Suppose G is a lottery where the payoff is equal to -1 with 0.5 probability and 1 with 0.5 probability. I'm trying to show that G is a mean preserving spread of a lottery with uniform distribution on [-1,1] and (stronger) that given any c.d.f. on [-1,1], G is a mean preserving spread of that c.d.f. Any help on either problem would be greatly appreciated.

I understand intuitively how this is true, but I'm having a hard time formalizing this into a proof. I haven't seen an example of how to show a lottery is a mean preserving spread of another.


1 Answer 1


Let $F$ be the cdf of the uniform distribution on $[-1,1]$. Since $F$ and $G$ have the same mean, $G$ is a mean-preserving spread of $F$ if and only if \begin{equation} \int_{-\infty}^x F(t)\mathrm dt \le \int_{-\infty}^x G(t)\mathrm dt\, \qquad \text{for all }x\in[-1,1]. \end{equation} Proving the inequality should be straightforward.

The key condition here is $F$ and $G$ have the same mean. But if $F$ is any cdf on $[-1,1]$, the "mean-preserving" part will not always hold. For example, suppose $F$ is a distribution that gives $-1$ a probability of $0.3$ and $1$ a probability of $0.7$. Then $G$ cannot be a "mean-preserving" spread of $F$ because the two have different means. (However, you can still say that $F$ second order stochastically dominates $G$, which is a slightly more general concept than MPS.)


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.