'Follow the Data' — What Data Says About Real-world Behavior in Commons Problems

We test the game-theoretic foundations of common-pool resources using an individual-level dataset of groundwater usage that accounts for 3% of US irrigated agriculture. Using necessary and sufficient revealed preference tests for dynamic games, we find: (i) a rejection of the standard game-theoretic arguments based on strategic substitutes, and instead (ii) support for models building on reciprocity-like behavior and strategic complements. By estimating strategic interactions directly, we find that reciprocity-like interactions drive behavior more than market and climate trends. Taken together, we take a step toward developing more realistic models to understand groundwater usage, and related issues pertaining to tragedy of the commons and commons governance.