University of California Water Resources Center

Given the importance of regional response in physical and economic terms to water resources planners, large scale regional models are a widely used component of water resource management. Given the absence of a sufficiently rich data base to estimate cost functions by econometric approaches, linear constrained optimization models have been extensively used to derive normative estimates of former response to water policy.

A long standing problems in linear models is the inevitable trade-off between the precision of calibration of the model and the constrained nature of the solution. Often model results significantly depart from empirical reality with deleterious effects on policy. This research shows that a nonlinear cost function formulation overcomes the constraint problem when the function parameters are estimated from actual farmer responses.

The Positive Quadratic Programming (PQP) theory developed was applied to the California Agricultural and Resource Model (CARM) and used to estimate demand functions for irrigation water by region.