各位大神，我在做面板数据（小T大N，T=10）的固定效应回归时遇到的这个问题，在做固定效应模型的时间效应（双向固定效应）检验时，按照书上说的，把时间变量当虚拟变量加进去做回归，我是做分行业做A股上市公司的研究，结果做了3个行业都是year3的被omitted说是因为多重共线性。想请问下大神们，为啥会出现这个问题呢？我的x1,x2,x3,x4都是财务报表数据求得的财务比率，每股每年理论上都不同。另外，出现这个问题有啥解决办法不？谢谢大家了，第一次用stata，很多不懂的。

出问题的回归结果如下：

xtreg y x1 x2 x3 x4 year2-year10, fe vce (cluster stc)
note: year3 omitted because of collinearity
Fixed-effects (within) regression Number of obs = 351
Group variable: stc Number of groups = 56
R-sq:within= 0.5139 Obs per group: min = 1
between = 0.5553 avg = 6.3
overall = 0.4931 max = 9
F(12,55) = 15.83
corr(u_i, Xb)= 0.2404 Prob > F = 0.0000
(Std. Err. adjusted for 56 clusters in stc)
------------------------------------------------------------------------------
| Robust
p | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
x1| 2.941548 .6315635 4.66 0.000 1.675867 4.20723
x2| 2.106469 2.407112 0.88 0.385 -2.717492 6.93043
x3| .6100122 .6736124 0.91 0.369 -.7399371 1.959962
x4|-2.036041 3.329725 -0.61 0.543 -8.708959 4.636876
year2 |-1.533012 .7260395 -2.11 0.039 -2.988028 -.0779966
year3 | 0(omitted)
year4 | 11.45015 2.046779 5.59 0.000 7.348316 15.55199
year5 |-.1459721 1.25433 -0.12 0.908 -2.659707 2.367762
year6 | 6.267309 1.287942 4.87 0.000 3.686216 8.848401
year7 | 10.63463 2.347288 4.53 0.000 5.93056 15.3387
year8 |-1.243413 1.368683 -0.91 0.368 -3.986315 1.499489
year9 |-.6609542 1.288508 -0.51 0.610 -3.243181 1.921273
year10 |-2.924296 1.479134 -1.98 0.053 -5.888547 .0399558
_cons | -1.68107 2.328777 -0.72 0.473 -6.348044 2.985903
-------------+----------------------------------------------------------------
sigma_u |10.390442
sigma_e |7.5034695
rho |.65724518 (fraction of variance due to u_i)
------------------------------------------------------------------------------