Hi,
I am running OLS regressions, firstly for 3 separate years (2006, 2010 and then 2015) and then for the whole data period (1995-2010). When I use -estat ovtest- the individual years (and even when I try five-year periods) fail to reject the null of no omitted variables,
however when I test the model that includes the whole 21 year period, I get a very low P-value, hence rejecting the null and implying my model has omitted variables.
What can be done to remedy this, particularly as I'm exactly replicating a paper that only uses 2006 values and then comparing the results from other years and the 21 year period?
Thanks everyone,
Hellie
I am running OLS regressions, firstly for 3 separate years (2006, 2010 and then 2015) and then for the whole data period (1995-2010). When I use -estat ovtest- the individual years (and even when I try five-year periods) fail to reject the null of no omitted variables,
Code:
reg AvgGrowth lagGII lagIncomeln lagCivilLiberty lagEthnic lagLinguistic lagReligious lagMuslim lagCatholic
> lagProtestant lagInflation lagLandlocked lagPopulationln lagTrade lagFinancialFreedom lagFiscalFreedom lagEd
> ucation lagEducGap GII_Income EducGap_Income if Year==2006, robust
Linear regression Number of obs = 81
F(19, 61) = 7.64
Prob > F = 0.0000
R-squared = 0.5197
Root MSE = 1.7863
-------------------------------------------------------------------------------------
| Robust
AvgGrowth | Coef. Std. Err. t P>|t| [95% Conf. Interval]
--------------------+----------------------------------------------------------------
lagGII | -57.32564 21.19778 -2.70 0.009 -99.71321 -14.93806
lagIncomeln | -1.039012 2.384421 -0.44 0.665 -5.806956 3.728932
lagCivilLiberty | -.0821918 .2667543 -0.31 0.759 -.6155999 .4512163
lagEthnic | .6690179 1.238369 0.54 0.591 -1.807253 3.145289
lagLinguistic | -.1083739 1.031271 -0.11 0.917 -2.170528 1.95378
lagReligious | -1.097978 1.089466 -1.01 0.318 -3.2765 1.080544
lagMuslim | -.0096811 .0105531 -0.92 0.363 -.0307833 .0114211
lagCatholic | -.0118937 .0093353 -1.27 0.207 -.0305607 .0067733
lagProtestant | -.0090421 .0111835 -0.81 0.422 -.0314049 .0133206
lagInflation | .0564553 .0748507 0.75 0.454 -.093218 .2061286
lagLandlocked | .618708 .5490238 1.13 0.264 -.479133 1.716549
lagPopulationln | .0032796 .2040642 0.02 0.987 -.4047719 .4113312
lagTrade | .0058239 .0059616 0.98 0.332 -.006097 .0177448
lagFinancialFreedom | .0001623 .0143011 0.01 0.991 -.0284345 .028759
lagFiscalFreedom | .03642 .021103 1.73 0.089 -.0057781 .078618
lagEducation | -.0126441 .0204858 -0.62 0.539 -.053608 .0283198
lagEducGap | 25.19475 15.74717 1.60 0.115 -6.293671 56.68316
GII_Income | 4.998349 2.108217 2.37 0.021 .7827081 9.213991
EducGap_Income | -2.940885 1.832571 -1.60 0.114 -6.605337 .7235669
_cons | 19.24102 22.57219 0.85 0.397 -25.89486 64.3769
-------------------------------------------------------------------------------------
. estat ovtest
Ramsey RESET test using powers of the fitted values of AvgGrowth
Ho: model has no omitted variables
F(3, 58) = 0.28
Prob > F = 0.8394
however when I test the model that includes the whole 21 year period, I get a very low P-value, hence rejecting the null and implying my model has omitted variables.
