[Limdep Nlogit List] Some help to confirm which output to use for a two way fixed / random effects panel data test
Darren Lee
d.lee at business.uq.edu.au
Thu Nov 25 17:28:08 EST 2010
Sorry the attached limdep output file did not come through. I have now added to the body of the email.
Hi Everyone,
I wanted to confirm that I was on the right track. I have a two way fixed and random effects panel data model as per section E11.10 of the manual. There are firm and time effects. The model specifications and output are below. I believe that the output that should be reported given the below results is for the fixed effects (least squares) model with Group and Period Effects. Am I correct in forming this conclusion? If not - could you please let me know why?
Thanks in advance.
cheers,
Darren
--> SKIP
--> REGRESS;Lhs=EXMRET;Rhs=ONE,TOTALCOM,LNMKTCAP,LNBTM_T_,PR1YRAVR
,AV_SHTUR,LNFIRMAG,FTSE,BETA;Het;Panel;Str=_STRATUM;Period=DATE$
--------------------------------------------------------------
Deleted 555 observations with missing data. N is now 9173
--------------------------------------------------------------
------------------------------------------------------------------
OLS Without Group Dummy Variables.................
Ordinary least squares regression ............
LHS=EXMRET Mean = .00894
Standard deviation = .08897
Number of observs. = 9173
Model size Parameters = 9
Degrees of freedom = 9164
Residuals Sum of squares = 71.50156
Standard error of e = .08833
Fit R-squared = .01509
Adjusted R-squared = .01423
Model test F[ 8, 9164] (prob) = 17.5(.0000)
Diagnostic Log likelihood = 9248.32521
Restricted(b=0) = 9178.59048
Chi-sq [ 8] (prob) = 139.5(.0000)
Info criter. LogAmemiya Prd. Crt. = -4.85234
Akaike Info. Criter. = -4.85234
Bayes Info. Criter. = -4.84535
Model was estimated on Nov 19, 2010 at 07:04:14 PM
Panel Data Analysis of EXMRET [ONE way]
Unconditional ANOVA (No regressors)
Source Variation Deg. Free. Mean Square
Between 3.21559 205. .01569
Residual 69.38142 8967. .00774
Total 72.59700 9172. .00792
--------+---------------------------------------------------------
| Standard Prob. Mean
EXMRET| Coefficient Error z z>|Z| of X
--------+---------------------------------------------------------
TOTALCOM| .00022*** .5697D-04 3.85 .0001 44.8753
LNMKTCAP| -.00613*** .00101 -6.07 .0000 8.06782
LNBTM_T_| .00445*** .00119 3.75 .0002 -.88314
PR1YRAVR| .01448 .03566 .41 .6846 .00834
AV_SHTUR|-.50313D-08 .3410D-07 -.15 .8827 12789.9
LNFIRMAG| -.00374*** .00130 -2.87 .0041 8.96880
FTSE| -.01160*** .00242 -4.79 .0000 .81849
BETA| .00522*** .00111 4.69 .0000 1.10322
Constant| .08972*** .01289 6.96 .0000
--------+---------------------------------------------------------
Note: nnnnn.D-xx or D+xx => multiply by 10 to -xx or +xx.
Note: ***, **, * ==> Significance at 1%, 5%, 10% level.
------------------------------------------------------------------
?
------------------------------------------------------------------
Least Squares with Group Dummy Variables..........
Ordinary least squares regression ............
LHS=EXMRET Mean = .00894
Standard deviation = .08897
Number of observs. = 9173
Model size Parameters = 214
Degrees of freedom = 8959
Residuals Sum of squares = 67.00956
Standard error of e = .08648
Fit R-squared = .07697
Adjusted R-squared = .05502
Model test F[213, 8959] (prob) = 3.5(.0000)
Diagnostic Log likelihood = 9545.91564
Restricted(b=0) = 9178.59048
Chi-sq [213] (prob) = 734.7(.0000)
Info criter. LogAmemiya Prd. Crt. = -4.87252
Akaike Info. Criter. = -4.87253
Bayes Info. Criter. = -4.70633
Model was estimated on Nov 19, 2010 at 07:04:14 PM
Estd. Autocorrelation of e(i,t) = .033680
White/Hetero. corrected covariance matrix was used
Panel:Groups Empty 0, Valid data 206
Smallest 2, Largest 70
Average group size in panel 44.53
--------+---------------------------------------------------------
| Standard Prob. Mean
EXMRET| Coefficient Error z z>|Z| of X
--------+---------------------------------------------------------
TOTALCOM| .00049*** .00012 4.11 .0000 44.8753
LNMKTCAP| -.04231*** .00518 -8.18 .0000 8.06782
LNBTM_T_| .01810*** .00351 5.15 .0000 -.88314
PR1YRAVR| .25435*** .07431 3.42 .0006 .00834
AV_SHTUR|-.27747D-06** .1159D-06 -2.39 .0167 12789.9
LNFIRMAG| -.01650 .01163 -1.42 .1562 8.96880
FTSE| -.00738* .00407 -1.81 .0698 .81849
BETA| .00442*** .00171 2.59 .0096 1.10322
--------+---------------------------------------------------------
Note: nnnnn.D-xx or D+xx => multiply by 10 to -xx or +xx.
