Hydropower Droughts Reveal Hidden Adjustment Risks in Global Electricity Transitions

We combine a 2005–2023 global country-year panel with capacity-weighted exposure from 7,156 hydroelectric plants and an independent crosswalk linking 3,516 plants to ReGeom/GRanD reservoirs within 25 km. Country and year fixed-effects models estimate conditional associations between reservoir-location scPDSI stress, predetermined hydropower dependence and electricity-system adjustment, with two-way clustered uncertainty, pre-specified Holm/BH-FDR correction, 999-country bootstrap refits, leave-one-out tests, spatial holdouts and a frozen failure backtest.

In 1,937 country-years across 102 countries, the stress–dependence interaction is positive for adjustment burden (0.00949; nominal p = 0.021; BH-FDR q = 0.0315; Holm p = 0.107; 95% CI, 0.00146–0.01751), negative for the hydropower margin (−0.00180; nominal p = 0.019), positive for net imports (0.00244; nominal p = 0.020) and positive for electricity carbon-intensity growth (0.00229; nominal p = 0.018).