Risk programming analysis with imperfect information

Gudbrand Lien

doi:10.1007/s10479-009-0555-y

Risk programming analysis with imperfect information

2009-06-12

7

OA

AI

M.A.P.M. van Asseldonk *

J

James W. Richardson

DOI：10.1007/s10479-009-0555-y

Original

Original request for help

Save

Abstract

En 中文

A Monte Carlo procedure is used to demonstrate the dangers of basing (farm) risk programming on only a few states of nature and to study the impact of applying alternative risk programming methods. Two risk programming formulations are considered, namely mean-variance (E,V) programming and utility efficient (UE) programming. For the particular example of a Norwegian mixed livestock and crop farm, the programming solution is unstable with few states, although the cost of picking a sub-optimal plan declines with increases in number of states. Comparing the E,V results with the UE results shows that there were few discrepancies between the two and the differences which do occur are mainly trivial, thus both methods gave unreliable results in cases with small samples.

Keywords:

Quadratic risk programming

States of nature

Sparse data

Kernel smoothing

Risk programming analysis with imperfect information

Abstract

Journal

Organization