, such as location quotients, representative coefficients, and the RAS method, often produce inconsistent results. The proposed approach constructs a block diagonal matrix of regional weights, allowing reliable estimation of regional coefficients even when some data are missing. Empirical analysis for 21 Indian regions demonstrates approximately 46% conformity with observed values and explains 65% of variations through regression. The study highlights a practical framework for regional economic planning in data-scarce contexts.