Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Extremely low R^2 (lower R^2 for FE as compared to Pooled OLS)

    Hi!

    My question might be too basic but I have very limited understanding of econometrics. Therefore, any help will be appreciated.

    These are my regressions:

    reg ROE_w LTD_w Q_w SIZE_w INV_w
    estimate store PO4

    *RE estimate
    xtreg ROE_w LTD_w Q_w SIZE_w INV_w

    *Test of Pool and RE
    xttest0

    *FE estimate
    xtreg ROE_w LTD_w Q_w SIZE_w INV_w, fe

    *RE Estimate
    xtreg ROE_w LTD_w Q_w SIZE_w INV_w, re

    *Hausman Test
    xtreg ROE_w LTD_w Q_w SIZE_w INV_w, fe
    estimate store FE4
    xtreg ROE_w LTD_w Q_w SIZE_w INV_w, re
    estimate store RE4
    hausman FE4 RE4

    Now theoretically, FE should have a higher R^2 and lower MSE. However, for my case, the R^2 for Pooled is 0.03 and for FE is 0.01 (this is for within)

    My questions are:

    1. The R^2 are extremely low - what could be the issue? How to resolve these or provide an explanation (if not solve)?
    2. How come the R^2 for FE is coming out to be lower (that too only 0.01). The Hausman test indicates that FE is the better model to use. How to solve this?

    The Hausman test results is as follows:

    Test: Ho: difference in coefficients not systematic

    chi2(4) = (b-B)'[(V_b-V_B)^(-1)](b-B)
    = 24.98
    Prob>chi2 = 0.0001


    I will be very grateful if someone can help me out in this. I have 4 days to submit my thesis and I can't figure out these things.

    Thank you!
    Last edited by Babar Javed; 16 May 2021, 15:10. Reason: Fixed-effects
Working...
X