Aggregate Technical Efficiency and Water Use in U.S. Agriculture

In the United States freshwater withdrawals for agriculture account for 80% of all out of stream water withdrawals from 1985 to 2005. To assess what drives water use in agriculture, we use the two error stochastic frontier analysis model of Battese and Coelli (1995) to estimate a translog production frontier for agriculture at the state level. The inclusion of non-negative technical inefficiency eff ects allows us to account for additional characteristics in our assessment of production inefficiency. The average marginal value of irrigation is $380, while we fi nd that on average $1 of intermediate inputs provides $0.96 of final output. These results are driven by a small subset of states with large negative values, indicating persistent misallocation of resources. The inefficiency eff ects regression finds that government subsidies increase in value of output of 0:083 per real dollar of subsidies and that that shifts from larger acreage farms to lower acreage ones will generally be efficiency increasing. This analysis highlights di fferences in water use and how they can have major implications for farm policy as a whole. Of particular note is the measured positive correlation between having a negative marginal product of intermediates and having a positive marginal product of irrigation, which suggests that shifts in inputs from intermediates to irrigation are a ripe target for efficiency gains in many states.