Dear list members,
I have an unbalanced long T dataset (38 quarters), and running a fixed-effects model. I also include the lagged dependent variable variable due to the theory I am testing. Again, due to the theory I am testing I include lags from 1 to 4 of all explanatory variables. As the dynamic panel bias increases with a long T panel, I do not prefer to estimate a GMM model. When I obtain my results, I see that the signs of the coefficients of some of the explanatory variables changes in the further lags (e.g. lag 1 is significantly positive, but lag 2 is significantly negative). Therefore, it becomes confusing to comment on the relation of the explanatory variables with the dependent variable. I therefore wanted to know whether it is possible to estimate long-run coefficients for a fixed-effects model. I may be asking for something econometrically wrong, but want to hear suggestions if any?
I have an unbalanced long T dataset (38 quarters), and running a fixed-effects model. I also include the lagged dependent variable variable due to the theory I am testing. Again, due to the theory I am testing I include lags from 1 to 4 of all explanatory variables. As the dynamic panel bias increases with a long T panel, I do not prefer to estimate a GMM model. When I obtain my results, I see that the signs of the coefficients of some of the explanatory variables changes in the further lags (e.g. lag 1 is significantly positive, but lag 2 is significantly negative). Therefore, it becomes confusing to comment on the relation of the explanatory variables with the dependent variable. I therefore wanted to know whether it is possible to estimate long-run coefficients for a fixed-effects model. I may be asking for something econometrically wrong, but want to hear suggestions if any?
Code:
xtreg ROA L1.ROA L(1/4).(RGDP INFL UNEM EA NPL_Gross CosttoIncome AbsoluteSize Current_Ratio NII_TA LSV), fe
Code:
Fixed-effects (within) regression Number of obs = 928
Group variable: bank Number of groups = 29
R-sq: within = 0.7081 Obs per group: min = 10
between = 0.8980 avg = 32.0
overall = 0.7636 max = 33
F(41,858) = 50.76
corr(u_i, Xb) = 0.3092 Prob > F = 0.0000
-------------------------------------------------------------------------------
ROA | Coef. Std. Err. t P>|t| [95% Conf. Interval]
--------------+----------------------------------------------------------------
ROA |
L1. | .7197828 .0201665 35.69 0.000 .6802014 .7593642
|
RGDP |
L1. | .0294273 .0216249 1.36 0.174 -.0130166 .0718712
L2. | -.0399119 .0188087 -2.12 0.034 -.0768283 -.0029954
L3. | -.0285918 .0243302 -1.18 0.240 -.0763455 .0191619
L4. | -.0026335 .0203985 -0.13 0.897 -.0426703 .0374034
|
INFL |
L1. | .0737775 .0305025 2.42 0.016 .0139092 .1336458
L2. | -.0086795 .0315864 -0.27 0.784 -.0706752 .0533162
L3. | -.0435002 .0356204 -1.22 0.222 -.1134135 .0264131
L4. | .0152626 .0222298 0.69 0.493 -.0283686 .0588937
|
UNEM |
L1. | -.0387789 .0534237 -0.73 0.468 -.1436353 .0660775
L2. | .0675899 .0431009 1.57 0.117 -.0170057 .1521854
L3. | -.0178881 .0403549 -0.44 0.658 -.097094 .0613177
L4. | -.0364572 .0404826 -0.90 0.368 -.1159138 .0429994
|
EA |
L1. | .0257591 .0125593 2.05 0.041 .0011084 .0504097
L2. | -.02367 .0149172 -1.59 0.113 -.0529485 .0056086
L3. | .0326956 .0154422 2.12 0.035 .0023868 .0630045
L4. | -.0297878 .0112333 -2.65 0.008 -.0518357 -.0077399
|
NPL_Gross |
L1. | -.0235523 .0094105 -2.50 0.013 -.0420226 -.0050819
L2. | -.0090256 .0112839 -0.80 0.424 -.0311728 .0131216
L3. | .0112926 .0107352 1.05 0.293 -.0097777 .0323629
L4. | .0132082 .0081891 1.61 0.107 -.0028648 .0292812
|
CosttoIncome |
L1. | -.0000295 .0000262 -1.12 0.261 -.000081 .000022
L2. | -.0000349 .0000262 -1.33 0.182 -.0000863 .0000164
L3. | -.0000639 .0000261 -2.45 0.014 -.0001151 -.0000127
L4. | -.0000105 .0000261 -0.40 0.687 -.0000617 .0000407
|
AbsoluteSize |
L1. | .0012674 .0026976 0.47 0.639 -.0040273 .0065621
L2. | -.0007526 .0033651 -0.22 0.823 -.0073574 .0058521
L3. | .0052451 .0034642 1.51 0.130 -.0015543 .0120444
L4. | -.0072977 .0027751 -2.63 0.009 -.0127445 -.001851
|
Current_Ratio |
L1. | .0010355 .0013443 0.77 0.441 -.001603 .003674
L2. | .0011543 .0015619 0.74 0.460 -.0019113 .0042199
L3. | -.0035439 .0014725 -2.41 0.016 -.0064339 -.0006539
L4. | .0003214 .0003754 0.86 0.392 -.0004154 .0010583
|
NII_TA |
L1. | .0072502 .0166707 0.43 0.664 -.02547 .0399704
L2. | -.0198808 .0142598 -1.39 0.164 -.0478691 .0081074
L3. | -.0366395 .0140463 -2.61 0.009 -.0642085 -.0090704
L4. | .0037115 .0123751 0.30 0.764 -.0205776 .0280005
|
LSV |
L1. | -.0308357 .0146235 -2.11 0.035 -.0595377 -.0021337
L2. | .0056811 .0120967 0.47 0.639 -.0180615 .0294237
L3. | -.0206869 .013424 -1.54 0.124 -.0470346 .0056608
L4. | -.0003904 .0153515 -0.03 0.980 -.0305212 .0297405
|
_cons | .0311564 .0149171 2.09 0.037 .0018781 .0604347
--------------+----------------------------------------------------------------
sigma_u | .00488001
sigma_e | .00940663
rho | .21206284 (fraction of variance due to u_i)
-------------------------------------------------------------------------------
F test that all u_i=0: F(28, 858) = 2.29 Prob > F = 0.0002
