I have a panel data of 27 countries from 1995-2016 (unbalanced panel). I am confused how to interpret whether autocorrelation is a minor concern or not and whether to use the command robust. Can you please help me interpret my results? Is correlation a concern and I need to use -robust-?
I use the command -regress-
Regression without -robust-:
Wooldridge Serial Correlation test gives:
Results with -robust- :
I use the command -regress-
Regression without -robust-:
Code:
Source | SS df MS Number of obs = 299
-------------+------------------------------ F( 18, 280) = 19.12
Model | 14.153387 18 .786299278 Prob > F = 0.0000
Residual | 11.5164346 280 .041130124 R-squared = 0.5514
-------------+------------------------------ Adj R-squared = 0.5225
Total | 25.6698216 298 .086140341 Root MSE = .20281
-----------------------------------------------------------------------------------
lnPATR1 | Coef. Std. Err. t P>|t| [95% Conf. Interval]
------------------+----------------------------------------------------------------
preelection | -.034093 .0394657 -0.86 0.388 -.1117801 .0435942
lninflation | -.0082857 .0260905 -0.32 0.751 -.0596442 .0430727
lngdp | -.0214539 .0155474 -1.38 0.169 -.0520585 .0091507
lnurbanisation | .5635988 .1471487 3.83 0.000 .2739406 .8532571
left | .1474364 .0380318 3.88 0.000 .0725718 .222301
right | .1450968 .0386415 3.75 0.000 .069032 .2211616
lndependency | 2.259958 4.325274 0.52 0.602 -6.254226 10.77414
lndependencyold | -4.3537 12.34348 -0.35 0.725 -28.65151 19.94411
lndependencyyoung | -2.010141 .9168129 -2.19 0.029 -3.814862 -.2054197
lnpopbelow14 | 0 (omitted)
lnpop1564 | 0 (omitted)
lnpop65andabove | 3.49226 12.26131 0.28 0.776 -20.6438 27.62832
eurozone | -.0949131 .0304085 -3.12 0.002 -.1547713 -.0350549
eu | -.2019194 .0575086 -3.51 0.001 -.3151235 -.0887153
lngovtexp | -.6823006 .1670674 -4.08 0.000 -1.011168 -.353433
lnwage | .0640134 .0635835 1.01 0.315 -.0611489 .1891758
plurality | -.1916017 .0437287 -4.38 0.000 -.2776804 -.105523
proportional | -.3352688 .0839047 -4.00 0.000 -.5004329 -.1701047
lnpit | .2566626 .0407955 6.29 0.000 .1763578 .3369675
lncorporate | .2705219 .0618869 4.37 0.000 .1486993 .3923445
_cons | 4.06789 10.44516 0.39 0.697 -16.49312 24.6289
-----------------------------------------------------------------------------------
Wooldridge Serial Correlation test gives:
Code:
Wooldridge test for autocorrelation in panel data
H0: no first-order autocorrelation
F( 1, 18) = 12.043
Prob > F = 0.0027
Code:
Linear regression Number of obs = 299
F( 18, 280) = 39.24
Prob > F = 0.0000
R-squared = 0.5514
Root MSE = .20281
-----------------------------------------------------------------------------------
| Robust
lnPATR1 | Coef. Std. Err. t P>|t| [95% Conf. Interval]
------------------+----------------------------------------------------------------
preelection | -.034093 .0403675 -0.84 0.399 -.1135553 .0453693
lninflation | -.0082857 .0280417 -0.30 0.768 -.063485 .0469135
lngdp | -.0214539 .0127689 -1.68 0.094 -.046589 .0036813
lnurbanisation | .5635988 .1326198 4.25 0.000 .3025404 .8246572
left | .1474364 .0290717 5.07 0.000 .0902096 .2046633
right | .1450968 .0329821 4.40 0.000 .0801725 .2100211
lndependency | 2.259958 3.854314 0.59 0.558 -5.327153 9.847069
lndependencyold | -4.3537 10.82639 -0.40 0.688 -25.66514 16.95774
lndependencyyoung | -2.010141 .8918769 -2.25 0.025 -3.765776 -.2545055
lnpopbelow14 | 0 (omitted)
lnpop1564 | 0 (omitted)
lnpop65andabove | 3.49226 10.73778 0.33 0.745 -17.64477 24.62929
eurozone | -.0949131 .0345734 -2.75 0.006 -.1629698 -.0268564
eu | -.2019194 .0646812 -3.12 0.002 -.3292426 -.0745963
lngovtexp | -.6823006 .1725209 -3.95 0.000 -1.021903 -.3426978
lnwage | .0640134 .0667391 0.96 0.338 -.0673606 .1953875
plurality | -.1916017 .0433451 -4.42 0.000 -.2769254 -.1062781
proportional | -.3352688 .0718241 -4.67 0.000 -.4766526 -.193885
lnpit | .2566626 .0354634 7.24 0.000 .186854 .3264712
lncorporate | .2705219 .0661607 4.09 0.000 .1402863 .4007575
_cons | 4.06789 9.097983 0.45 0.655 -13.84124 21.97702
-----------------------------------------------------------------------------------
Comment