Using the mills ratio result, let $X \sim N(\mu, \sigma^2)$, then

$E(X| X<\alpha) = \mu - \sigma\frac{\phi(\frac{a- \mu}{\sigma})}{\Phi(\frac{a-\mu}{\sigma})}$

However, when calculating it in R. I don't obtain the correct results as

> mu <- 1
> sigma <- 2 
> a <- 3 
> x <- rnorm(1000000, mu, sigma) 
> x <- x[x < a] 
> mean(x)
[1] 0.4254786
> mu -  sigma * dnorm(a, mu, sigma) / pnorm(a, mu, sigma)
[1] 0.7124

What am I doing wrong?


Your formula implementation is wrong because, $$\phi\left(\frac{x-\mu}{\sigma}\right)=\frac{1}{\sqrt{2\pi}}e^{-\frac{1}{2}\left(\frac{x-\mu}{\sigma}\right)^2}\neq f_{X,\mu,\sigma}(x)=\frac{1}{\sqrt{2\pi}\sigma}e^{-\frac{1}{2}\left(\frac{x-\mu}{\sigma}\right)^2}$$ As you can see, we have an extra $\sigma$ in the denominator of $f_{X,\mu,\sigma}(x)$, which yields: $$\phi\left(\frac{x-\mu}{\sigma}\right)=\sigma f_{X,\mu,\sigma}(x)$$ dnorm method gives you $f_{X,\mu,\sigma}(x)$, where you need to multiply it with $\sigma$ to obtain $\phi$. Since your $\sigma=2$, this can be practically done via subtracting the second term again, which is $1-0.7124=0.2876$: $$1-0.2876-0.2876=0.4247$$ which is close to your estimate.

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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