Durbin-Watson Stat in GMM and Bayesian Fixed Effects
Posted: Wed Oct 16, 2013 4:40 am
Hi Tom!!!
I have 2 unrelated questions about panel data:
1. I have always used STATA for GMM and just recently switched to RATS for panel exercies. I always used the Arellano-Bond autocorrekation test "estat abond" and the sargan tests in Stata. Was just wondering, whether
the Durbin Watson statistic is equivalent to the Arellano-Bond test for zero autocorrelation? Below is a sample output. I have a DW statistic of 2.2772 and I have N=16 (Also i found it weird that the J statistic is always at 15 and the significance is always 1, no matter how I play around with sometimes illogical instruments .
2. I just bought your book, Bayesian Econometrics Course in Estima, and came across the Bayesian Fixed effects model. Does it make sense to try Bayesian methods to my macro panel data (at the baseline, I used FE with cluistered standard errors and for the second step, I did GMM as shown below) with T=64 and N=16? What might be the justification to use Bayesian FE instead of the usual classical FE?
Thanks so much for all your help!!! Its great being in this forum!
dnfloro1
Linear Regression - Estimation by Arellano-Bond One-Step Estimator
With Clustered Standard Error Calculations
Dependent Variable POLICY
Panel(64) of Quarterly Data From 1//1995:01 To 17//2009:04
Usable Observations 848
Degrees of Freedom 842
Skipped/Missing (from 1084) 236
Mean of Dependent Variable 6.4304127358
Std Error of Dependent Variable 5.5854547733
Standard Error of Estimate 1.6662542074
Sum of Squared Residuals 2337.7313966
J-Specification(75) 15.0000
Significance Level of J 1.0000000
Durbin-Watson Statistic 2.2772
Variable Coeff Std Error T-Stat Signif
************************************************************************************
1. POLICY{1} 0.927282343 0.034286142 27.04540 0.00000000
2. GROWTH{1} -0.011244108 0.024521615 -0.45854 0.64656550
3. INAXFSI11{1} -0.028745934 0.005942821 -4.83709 0.00000132
4. INAXINDEP1 0.117711013 0.093417820 1.26005 0.20765175
5. INAXFISCAL3 -2.990367233 1.057039942 -2.82900 0.00466935
6. INAXOPEN1 0.268030093 0.074459028 3.59970 0.00031859
I have 2 unrelated questions about panel data:
1. I have always used STATA for GMM and just recently switched to RATS for panel exercies. I always used the Arellano-Bond autocorrekation test "estat abond" and the sargan tests in Stata. Was just wondering, whether
the Durbin Watson statistic is equivalent to the Arellano-Bond test for zero autocorrelation? Below is a sample output. I have a DW statistic of 2.2772 and I have N=16 (Also i found it weird that the J statistic is always at 15 and the significance is always 1, no matter how I play around with sometimes illogical instruments .
2. I just bought your book, Bayesian Econometrics Course in Estima, and came across the Bayesian Fixed effects model. Does it make sense to try Bayesian methods to my macro panel data (at the baseline, I used FE with cluistered standard errors and for the second step, I did GMM as shown below) with T=64 and N=16? What might be the justification to use Bayesian FE instead of the usual classical FE?
Thanks so much for all your help!!! Its great being in this forum!
dnfloro1
Linear Regression - Estimation by Arellano-Bond One-Step Estimator
With Clustered Standard Error Calculations
Dependent Variable POLICY
Panel(64) of Quarterly Data From 1//1995:01 To 17//2009:04
Usable Observations 848
Degrees of Freedom 842
Skipped/Missing (from 1084) 236
Mean of Dependent Variable 6.4304127358
Std Error of Dependent Variable 5.5854547733
Standard Error of Estimate 1.6662542074
Sum of Squared Residuals 2337.7313966
J-Specification(75) 15.0000
Significance Level of J 1.0000000
Durbin-Watson Statistic 2.2772
Variable Coeff Std Error T-Stat Signif
************************************************************************************
1. POLICY{1} 0.927282343 0.034286142 27.04540 0.00000000
2. GROWTH{1} -0.011244108 0.024521615 -0.45854 0.64656550
3. INAXFSI11{1} -0.028745934 0.005942821 -4.83709 0.00000132
4. INAXINDEP1 0.117711013 0.093417820 1.26005 0.20765175
5. INAXFISCAL3 -2.990367233 1.057039942 -2.82900 0.00466935
6. INAXOPEN1 0.268030093 0.074459028 3.59970 0.00031859