Note: ***, **, * ==> Significance at 1%, 5%, 10% level.
------------------------------------------------------------------
+--------------------------------------------------------------------+
| Test Statistics for the Classical Model |
+--------------------------------------------------------------------+
| Model Log-Likelihood Sum of Squares R-squared |
|(1) Constant term only 9178.59061 72.59700 .00000 |
|(2) Group effects only 9386.38014 69.38142 .04429 |
|(3) X - variables only 9248.32535 71.50156 .01509 |
|(4) X and group effects 9545.91578 67.00956 .07697 |
+--------------------------------------------------------------------+
| Hypothesis Tests |
| Likelihood Ratio Test F Tests |
| Chi-squared d.f. Prob F num denom P value |
|(2) vs (1) 415.58 205 .0000 2.03 205 8967 .00000 |
|(3) vs (1) 139.47 8 .0000 17.55 8 9164 .00000 |
|(4) vs (1) 734.65 213 .0000 3.51 213 8959 .00000 |
|(4) vs (2) 319.07 8 .0000 39.64 8 8959 .00000 |
|(4) vs (3) 595.18 205 .0000 2.93 205 8959 .00000 |
+--------------------------------------------------------------------+
------------------------------------------------------------------
Random Effects Model: v(i,t) = e(i,t) + u(i)
Estimates: Var[e] = .007480
Var[u] = .000323
Corr[v(i,t),v(i,s)] = .041379
Lagrange Multiplier Test vs. Model (3) = 10.74
( 1 degrees of freedom, prob. value = .001049)
(High values of LM favor FEM/REM over CR model)
Baltagi-Li form of LM Statistic = 3.13
Moulton/Randolph form:SLM N[0,1]= 3.78
Fixed vs. Random Effects (Hausman) = .00
( 8 degrees of freedom, prob. value = 1.000000)
(High (low) values of H favor F.E.(R.E.) model)
Sum of Squares 72.073697
R-squared .007228
--------+---------------------------------------------------------
| Standard Prob. Mean
EXMRET| Coefficient Error z z>|Z| of X
--------+---------------------------------------------------------
TOTALCOM| .00030*** .7962D-04 3.73 .0002 44.8753
LNMKTCAP| -.01139*** .00157 -7.24 .0000 8.06782
LNBTM_T_| .00936*** .00171 5.47 .0000 -.88314
PR1YRAVR| .03546 .03852 .92 .3572 .00834
AV_SHTUR|-.90188D-08 .5491D-07 -.16 .8695 12789.9
LNFIRMAG| -.00522** .00212 -2.46 .0138 8.96880
FTSE| -.01373*** .00328 -4.19 .0000 .81849
BETA| .00496*** .00121 4.09 .0000 1.10322
Constant| .14778*** .02006 7.37 .0000
--------+---------------------------------------------------------
Note: nnnnn.D-xx or D+xx => multiply by 10 to -xx or +xx.
Note: ***, **, * ==> Significance at 1%, 5%, 10% level.
------------------------------------------------------------------
?
------------------------------------------------------------------
Least Squares with Group and Period Effects.......
Ordinary least squares regression ............
LHS=EXMRET Mean = .00894
Standard deviation = .08897
Number of observs. = 9173
Model size Parameters = 284
Degrees of freedom = 8889
Residuals Sum of squares = 51.99512
Standard error of e = .07648
Fit R-squared = .28370
Adjusted R-squared = .26090
Model test F[283, 8889] (prob) = 12.4(.0000)
Diagnostic Log likelihood = 10709.44372
Restricted(b=0) = 9178.59048
Chi-sq [283] (prob) = 3061.7(.0000)
Info criter. LogAmemiya Prd. Crt. = -5.11093
Akaike Info. Criter. = -5.11095
Bayes Info. Criter. = -4.89039
Model was estimated on Nov 19, 2010 at 07:04:15 PM
Estd. Autocorrelation of e(i,t) = -.009535
Panel:Groups Empty 0, Valid data 206
Smallest 2, Largest 70
Average group size in panel 44.53
Panel: Prds: Empty 0, Valid data 70
Smallest 0, Largest 169
Average group size in panel 131.04
--------+---------------------------------------------------------
| Standard Prob. Mean
EXMRET| Coefficient Error z z>|Z| of X
--------+---------------------------------------------------------
TOTALCOM| .58983D-04 .00012 .51 .6118 44.8753
LNMKTCAP| -.03045*** .00430 -7.09 .0000 8.06782
LNBTM_T_| .02281*** .00280 8.15 .0000 -.88314
PR1YRAVR| .07127 .05225 1.36 .1726 .00834
AV_SHTUR|-.22719D-06*** .8059D-07 -2.82 .0048 12789.9
LNFIRMAG| -.01927** .00961 -2.00 .0451 8.96880
FTSE| .00334 .00416 .80 .4219 .81849
BETA| .00268** .00119 2.26 .0238 1.10322
Constant| .44154*** .08954 4.93 .0000
--------+---------------------------------------------------------
Note: nnnnn.D-xx or D+xx => multiply by 10 to -xx or +xx.
