Often analysts must conduct risk analysis based on a small number of observations. This paper describes and illustrates the use of a kernel density estimation procedure to smooth out irregularities in such a sparse data set for simulating univariate and multivariate probability distributions.

The aim of this study was to compare production and policy risk of organic, integrated and conventional cropping systems in Norway. Experimental cropping system data (1991-1999) from eastern Norway were combined with budgeted data. Empirical distributions of total farm income for different...