Accounting for Weather Probabilities in Crop Insurance Rating

The US crop insurance program previously used a simple average of equally weighted historical loss cost data to serve as the backbone for estimating crop insurance premium rates. This article develops a procedure for weighting the historical loss cost experience based on longer time-series weather information and improve statistical validity of estimated premium rates. It was determined that the best weather data to account for weather probabilities in crop insurance premium rating is the National Climatic Data Center’s Time Bias Corrected Divisional Temperature-Precipitation-Drought Index data, also called the Climate Division Data. Using fractional logit and out-of-sample competitions, weather variables can be selected to construct an index that would allow proper assessment of the relative probability of weather events that drive production losses and to construct proper “weather weights” that can be applied when averaging historical loss cost data to calculate rates. A variable width binning approach with equal probabilities was determined as the best approach for classifying each year in the shorter historical loss cost data used for rating. When the weather weighting approach described above is applied, we find that for apples, barley, cotton, potatoes, rice, and spring/winter wheat, the weather weighted average loss costs at the national level tend to be smaller than the calculated average loss costs without weather weighting. However, for corn, cotton, sorghum, and soybeans, the weather weighted average loss costs at the national level tend to be larger. Around 51% of the counties have weather weighted average loss costs lower than the average loss costs without weather weighting


Issue Date:
2013
Publication Type:
Conference Paper/ Presentation
PURL Identifier:
http://purl.umn.edu/156825
Total Pages:
31
JEL Codes:
G22; Q10; Q18




 Record created 2017-04-01, last modified 2017-08-22

Fulltext:
Download fulltext
PDF

Rate this document:

Rate this document:
1
2
3
 
(Not yet reviewed)