On Sunday, Daniel asked:
> I hope I'll find somebady which is able to get through a specification
> problem with panel data.
>
> I want to compute a two-way random and fixed effect model like this one:
>
> yi,t= a0+consi+const+ b1Xi,t+epsiloni,t
>
> where a0 is a common constant and consi is a country specific constant
> and const is a specific time constant.
>
> I used xtreg y x dummy (for time period),re (fe) i(country) but I'm not
> sure that it's the correct way to formulate this two-way model...
Svetlana suggested:
> You can try
>
> xi: reg y i.firm i.time x1
> or
> xi: reg y i.firm i.time x1, robust
>
> for the robust version.
You can certainly used -xtreg,fe- and include the time dummies. You will get
the same results as the ones obtained using -xi:regress- See the example
below
. clear
. set mem 10m
(10240k)
. webuse nlswork
(National Longitudinal Survey. Young Women 14-26 years of age in 1968)
. keep if id<=20
(28357 observations deleted)
. tsset idcode year
panel variable: idcode, 1 to 20
time variable: year, 68 to 88, but with gaps
. xi:xtreg ln_w age tenure i.year, fe i(idcode)
i.year _Iyear_68-88 (naturally coded; _Iyear_68 omitted)
Fixed-effects (within) regression Number of obs =
175
Group variable (i): idcode Number of groups =
18
R-sq: within = 0.3069 Obs per group: min =
2
between = 0.0696 avg =
9.7
overall = 0.0312 max =
15
F(16,141) =
3.90
corr(u_i, Xb) = -0.7785 Prob > F =
0.0000
----------------------------------------------------------------------------
--
ln_wage | Coef. Std. Err. t P>|t| [95% Conf.
Interval]
-------------+--------------------------------------------------------------
--
age | .2163456 .1259685 1.72 0.088 -.0326854
.4653766
tenure | .029208 .0080542 3.63 0.000 .0132854
.0451307
_Iyear_69 | -.0610705 .2066732 -0.30 0.768 -.4696492
.3475083
_Iyear_70 | -.4496718 .2829069 -1.59 0.114 -1.008959
.1096159
_Iyear_71 | -.6838077 .3932548 -1.74 0.084 -1.461245
.09363
_Iyear_72 | -.8756372 .5113781 -1.71 0.089 -1.886597
.1353222
_Iyear_73 | -1.005979 .6235475 -1.61 0.109 -2.23869
.2267316
_Iyear_75 | -1.471403 .8757545 -1.68 0.095 -3.20271
.2599033
_Iyear_77 | -1.941271 1.122908 -1.73 0.086 -4.161183
.2786407
_Iyear_78 | -2.076393 1.248898 -1.66 0.099 -4.545379
.3925926
_Iyear_80 | -2.601959 1.497365 -1.74 0.084 -5.562148
.3582295
_Iyear_82 | -3.131761 1.746915 -1.79 0.075 -6.585293
.321771
_Iyear_83 | -3.135278 1.87198 -1.67 0.096 -6.836054
.5654988
_Iyear_85 | -3.441114 2.120773 -1.62 0.107 -7.633738
.7515099
_Iyear_87 | -4.010854 2.372786 -1.69 0.093 -8.70169
.6799811
_Iyear_88 | -4.251622 2.571628 -1.65 0.100 -9.335555
.8323111
_cons | -2.3679 2.420624 -0.98 0.330 -7.153307
2.417507
-------------+--------------------------------------------------------------
--
sigma_u | .51542996
sigma_e | .25797362
rho | .79967882 (fraction of variance due to u_i)
----------------------------------------------------------------------------
--
F test that all u_i=0: F(17, 141) = 12.34 Prob > F =
0.0000
. xi:regres ln_w age tenure i.year i.idcode
i.year _Iyear_68-88 (naturally coded; _Iyear_68 omitted)
i.idcode _Iidcode_1-20 (naturally coded; _Iidcode_1 omitted)
Source | SS df MS Number of obs =
175
-------------+------------------------------ F( 33, 141) =
9.45
Model | 20.7606901 33 .62911182 Prob > F =
0.0000
Residual | 9.38360513 141 .066550391 R-squared =
0.6887
-------------+------------------------------ Adj R-squared =
0.6159
Total | 30.1442952 174 .173243076 Root MSE =
.25797
----------------------------------------------------------------------------
--
ln_wage | Coef. Std. Err. t P>|t| [95% Conf.
