[Carlo Lazzaro]

Niels:
cross-sectional interdependence can be acccomodate via (say) the user-written programme -xtscc- (type -search xtscc- from withn Stata to install it).
The following should help: http://www.stata-journal.com/sjpdf.h...iclenum=st0113

[Niels Meijer]

Thank you very much Carlo

[Martin Knipp]

Niels:
before switching to -xtgls- despite havibng a large N, small T dataset,, please note the dramatically different times (in seconds) taken by -xtreg- and -xtgls- to perform the same simple panel data regression:

. set rmsg on
r; t=0.00 15:48:21

. xtreg ln_wage i.race, re

Random-effects GLS regression Number of obs = 28,534
Group variable: idcode Number of groups = 4,711

R-sq: Obs per group:
within = 0.0000 min = 1
between = 0.0198 avg = 6.1
overall = 0.0186 max = 15

Wald chi2(2) = 99.02
corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0000

------------------------------------------------------------------------------
ln_wage | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
race |
black | -.1300382 .013486 -9.64 0.000 -.1564702 -.1036062
other | .1011474 .0562889 1.80 0.072 -.0091768 .2114716
|
_cons | 1.691756 .0071865 235.41 0.000 1.677671 1.705841
-------------+----------------------------------------------------------------
sigma_u | .38195681
sigma_e | .32028665
rho | .58714668 (fraction of variance due to u_i)
------------------------------------------------------------------------------
r; t=0.61 15:48:28

. xtgls ln_wage i.race

Cross-sectional time-series FGLS regression

Coefficients: generalized least squares
Panels: homoskedastic
Correlation: no autocorrelation

Estimated covariances = 1 Number of obs = 28,534
Estimated autocorrelations = 0 Number of groups = 4,711
Estimated coefficients = 3 Obs per group:
min = 1
avg = 6.056888
max = 15
Wald chi2(2) = 542.80
Log likelihood = -19162 Prob > chi2 = 0.0000

------------------------------------------------------------------------------
ln_wage | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
race |
black | -.1427862 .006243 -22.87 0.000 -.1550222 -.1305502
other | .080671 .0274112 2.94 0.003 .026946 .134396
|
_cons | 1.714338 .0033339 514.21 0.000 1.707804 1.720873
------------------------------------------------------------------------------
r; t=692.49 16:00:07
.

A possible work-around could be:
-skipping -xttest2- and -xttest3-;
- graphically inspect your residual distribution;
-robustify/cluster your standard errors if you suspect that (especially) heteroskedasticity can bite your results (as said, serial correlation is expected to be a minor nuisance with a short T dimension).

Otherwise, as many econometricians usually do, go -cluster-/-robust- from scratch; with 200 -panelid- you have enough clusters to survive.

I'd like to use this thread, because you are referring to it here, Carlo.
I have read this a lot recently, that serial correlation/autocorrelation is not a big problem in short panel data. Unfortunately I was not able to find a reliable (high ranked) paper giving proof to that assumption.
Carlo, would you be so kind as to cite a study if you know of one?

Thanks a lot!
Martin

[Carlo Lazzaro]

Martin:
in the post you quoted, clustering is recommended as the number of panels (i.e., cluster) is actualy relevant, no matter the T dimension (that, along with other conditions, can have a role in increasing within-cluster correlation pattern of the systematic error). You may want to take a look at:
https://www.stata.com/meeting/wcsug07/cameronwcsug.pdf and related references (actually, I'm not aware of an article that focuses specifically on this particular topic).

[Martin Knipp]

Martin:
in the post you quoted, clustering is recommended as the number of panels (i.e., cluster) is actualy relevant, no matter the T dimension (that, along with other conditions, can have a role in increasing within-cluster correlation pattern of the systematic error). You may want to take a look at:
https://www.stata.com/meeting/wcsug07/cameronwcsug.pdf and related references (actually, I'm not aware of an article that focuses specifically on this particular topic).

Thank you, Carlo.
Here it is also reffered to this fact: http://www.princeton.edu/~otorres/Panel101.pdf on slide 36.
Unfortunately, there are no references.

Cheers
Martin

[Martin Knipp]

For those interested in a reference to cite that serial correlation is not a big issue in short panels (p.332):
Akel, V., & Torun, T. (2017). Stock market development and economic growth: the case of MSCI emerging market index countries. In Global Financial Crisis and Its Ramifications on Capital Markets (pp. 323-336). Springer, Cham.