Incorrect (?) confidence intervals when distribution is not symmetric

ZoltanRacz / MomentMatching.jl

This is a package to estimate parameters of macroeconomic models by moment matching methods, such as GMM or SMM.

MIT License

1 stars 0 forks source link

Incorrect (?) confidence intervals when distribution is not symmetric #48

Open ZoltanRacz opened 9 months ago

ZoltanRacz commented 9 months ago

If I understand correctly, merge #43 introduced what Christensen calls Efron's confidence intervals in section 5.4. He seems to think that this is incorrect when the distribution of the statistic is asymmetric. Do you agree?

I haven't looked in the more sophisticated solution, but we might want to implement that one later on. I don't think this is extremely urgent though. @gualtiazza

ZoltanRacz commented 9 months ago

Added milestone 0.2.0, since if this is indeed incorrect in general, we have to fix it soon. Otherwise, we can remove the milestone