1

I'm struggling to understand what a fat tail in the error distribution means intuitively. Therefore I am also not completely sure, why this is relevant for nonparametric estimation. Clearly we are estimating the conditional mean of the dependent variable, so the error will disappear, provided $E(\epsilon_{i}|x_{i}) = 0$. So when using an asymptotically consistent estimator, does the fat tail simply mean that convergence takes longer, i.e. finite sample bias is larger? Or does it have to do something with the standard error?

Best Thomas

Hello
  • 11
  • 1
  • Have a look at this: http://stats.stackexchange.com/questions/120776/why-should-we-use-t-errors-instead-of-normal-errors/120787#120787 for some intuitive explanations. – kjetil b halvorsen Oct 01 '15 at 08:12

0 Answers0