I am running a panel data with fixed effects. The model I'm applying is a partial adjustment model to see if firms have a target debt structure. In the model I include a dummy variable that is equal to 1 if the firm's intangible asset (variable name IVA_TA) has a missing value (so the dummy variable is 1 when there is a dot (.) in the data), 0 otherwise.
However, when I try to run the xtreg command with the dummy variable together with the variable IVA_TA, the dummy variable gets omitted because of collinearity. If I leave out either one of the two variables, the command runs perfectly fine.
This is the output with both variables IVA_TA and IVA_DUMMY included
Code:
xtreg MDR_win l1.MDR_win l1.EBIT_TA_win l1.MB_win l1.DEP_TA_win l1.LnTA_win l1 > .FA_TA_win IVA_TA_win IVA_DUMMY_win , fe note: IVA_DUMMY_win omitted because of collinearity. Fixed-effects (within) regression Number of obs = 182 Group variable: firmid Number of groups = 47 R-squared: Obs per group: Within = 0.1705 min = 1 Between = 0.7017 avg = 3.9 Overall = 0.5644 max = 4 F(7,128) = 3.76 corr(u_i, Xb) = 0.5712 Prob > F = 0.0010 ------------------------------------------------------------------------------- MDR_win | Coefficient Std. err. t P>|t| [95% conf. interval] --------------+---------------------------------------------------------------- MDR_win | L1. | .3768109 .0868534 4.34 0.000 .2049565 .5486652 | EBIT_TA_win | L1. | -.1659445 .2490226 -0.67 0.506 -.6586783 .3267893 | MB_win | L1. | .0161656 .0058284 2.77 0.006 .0046331 .0276981 | DEP_TA_win | L1. | -.1953914 .4344386 -0.45 0.654 -1.055003 .6642197 | LnTA_win | L1. | -.00083 .0392252 -0.02 0.983 -.0784437 .0767837 | FA_TA_win | L1. | .0108168 .0872136 0.12 0.901 -.1617502 .1833838 | IVA_TA_win | -.2374304 .2070157 -1.15 0.254 -.6470462 .1721855 IVA_DUMMY_win | 0 (omitted) _cons | .0707918 .2471637 0.29 0.775 -.4182638 .5598474 --------------+---------------------------------------------------------------- sigma_u | .11470965 sigma_e | .08096757 rho | .66745816 (fraction of variance due to u_i) ------------------------------------------------------------------------------- F test that all u_i=0: F(46, 128) = 2.16 Prob > F = 0.0004
Code:
xtreg MDR_win l1.MDR_win l1.EBIT_TA_win l1.MB_win l1.DEP_TA_win l1.LnTA_win l1 > .FA_TA_win IVA_DUMMY_win , fe Fixed-effects (within) regression Number of obs = 208 Group variable: firmid Number of groups = 52 R-squared: Obs per group: Within = 0.1462 min = 4 Between = 0.3561 avg = 4.0 Overall = 0.3196 max = 4 F(7,149) = 3.65 corr(u_i, Xb) = 0.2243 Prob > F = 0.0012 ------------------------------------------------------------------------------- MDR_win | Coefficient Std. err. t P>|t| [95% conf. interval] --------------+---------------------------------------------------------------- MDR_win | L1. | .3422164 .0827372 4.14 0.000 .1787267 .5057061 | EBIT_TA_win | L1. | -.1749656 .2131508 -0.82 0.413 -.5961543 .2462232 | MB_win | L1. | .0138445 .0057302 2.42 0.017 .0025216 .0251674 | DEP_TA_win | L1. | -.2009779 .416863 -0.48 0.630 -1.024705 .6227489 | LnTA_win | L1. | -.0222294 .0380439 -0.58 0.560 -.0974046 .0529458 | FA_TA_win | L1. | .0164623 .0835363 0.20 0.844 -.1486065 .1815311 | IVA_DUMMY_win | .0591311 .051265 1.15 0.251 -.0421691 .1604313 _cons | .197463 .2385291 0.83 0.409 -.2738736 .6687996 --------------+---------------------------------------------------------------- sigma_u | .14593521 sigma_e | .08039293 rho | .76718305 (fraction of variance due to u_i) ------------------------------------------------------------------------------- F test that all u_i=0: F(51, 149) = 2.47 Prob > F = 0.0000
0 Response to Omitted dummy variable because of collinearity
Post a Comment