This study aims to determine whether areas and intervals with lower average precipitation exhibit greater rainfall index variability in the Pasture, Rangeland and Forage Rainfall Index (PRF-RI) program, and whether this variability widens the gap between expected payouts and premium rates, thereby raising loss ratios.
Using PRF-RI and precipitation data for 66 South Dakota counties (1948–2024), we apply a linear mixed model that accounts for correlation among overlapping bi-monthly intervals to relate index variability (coefficient of variation) to long-run average precipitation. We then link this variability to expected payouts and loss ratios by combining the PRF-RI payment calculation factor with 2024 premium rates, comparing outcomes under the empirical distribution of the rainfall index and a truncated normal distribution. Finally, we estimate the relationship between average annual precipitation and cumulative loss ratios (2016–2024) across 43 states.
We find that intervals and regions with lower average precipitation exhibit greater variability in the PRF-RI. Loss ratios are higher in drier intervals and regions, but this pattern is not driven by non-normality alone: the truncated normal assumption overstates expected payouts relative to the empirical distribution and can push premium rates too high. Across states, average annual precipitation is negatively related to cumulative loss ratios – states where recent precipitation (2016–2024) has fallen below long-run averages tend to experience higher losses.
This paper fills a gap in understanding PRF-RI program design by showing how long-run precipitation levels and bi-monthly rainfall variability widen the gap between expected payouts and premium rates and affect actual loss ratios – offering insights for premium rating, interval design, and the program's long-term viability under climate change.