Code:
. reg AvgGrowth lagGII lagIncomeln lagCivilLiberty lagEthnic lagLinguistic lagReligious lagMuslim lagCatholic
> lagProtestant lagInflation lagLandlocked lagPopulationln lagTrade lagFinancialFreedom lagFiscalFreedom lagEd
> ucation lagEducGap GII_Income EducGap_Income i.Year, robust
Linear regression Number of obs = 1,544
F(38, 1505) = 20.84
Prob > F = 0.0000
R-squared = 0.3044
Root MSE = 2.0518
-------------------------------------------------------------------------------------
| Robust
AvgGrowth | Coef. Std. Err. t P>|t| [95% Conf. Interval]
--------------------+----------------------------------------------------------------
lagGII | -42.08487 4.341376 -9.69 0.000 -50.60066 -33.56908
lagIncomeln | -2.538325 .6338862 -4.00 0.000 -3.781719 -1.294931
lagCivilLiberty | -.1399923 .0649339 -2.16 0.031 -.2673629 -.0126218
lagEthnic | -.2943895 .3422877 -0.86 0.390 -.965801 .377022
lagLinguistic | -.3334241 .3096355 -1.08 0.282 -.9407871 .2739388
lagReligious | -.7723125 .2628651 -2.94 0.003 -1.287933 -.2566917
lagMuslim | -.0032381 .0026514 -1.22 0.222 -.0084389 .0019628
lagCatholic | -.008598 .0022728 -3.78 0.000 -.0130563 -.0041398
lagProtestant | -.0008512 .0028165 -0.30 0.763 -.0063759 .0046734
lagInflation | -.0051196 .0022192 -2.31 0.021 -.0094727 -.0007665
lagLandlocked | .5637036 .1505337 3.74 0.000 .2684255 .8589816
lagPopulationln | .3095838 .0470946 6.57 0.000 .2172058 .4019617
lagTrade | .0114543 .0017344 6.60 0.000 .0080521 .0148564
lagFinancialFreedom | -.0133614 .0040853 -3.27 0.001 -.0213749 -.005348
lagFiscalFreedom | .004272 .0051197 0.83 0.404 -.0057706 .0143145
lagEducation | .0114035 .0042116 2.71 0.007 .0031422 .0196648
lagEducGap | 4.518104 4.82945 0.94 0.350 -4.955062 13.99127
GII_Income | 3.98745 .4390719 9.08 0.000 3.126192 4.848708
EducGap_Income | -.5881582 .5614429 -1.05 0.295 -1.689452 .5131353
|
Year |
1997 | -.1178138 .3623058 -0.33 0.745 -.8284917 .5928641
1998 | .0913023 .3949375 0.23 0.817 -.683384 .8659885
1999 | -.4652848 .3490904 -1.33 0.183 -1.15004 .2194704
2000 | -.7425016 .3171268 -2.34 0.019 -1.364559 -.1204442
2001 | -.6776028 .3271859 -2.07 0.039 -1.319391 -.0358142
2002 | -.19607 .2945478 -0.67 0.506 -.7738377 .3816977
2003 | .1985168 .2822417 0.70 0.482 -.355112 .7521456
2004 | .7939147 .3101818 2.56 0.011 .1854803 1.402349
2005 | .8820597 .3151816 2.80 0.005 .2638179 1.500302
2006 | .5161398 .3186726 1.62 0.106 -.1089496 1.141229
2007 | -1.552917 .3347261 -4.64 0.000 -2.209496 -.8963382
2008 | -1.878248 .3355065 -5.60 0.000 -2.536358 -1.220138
2009 | -1.771681 .3400277 -5.21 0.000 -2.438659 -1.104702
2010 | -.4046335 .3357402 -1.21 0.228 -1.063202 .2539349
2011 | -1.064553 .3721843 -2.86 0.004 -1.794608 -.3344979
2012 | -1.235579 .3735256 -3.31 0.001 -1.968265 -.5028927
2013 | -1.227063 .3800548 -3.23 0.001 -1.972556 -.4815694
2014 | -1.205918 .3389585 -3.56 0.000 -1.870799 -.5410364
2015 | -.7814795 .3455992 -2.26 0.024 -1.459387 -.1035722
|
_cons | 24.86397 5.923684 4.20 0.000 13.24442 36.48352
-------------------------------------------------------------------------------------
. estat ovtest
Ramsey RESET test using powers of the fitted values of AvgGrowth
Ho: model has no omitted variables
F(3, 1502) = 5.61
Prob > F = 0.0008
What can be done to remedy this, particularly as I'm exactly replicating a paper that only uses 2006 values and then comparing the results from other years and the 21 year period?
Thanks everyone,
Hellie
Comment