Prakash, Chang, Hamid, and Smyser (1996) purport to show that with suf
ficient preference for positive skewness, risk-averse managers might e
lect to engage in unfair gambles. Their ''proof'' of the impossible as
sumes the existence of a differentiable risk-preference function that
satisfies certain assumptions about the first three derivatives. The f
undamental flaw in their argument is that no differentiable function e
xists for which risk aversion and the participation in an unfair gambl
e are compatible.