Decomposition of sectoral water consumption: a subsystem SAM model for Extremadura, Spain

Though most direct water use is generally associated with agriculture, and therefore with the production of food, there is a clear consensus that considering indirect water uses is a relevant factor to be considered in any environmental analysis. In this regard, the subsystems input-output approach makes it possible to isolate the water relations of a limited number of activities as part of the entire production sphere and thus obtain specific production information of them. However, not directly related with production, the demand of goods and services closely linked to available income, also play a significant role in water consumption. Hence, in this paper we propose an extension of the input-output subsystem model to account for all transactions within a Social Accounting Matrix (SAM) and capture the entire flow by which water depletion is caused and transmitted throughout the economic system. Our approach, which divides the water consumption into different components, illustrates how the individual patterns of water depletion differ among economic agents. The starting point of the subsystems representation consists of the decomposition of the N accounts of a SAM system into two categories of institutions: m endogenous and s exogenous. By taking into account this separation of the accounts, in this paper, we analyze separately all the endogenous economic agents, and for each one, we apply a subsystem division of its water consumption. The empirical application is for the Spanish regional economy of Extremadura, and the economic and environmental data are for the year 2005. Our results show that large asymmetries exist not only in the quantitative contribution of the different economic agents to water consumption, but also in the decomposed effects of this contribution. In particular, this decomposition allows illustration of the pull effects of the SAM economic agents on the others, and this shows the effective consumption caused by one agent, no matter which is the agent that causes the end water consumption. Models such as the one presented here also have limitations, like the shortage of data or parameter specification. However, they are of great interest to define the direction that environmental or economic policies should take if we are to reduce water consumption or increase water use efficiency levels.