Dynamics of residential water consumption in a panel of Italian municipalities

Abstract

The purpose of this study is to estimate a residential water demand function in a dynamic framework for a panel of Italian municipalities and to calculate both short-run and long-run price elasticities. The Generalized Method of Moments (GMM) provides a suitable framework for obtaining asymptotically efficient estimators in this context. Specifically, the system GMM estimator is applied, which improves the properties of the standard first-difference estimator. The most relevant result shows that persistence of habits is coupled with high long-term price elasticity which is higher, in absolute value, that the instantaneous (one-year) price elasticity. Under an economic policy perspective, the effects of policy interventions can be suitably evaluated only in the long-run.

Acknowledgements

This study has been carried out for the Research Project Genesto (Gestione integrata e sostenibile delle risorse idriche in differenti contesti territoriali). Financed by the Italian Ministry of Education, University and Reserch, Decree 10 May 2000. We are grateful to Padania Acque for data provision. We wish to thank R. Zoboli and A. Thomas for their helpful comments. The usual caveats apply.

Notes

¹ Martínez-Espiñeira and Nauges, 2004, for instance, take into account the presence of habits in the interesting theoretical framework of a Stone-Geary utility function.

² Until the 1974 the price of water was designed by a government committee without any reference to management costs. In 1984 a norm established that any increase in the price could not be larger that the forecasted inflation rate.

³ Musolesi and Nosvelli (2005) and Mazzanti and Montini (2006) are among the few authors who estimate a water demand function for a set of Italian municipalities.

⁴ As Nauges and Thomas (2003) show, this type of dynamic formulation can be derived starting from a structural optimization programme solved by the municipality.

⁵ For T=4 and , the asymptotic variance ratio of the first-difference GMM estimator to the system GMM is 1.75 for δ = 0 and increases to 3.26 for δ = 0.5 and 55.4 for δ = 0.9.