.

. 为什么会得到负的卡方检验值呢？这种情况下可否认为固定效应是合理的呢？

. hausman fe

Note: the rank of the differenced variance matrix (9) does not equal the number of coefficients being
        tested (10); be sure this is what you expect, or there may be problems computing the test.  Examine
        the output of your estimators for anything unexpected and possibly consider scaling your variables
        so that the coefficients are on a similar scale.

                 ---- Coefficients ----
             |      (b)          (B)            (b-B)     sqrt(diag(V_b-V_B))
             |       fe           re         Difference          S.E.
-------------+----------------------------------------------------------------
         dar |   -.0442345    -.0527817        .0085472         .004907
       fsize |   -.2448338     .2065122        -.451346        .5233349
       gsale |    .0056962     .0061054       -.0004092        .0002903
        herf |    .0012748     .0004844        .0007904        .0003361
       herf2 |   -1.18e-07    -4.56e-08       -7.20e-08        3.70e-08
       dumt2 |    1.329218     1.295918        .0333008        .0247267
       dumt3 |    2.502135     2.461852        .0402829        .0260247
       dumt4 |    3.334255      3.26244        .0718158         .028847
       dumt5 |    3.161338     3.074924        .0864145        .0404706
       dumt6 |    3.496019     3.413124        .0828951        .0449345
------------------------------------------------------------------------------
                           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(9) = (b-B)'[(V_b-V_B)^(-1)](b-B)
                          =   -21.43    chi2<0 ==> model fitted on these
                                        data fails to meet the asymptotic
                                        assumptions of the Hausman test;
                                        see suest for a generalized test