• The editor in charge of this paper was Nicola Gennaioli.

  • Acknowledgments: We thank Daron Acemoglu, Oana Borcan, Antonio Ciccone, Carl-Johan Dalgaard, Jacob Gerner Hariri, James Robinson, Hans-Joachim Voth, David Weil, seminar participants at Brown University, ISET, Pompeu Fabra, and University of Copenhagen; participants at the Growth and Development Conference at ISI, New Delhi; the Nordic Conference in Development Economics, and the Congress of the European Economic Association; and four anonymous referees for useful comments. We gratefully acknowledge financial support from the Carlsberg Foundation and the Danish Council for Independent Research.


Irrigated agriculture makes societies more likely to be ruled by authoritarian regimes. Ancient societies have long been thought to follow this pattern. We empirically show that irrigation affects political regimes even in the present. To avoid endogeneity, we use geographical and climatic variation to identify irrigation dependent societies. We find that countries whose agriculture depended on irrigation are about six points less democratic on the 21-point polity2 scale than countries where agriculture has been rainfed. We find qualitatively similar results across regions within countries. We argue that the effect has historical origins: irrigation allowed landed elites in arid areas to monopolize water and arable land. This made elites more powerful and better able to oppose democratization. Consistent with this conjecture, we show that irrigation dependence predicts land inequality both at the country level, and in premodern societies surveyed by ethnographers. (JEL: O11, N50, Q15)