I try to replicate the table 8.1 of the book "Econometric Analysis of Panel Data" (Baltagi, 3rd. Edition)
For the FD2SLS estimation I get the wrong answer.
. xtivreg lnc lnp lnpn lny (l.lnc=l.lnp l.lnpn l.lny) ydum2-ydum30, fd
First-differenced IV regression Number of obs = 1288
Group variable: state Number of groups = 46
R-sq: within = 0.7214 Obs per group: min = 28
between = 0.9993 avg = 28.0
overall = 0.7746 max = 28
chi2(31) = 517.04
corr(u_i, Xb) = 0.5481 Prob > chi2 = 0.0000
------------------------------------------------------------------------------
d.lnc | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
lnc |
LD. | .2667006 .0664638 4.01 0.000 .1364341 .3969672
lnp |
D1. | -.3857637 .0287328 -13.43 0.000 -.442079 -.3294485
lnpn |
D1. | .01763 .0374006 0.47 0.637 -.0556738 .0909339
lny |
D1. | .1684187 .046402 3.63 0.000 .0774725 .2593649
------------------------------------------------------------------------------------------------
The correct answer is much bigger than mine, and I do not understand why?
The modified data is uploaded for your convenience.
I apprecaite your help
- cigar.dta