The design of weather index insurance for forage: the case of basis risk for the Canadian province of Ontario

Abstract

This thesis examines weather index area-yield basis risk for forage insurance. The first focus of the research is to determine which weather variables (e.g. rainfall, temperature, sunshine, etc) should be included in the multivariable weather index, given the limited yield data and multicollinearity among weather variables. The second focus is to analyze the effect of the geographical scale (number of counties used in the index) on basis risk. Daily weather data and actual forage yield are from Ontario’s rainfall index-based forage insurance plan. Both principal component regression (PCR) and partial least squares regression (PLSR) are used to select the weather index variables. Results show that the two regression models generate similar weather variable selection for the weather index. Both models can be considered suitable depending on the choice of criteria. Further, the results show that as the number of counties of the index decreases, area-yield basis risk is reduced substantially.