用两年双向固定效应回归时,一些省份虚拟变量被ommt,第二年被ommit掉了,这不就等于没加时间效应一样吗,请问我该怎么处理这种问题呢?而且我发现不加个体年龄2016就不会被共线掉,加上年龄year2016就会被共线掉(换成年龄平方一样),但是一般不都会有年龄这一控制变量嘛,请问有没有解决办法,还是我不用管这种情况呢?xtset fid year
panel variable: fid (strongly balanced)
time variable: year, 2014 to 2016, but with gaps
delta: 1 unit
tabulate provcd, gen(dummy)
xtreg y x1 huzhu urban14 ln_shouru xiaofei familysize join_market debt age employ2014 jobclass_base badhealth goodhealth
> yanglao yibao dummy1- dummy27 i.year,fe r
note: dummy2 omitted because of collinearity
note: dummy3 omitted because of collinearity
note: dummy6 omitted because of collinearity
note: dummy8 omitted because of collinearity
note: dummy10 omitted because of collinearity
note: dummy19 omitted because of collinearity
note: dummy20 omitted because of collinearity
note: dummy22 omitted because of collinearity
note: dummy23 omitted because of collinearity
note: dummy24 omitted because of collinearity
note: dummy25 omitted because of collinearity
note: dummy26 omitted because of collinearity
note: dummy27 omitted because of collinearity
note: 2016.year omitted because of collinearity
Fixed-effects (within) regression Number of obs = 9,568
Group variable: fid Number of groups = 4,784
R-sq: Obs per group:
within = 0.6668 min = 2
between = 0.4745 avg = 2.0
overall = 0.5365 max = 2