Note: ***, **, * ==> Significance at 1%, 5%, 10% level.
------------------------------------------------------------------
+--------------------------------------------------------------------+
| Test Statistics for the Classical Model |
+--------------------------------------------------------------------+
| Model Log-Likelihood Sum of Squares R-squared |
|(1) Constant term only 9178.59061 72.59700 .00000 |
|(2) Group effects only 9386.38014 69.38142 .04429 |
|(3) X - variables only 9248.32535 71.50156 .01509 |
|(4) X and group effects 9545.91578 67.00956 .07697 |
|(5) X ind.&time effects 10708.92791 52.00097 .28370 |
+--------------------------------------------------------------------+
| Hypothesis Tests |
| Likelihood Ratio Test F Tests |
| Chi-squared d.f. Prob F num denom P value |
|(2) vs (1) 415.58 205 .0000 2.03 205 8967 .00000 |
|(3) vs (1) 139.47 8 .0000 17.55 8 9164 .00000 |
|(4) vs (1) 734.65 213 .0000 3.51 213 8959 .00000 |
|(4) vs (2) 319.07 8 .0000 39.64 8 8959 .00000 |
|(4) vs (3) 595.18 205 .0000 2.93 205 8959 .00000 |
|(5) vs (4) 2326.02 69 .0000 37.19 69 8890 .00000 |
|(5) vs (3) 2921.21 275 .0000 12.12 275 8890 .00000 |
+--------------------------------------------------------------------+
------------------------------------------------------------------
Random Effects Model: v(i,t) = e(i,t)+u(i)+w(t)
Estimates: Var[e] = .005849
Var[u] = .002790
Corr[v(i,t),v(i,s)] = .322939
Var[w] = .001606
Corr[v(i,t),v(j,t)] = .215455
Lagrange Multiplier Test vs. Model (3) =*******
( 2 degrees of freedom, prob. value = .000000)
(High values of LM favor FEM/REM over CR model)
Fixed vs. Random Effects (Hausman) = 62.19
( 8 degrees of freedom, prob. value = .000000)
(High (low) values of H favor F.E.(R.E.) model)
Sum of Squares 72.073697
R-squared .007228
--------+---------------------------------------------------------
| Standard Prob. Mean
EXMRET| Coefficient Error z z>|Z| of X
--------+---------------------------------------------------------
TOTALCOM| .00023** .00010 2.33 .0199 44.8753
LNMKTCAP| -.01384*** .00261 -5.30 .0000 8.06782
LNBTM_T_| .01920*** .00240 8.01 .0000 -.88314
PR1YRAVR| -.03707 .04634 -.80 .4238 .00834
AV_SHTUR|-.11910D-06 .7270D-07 -1.64 .1014 12789.9
LNFIRMAG| -.00857** .00430 -1.99 .0461 8.96880
FTSE| .00220 .00387 .57 .5693 .81849
BETA| .00267** .00116 2.30 .0217 1.10322
Constant| .19762*** .03991 4.95 .0000
--------+---------------------------------------------------------
Note: nnnnn.D-xx or D+xx => multiply by 10 to -xx or +xx.
Note: ***, **, * ==> Significance at 1%, 5%, 10% level.
------------------------------------------------------------------
Best Regards,
Darren Lee, PhD
Lecturer in Finance
UQ Business School
The University of Queensland
Room 328,Colin Clark Blg
St Lucia, Blair Drive
Queensland, Australia 4072
PH: 61 7 33468048
Fax: 61 7 33468166
Email: d.lee2 at uq.edu.au
Website: www.business.uq.edu.au/display/teach/Darren+Lee <https://exchange.uq.edu.au/exchweb/bin/redir.asp?URL=http://www.business.uq.edu.au/display/teach/Darren%2BLee>
PLEASE CONSIDER OUR ENVIRONMENT BEFORE PRINTING THIS EMAIL. PRIVATE AND CONFIDENTIAL: This email and any files transmitted with it are intended solely for the use of the addressee(s) and may contain information that is confidential, copyright or privileged and should not be distributed without the authors express consent. If you receive this email and you are not the addressee(s), please disregard the contents of the email, delete the email and notify the author immediately.
More information about the Limdep
mailing list