Interval]
-------------+--------------------------------------------------------------
--
age | .2163456 .1259685 1.72 0.088 -.0326854
.4653766
tenure | .029208 .0080542 3.63 0.000 .0132854
.0451307
_Iyear_69 | -.0610705 .2066732 -0.30 0.768 -.4696492
.3475083
_Iyear_70 | -.4496718 .2829069 -1.59 0.114 -1.008959
.1096159
_Iyear_71 | -.6838077 .3932548 -1.74 0.084 -1.461245
.09363
_Iyear_72 | -.8756372 .5113781 -1.71 0.089 -1.886597
.1353222
_Iyear_73 | -1.005979 .6235475 -1.61 0.109 -2.23869
.2267316
_Iyear_75 | -1.471403 .8757545 -1.68 0.095 -3.20271
.2599033
_Iyear_77 | -1.941271 1.122908 -1.73 0.086 -4.161183
.2786407
_Iyear_78 | -2.076393 1.248898 -1.66 0.099 -4.545379
.3925926
_Iyear_80 | -2.601959 1.497365 -1.74 0.084 -5.562148
.3582295
_Iyear_82 | -3.131761 1.746915 -1.79 0.075 -6.585293
.321771
_Iyear_83 | -3.135278 1.87198 -1.67 0.096 -6.836054
.5654988
_Iyear_85 | -3.441114 2.120773 -1.62 0.107 -7.633738
.7515099
_Iyear_87 | -4.010854 2.372786 -1.69 0.093 -8.70169
.6799811
_Iyear_88 | -4.251622 2.571628 -1.65 0.100 -9.335555
.8323111
_Iidcode_2 | -.4052557 .1065452 -3.80
0.000 -.6158883 -.1946231
_Iidcode_3 | -1.779763 .7502473 -2.37
0.019 -3.26295 -.2965755
_Iidcode_4 | -1.367843 .762298 -1.79 0.075 -2.874854
.1391677
_Iidcode_5 | -1.435239 .7573151 -1.90 0.060 -2.932399
.0619212
_Iidcode_6 | -1.471493 .6282069 -2.34
0.021 -2.713415 -.2295713
_Iidcode_7 | -1.251497 .3362335 -3.72
0.000 -1.916208 -.5867862
_Iidcode_9 | -.1895229 .1203334 -1.57 0.118 -.4274137
.0483679
_Iidcode_10 | -.6424414 .1163439 -5.52
0.000 -.8724453 -.4124375
_Iidcode_12 | -.3507719 .4964787 -0.71 0.481 -1.332276
.6307325
_Iidcode_13 | -.6464649 .5042873 -1.28 0.202 -1.643406
.3504766
_Iidcode_14 | -.8292302 .5097123 -1.63 0.106 -1.836897
.1784361
_Iidcode_15 | -.2628088 .3821977 -0.69 0.493 -1.018388
.49277
_Iidcode_16 | -.6522392 .3865861 -1.69 0.094 -1.416494
.1120152
_Iidcode_17 | -.4976363 .3920352 -1.27 0.206 -1.272663
.2773905
_Iidcode_18 | -1.359985 .4139419 -3.29
0.001 -2.17832 -.5416504
_Iidcode_19 | -.7293873 .3909297 -1.87 0.064 -1.502228
.0434539
_Iidcode_20 | -.6100004 .3804356 -1.60 0.111 -1.362096
.1420947
_cons | -1.534948 2.045132 -0.75 0.454 -5.578033
2.508138
----------------------------------------------------------------------------
--
You can also use the -areg- command and include the time dummies. -xtreg,fe-
does not support the robust option but -areg- does. See the example below,
which produces the same output as the previous two estimation results.
. xi:areg ln_w age tenure i.year, absorb(idcode)
i.year _Iyear_68-88 (naturally coded; _Iyear_68 omitted)
Number of obs =
175
F( 16, 141) =
3.90
Prob > F =
0.0000
R-squared =
0.6887
Adj R-squared =
0.6159
Root MSE =
.25797
----------------------------------------------------------------------------
--
ln_wage | Coef. Std. Err. t P>|t| [95% Conf.
Interval]
-------------+--------------------------------------------------------------
--
age | .2163456 .1259685 1.72 0.088 -.0326854
.4653766
tenure | .029208 .0080542 3.63 0.000 .0132854
.0451307
_Iyear_69 | -.0610705 .2066732 -0.30 0.768 -.4696492
.3475083
_Iyear_70 | -.4496718 .2829069 -1.59 0.114 -1.008959
.1096159
_Iyear_71 | -.6838077 .3932548 -1.74 0.084 -1.461245
.09363
_Iyear_72 | -.8756372 .5113781 -1.71 0.089 -1.886597
.1353222
_Iyear_73 | -1.005979 .6235475 -1.61 0.109 -2.23869
.2267316
_Iyear_75 | -1.471403 .8757545 -1.68 0.095 -3.20271
.2599033
_Iyear_77 | -1.941271 1.122908 -1.73 0.086 -4.161183
.2786407
_Iyear_78 | -2.076393 1.248898 -1.66 0.099 -4.545379
.3925926
_Iyear_80 | -2.601959 1.497365 -1.74 0.084 -5.562148
.3582295
_Iyear_82 | -3.131761 1.746915 -1.79 0.075 -6.585293
.321771
_Iyear_83 | -3.135278 1.87198 -1.67 0.096 -6.836054
.5654988
_Iyear_85 | -3.441114 2.120773 -1.62 0.107 -7.633738
.7515099
_Iyear_87 | -4.010854 2.372786 -1.69 0.093 -8.70169
.6799811
_Iyear_88 | -4.251622 2.571628 -1.65 0.100 -9.335555
.8323111
_cons | -2.3679 2.420624 -0.98 0.330 -7.153307
2.417507
-------------+--------------------------------------------------------------
--
idcode | F(17, 141) = 12.342 0.000 (18
categories)
-- Gustavo
[email protected]
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
