UC Berkeley

We study the optimality of human groundwater extraction from many of the world's 3,400 aquifers, where institutional or market failures may produce non-optimal extraction decisions. We use remote sensing and economic data to estimate a dynamic model of water extraction for each aquifer, recover the discount factor that rationalizes observed groundwater extraction, and compare it against Ramsey and market benchmarks. Three- to four-fifths of aquifers are extracted more rapidly than is optimal, decreasing aquifer present value by several percent, and creating trillions of dollars in present value welfare losses. Policies that guarantee users constant water quantities indefinitely, through subsidies or prior appropriation, or policy rules that fully discount future extraction, generate larger welfare costs. A sustainable extraction path that maintains current fill forever generates meaningfully lower welfare than the optimal extraction path, because most aquifer stocks are already inefficiently low.