Seminonparametric Estimation of Conditional Yield Densities

Government sponsored agricultural insurance programs are susceptible to losses due to moral hazard and adverse selection problems. To lessen the problems, the Group Risk Plan (GRP) of insurance has been offered by the Federal Crop Insurance Corporation since 1993. Under this product, losses are measured on the basis of a county’s mean yield. Accurate pricing of insurance products like GRP requires the distribution of future yields conditional on the current information set. The main objective of this thesis is to improve the accuracy of premium rates thus improving the efficiency of GRP program. In satisfying this objective, a seminonparametric (SNP)
maximum likelihood method is utilized in an attempt to reduce program inefficiencies induced by distributional assumptions in determining premium rates. A spline model with one knot point is used to capture the central tendency and to predict mean yields. A mixture of two normal distributions is used to represent the disturbance distribution.

National Agricultural Statistics Service (NASS) county mean yield data over the period 1955 to 2007 of 102 Illinois counties are used in the analysis. The crop analyzed is corn for grain. The premium rates are calculated numerically from the recovered yield distributions of each county. The SNP maximum likelihood approach makes more efficient use of the data in that yield correlation among counties is explicitly modeled. As a result, SNP rates tend to be more consistent among counties.

Hao Lu_Masters Thesis_Final.pdf (1.27 MB)

Author(s)

Publication Date