Proofreading the Theory Section Again

ijyliu / ECMA-31330-Project

Econometrics and Machine Learning Group Project

2 stars 1 forks source link

Proofreading the Theory Section Again #70

Closed ijyliu closed 3 years ago

ijyliu commented 3 years ago

Mostly looks very good!

[x] I think the toy "For a more concrete example" section isn't needed, since it's been superseded with an actual proof
[x] Unclear to me how 20 shows Z is exogenous
[ ] I guess if there is a part you want to discuss with nadav it's the principal components derivation part, the rest looks very standard and looks right to me
[ ] there's a typo in the statement under (27)? I don't think you mean (13)? I guess the equation numbering on the version of the draft I'm looking at is kind of messed up.
[x] Are we keeping the "Estimation of the Principal Component Regression" section? Not clear it leads to anything at the moment?

nicomarto commented 3 years ago

exogeinity proof was taken directly from Shaik's notes of EAI

ijyliu commented 3 years ago

I’m not saying it’s not true, just that it’s not too clear in my opinion

On Fri, May 28, 2021 at 3:11 PM nicomarto @.***> wrote:

exogeinity proof was taken directly from Shaik's notes of EAI

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/ijyliu/ECMA-31330-Project/issues/70#issuecomment-850618060, or unsubscribe https://github.com/notifications/unsubscribe-auth/AQCGE4PDQUVM4HEDYZ3F5C3TP7TGHANCNFSM45VF557A .

nicomarto commented 3 years ago

Oh I understand you wrote it is unclear, I just was letting you know that I took it from his lecture notes so I believe it is sufficient given that it is a proof given in a lecture. Which par of the proof of E(Z_ie_i^*) is not clear to you?

ijyliu commented 3 years ago

eq. (19)

= 0 right? clear E(we) = 0 and E(wn) = 0 clear from the covariances. But E(Xe) and E(Xn) = 0 is a little less clear. Though I guess it's related to assuming classical measurement error

nicomarto commented 3 years ago

oh, yes, I forgot the =0. E(Xe)=0 comes for the assumption of the real DGP, if we had X instead of X^* then we would have no problem (see eq 1). E(Xn)=0 comes from classical measurement error as you suggest!

nicomarto commented 3 years ago

@ijyliu let me know if now is clearer

ijyliu commented 3 years ago

@nicomarto did you change anything? is the latest pdf on github updated? (proof of claim 2.1 looks the same to me as earlier)

also it seems like the equation numbers are still messed up

ijyliu commented 3 years ago

eh this is not too important it's just in the appendix anyway and we know it's right