We study the effect of water scarcity decrees (WSDs) on the prices of water property rights (WPRs) in Chile. Using a quasi-experimental design, we estimate a probabilistic policy rule via a machine learning algorithm trained on two official proxies of water supply, the standardized precipitation index, and the river flow index, which guided WSD enforcement. This predicted rule serves as an instrument for actual WSDs. We find a price increase of 9.9% per decree, with stronger effects from the first treatments. The results suggest a negative risk premium, consistent with the role of WPRs as hedge assets against climate risk.