Distorted stochastic dominance: a generalized family of stochastic orders (Q2201711)

An alternative to expected utility model, in which the utilities are replaced by distortions of the corresponding cummulative distribution function (further abbreviated as CDF) is proposed. Distortion is a nondecreasing function \(\Phi:[0,1]\rightarrow [0,1]\) with the values \(\Phi(0)=0,\Phi(1)=1\). The authors study random variables of CDFs. CDFs are transformed through a distortion function \(H\). A semiparametric family of stochastic orders is defined. It is called \(H\)-distorted stochastic dominance (DSD) and denoted by \(\geq_H\). Relationships among \(\geq_H\) for different choices of \(H\) are studied. The same approach is applied with the aim to generalize the increasing and convex order. Further DSD, risk-loving-DSD and mixed-DSD are characterized in terms of isotonic distorted expectations. Finally some results of non-expected utility theory are generalized.