一、面板数据如何从混合最小二乘、固定效应、随即效应中选择
1、混合最小二乘、固定效应
.xtreg vol1 FI share1 share2 share3 , fe Fixed-effects (within) regreion Number of obs = 289Group variable: code Number of groups = 36R-sq: within = 0.0210 Obs per group: min = 7 between = 0.0259 avg = 8.0 overall = 0.0095 max = 11 F(4,249) = 1.34corr(u_i, Xb) = -0.7580 Prob > F = 0.2569 vol1 Coef.Std.Err.t P>|t| [95% Conf.Interval] FI .0274564 .0192466 1.43 0.155 -.0104505 .0653634 share1 .0043386 .1112655 0.04 0.969 -.2148029 .2234801 share2 .0075372 .111155 0.07 0.946 -.2113867 .2264611 share3 .0058343 .1113069 0.05 0.958 -.2133888 .2250574 _cons -.5947285 11.1241 -0.05 0.957 -22.50405 21.31459 sigma_u .02625677 sigma_e .05223138 rho .20172957 (fraction of variance due to u_i) F test that all u_i=0: F(35, 249) = 0.87 Prob > F = 0.6879
F检验，原假设为个固定效应都相同，拒绝则选择固定效应，反之为混合最小二乘。——此时选混合最小二乘
2、随即效应还是混合最小二乘
.xtreg roa_a stateshr size tl , reRandom-effects GLS regreion Number of obs = 77Group variable: id Number of groups = 14R-sq: within = 0.4859 Obs per group: min = 3 between = 0.8053 avg = 5.5 overall = 0.6927 max = 11Random effects u_i ~ Gauian Wald chi2(3) = 106.77corr(u_i, X) = 0 (aumed) Prob > chi2 = 0.0000 roa_a Coef.Std.Err.z P>|z| [95% Conf.Interval] stateshr -.0023371 .0011081 -2.11 0.035 -.0045089 -.0001653 size .0008558 .000203 4.22 0.000 .0004578 .0012537 tl -.1132165 .0112583 -10.06 0.000 -.1352823 -.0911507 _cons .0922799 .0109761 8.41 0.000 .0707672 .1137926 sigma_u .00103144 sigma_e .00170742 rho .2673593 (fraction of variance due to u_i) ..xttest0Breusch and Pagan Lagrangian multiplier test for random effects roa_a[id,t] = Xb + u[id] + e[id,t] Estimated results: Var sd = sqrt(Var) roa_a .0000114 .0033722 e 2.92e-06 .0017074 u 1.06e-06 .0010314 Test: Var(u) = 0 chi2(1) = 2.87 Prob > chi2 = 0.0903.
原假设为var(u)=0，拒绝则采用随即效应——此时选随机
3、固定还是随即 .hausman fe re Coefficients (b) (B) (b-B) sqrt(diag(V_b-V_B)) fe re Difference S.E. stateshr -.0208393 -.0416243 .0207849 .022253 size .0287271 .0156759 .0130511 .0068713 tl .0134735 .2663396 -.252866 .2696783 b = consistent under Ho and Ha; obtained from xtreg B = inconsistent under Ha, efficient under Ho; obtained from xtreg Test: Ho: difference in coefficients not systematic chi2(3) = (b-B)\'[(V_b-V_B)^(-1)](b-B) = 3.61 Prob>chi2 = 0.3070
原假设为随机，拒绝则选固定——此时选随机
二、缩尾会使显著性降低
