Wajdi BelgacemApplied Economist
Selected projects

Insurance evaluation · Geospatial analysis

Insurance Performance & Risk Assessment

Evaluating whether rainfall-index insurance responds to forage shortfalls in the Northern Great Plains.

Program
Pasture, Rangeland & ForageRainfall-index insurance
Region
Northern Great PlainsNebraska · North Dakota · South Dakota
Study period
1990–202435 years of historical data
Analytical sample
548 geographic grids19,180 grid-year observations

The problem

Rainfall-index insurance triggers payments using rainfall measurements. A producer’s forage production can fall even when the rainfall index does not trigger a payment. This mismatch is one form of basis risk.

The study evaluates how closely the rainfall index tracks forage production, how often payment triggers miss forage shortfalls, and how calculated indemnities compare with estimated losses.

Findings

How closely rainfall tracks forage production.

Map of grid-level historical rainfall–forage model fit in Nebraska, North Dakota, and South Dakota. R-squared ranges from zero to 0.7; higher values indicate a closer fit.
Historical model fit by insurance grid. Higher R² values indicate a closer fit between the selected rainfall indices and the forage-production index. White cells are outside the analyzed sample. This is not a forecast of individual farm losses.
View full-size figure
≈40%

of annual forage variation explained, on average

Even each grid’s best-fitting rainfall-interval pair leaves substantial forage variation unexplained. The strength of the relationship varies across the region.

Historical fit within the study sample; not out-of-sample forecasting accuracy.

8.4%

average probability of no payment trigger during a qualifying forage shortfall

At 90% coverage, when the forage index fell below 90% of its historical baseline, the estimated average probability that neither selected interval triggered a payment was 8.4%. Each grid uses its own best-fitting interval pair.

Table 2; average across 548 grids. State averages: Nebraska 10.0%, South Dakota 7.5%, North Dakota 6.0%.

Analytical approach

  1. Data preparation

    I combined Rangeland Analysis Platform forage estimates, USDA RMA rainfall-index records, and land-cover information, then constructed a growing-season forage index relative to each grid’s historical production baseline.

  2. Rainfall-interval evaluation

    I estimated models for 45 admissible pairs of non-overlapping rainfall intervals per grid and identified the pair with the highest historical fit to forage production, measured by R².

  3. Insurance performance assessment

    I estimated the frequency of missed payment triggers during forage shortfalls and compared calculated indemnities with estimated forage-loss values under the study’s contract assumptions.

Practical relevance

The results show where the rainfall index tracks forage production closely and where it does not. They can be used to test alternative interval choices, weights, and program designs, and to develop location-specific guidance.

Interpreting the results

The results are historical, grid-level estimates using satellite-derived forage production, rather than individual farm yield or claims records. The analysis assumes two equally weighted intervals and a productivity factor of 1.

The best-fitting intervals are selected using the historical sample; their fit is not a measure of out-of-sample forecasting accuracy. A triggered payment does not necessarily compensate the full estimated loss.