Abstract
Multiplier analysis based upon the information contained in Leontief's inverse is undoubtedly part of the core of the input–output methodology and numerous applications and extensions have been developed that exploit its informational content, both at the national and regional levels. Nonetheless there are some implicit theoretical assumptions whose policy implications need to be assessed. This is the case for the ‘excess capacity’ assumption, which implies that resources are available as needed to adjust production to new equilibrium states. In an actual economy, however, new resources are often scarce and always costly. When supply constraints intervene, the assessment of the effects of government demand policies may be substantially different from that of the standard Leontief multiplier matrix. Using a closed general equilibrium model that incorporates supply constraints, we perform some simple numerical exercises and proceed to derive two ‘constrained’ multiplier matrices, based upon the implicit Jacobian matrix, that can be compared with the standard ‘unconstrained’ Leontief matrix.
Acknowledgments
Support from research grants MICINN-ECO2009-11857 and SGR2009-578 is gratefully acknowledged. We thank participants at the Sydney 2010 Input–Output conference for helpful comments and Mark Partridge for providing us with very useful references and the referees and editor Bart Los for their insightful comments.
Notes
1 See Miller and Blair (Citation2009, Chapter 2, .15).
2 Ballard et al. Citation(1985) report that Cobb-Douglas unitary elasticity for value-added generation is more realistic than for the Armington function. We use Cobb-Douglas throughout as an approximation to illustrate how competing multiplier matrices can be derived. Also, Cobb-Douglas and Leontief models are related in the sense that they share the same calibrated share coefficients. See El-Hodiri and Nourzad (Citation1988).
3 We adopt here the simple activity analysis approach first proposed by Kehoe and Serra-Puche Citation(1983).
4 Summation is possible because of the standard normalization of data in the model implementation: each unit is defined as having a worth of one euro.