Interview design for revealing preferences of policy makers (Q1811253)

In this paper methods are presented to represent preferences by quadratic utility functions. The methods are based on nonlinear programming results. The author defines questionnaires such that the derived data result in an unambiguous non-trivial quadratic function. In section 5 a design of the model input for estimating cardinal utility is given, whereas in section 6 ordinal utilities are considered. In the latter case the quadratic function computed can be regarded as a numerical representation of the ordinal preferences. Finally, stability properties for the ordinal approach are given. The results of the paper are illustrated by a rather artificial example concerning German economic policy. A computer model for the methods is presented.