Probably a naive question. I am quoting this from Greene's econometrics book:

"The occasional statement that the properties of the MLE are only optimal in large samples is not true, however. It can be shown that when sampling is from an exponential family of distributions, there will exist sufficient statistics. If so, MLEs will be functions of them, which means that when minimum variance unbiased estimators exist, they will be MLEs. [See Stuart and Ord (1989).] Most applications in econometrics do not involve exponential families, so the appeal of the MLE remains primarily its asymptotic properties."

So why the variance estimate is downward biased in MLE for a normal distribution?

  • 2
    $\begingroup$ Why don't you post here: stats.stackexchange.com This is a statistics question. $\endgroup$ – Ric Mar 12 '15 at 15:43
  • 2
    $\begingroup$ The MLE variance estimator for normal distributions is biased because it divides by $n$ rather than $n-1$, see ee.columbia.edu/~dliang/files/mle_biased.pdf. Not sure how much that relates to the above quote. $\endgroup$ – John Mar 12 '15 at 18:15
  • $\begingroup$ You argument is irrelevant to my question. The statement is that if there is a sufficient statistic for an exponential distribution that would be MLE. $\endgroup$ – Amir Yousefi Mar 13 '15 at 17:59
  • $\begingroup$ In fact this sentence "which means that when minimum variance unbiased estimators exist, they will be MLEs." is what does not make sense. $\endgroup$ – Amir Yousefi Mar 13 '15 at 18:03
  • $\begingroup$ Richard - I agree to some sense. However, I see people using MLE a lot in practice, but the statement indicates that the estimates are not close enough for small samples and specially for non-expoential distributions. $\endgroup$ – Amir Yousefi Mar 13 '15 at 18:08

Your Answer

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

Browse other questions tagged or ask your own question.