Local Impact of Hurricane Exposure on Agricultural Insurance Behavior and Losses: Evidence from Spatial Boundary Discontinuities

Understanding how hurricanes translate into agricultural insurance losses is important for crop insurance design and agricultural risk management. We estimate the local causal effect of hurricane wind exposure using a spatial boundary regression discontinuity design. Using NOAA best-track and wind-radius data, we construct realized hurricane wind-field boundaries and compare counties just inside and just outside these boundaries within narrow geographic bandwidths. Combining these exposure measures with county-level crop insurance data from the USDA Risk Management Agency over 2004–2024, we find that crossing into the realized 34-knot wind field increases the hurricane loss-cost ratio by approximately 0.85 percentage points at the boundary. A broader storm-related loss measure produces a larger estimate, suggesting that narrow hurricane cause-of-loss categories may understate the insured loss response. A fuzzy RD specification further indicates that insured losses increase with within-county exposure share, and crop-specific estimates show meaningful heterogeneity across major crops. Estimates at the 64-knot boundary are larger but less precise. Descriptive comparisons with HIP-WI suggest that its effective indemnity footprint may not fully align with the broader realized exposure footprint over which agricultural losses occur. Our findings provide local causal evidence on hurricane losses and inform hurricane-specific crop insurance design.