[回归分析求助] 双向固定效应下时间虚拟变量结果与引入时间趋势结果相差过大？ [推广有奖]

11楼

pheadena 发表于 2013-9-23 01:51:28

Fixed-effects (within) regression             Number of obs    =    598
Group variable: country                      Number of groups =       26

R-sq:  within  = 0.3623                      Obs per group: min =       23
   between = 0.0242                                     avg =    23.0
   overall = 0.1293                                     max =       23

                                             F(15,25)          = 105.47
corr(u_i, Xb)  = -0.7658                      Prob > F          = 0.0000

                           (Std. Err. adjusted for 26 clusters in country)
------------------------------------------------------------------------------
         |             Robust
   growth |    Coef. Std. Err.    t P>|t|    [95% Conf. Interval]
-------------+----------------------------------------------------------------
   ylag1 | .1524533 .0331354    4.60 0.000    .0842097 .2206969
   ylag2 | .1649073 .0242579    6.80 0.000    .1149472 .2148673
   ylag3 |  -.1066755 .0423174 -2.52 0.018 -.1938298 -.0195212
   lineq | 20.39012 5.606949    3.64 0.001    8.842395 31.93785
   lineq2 |  -1.937149 .7185533 -2.70 0.012 -3.417037 -.4572606
      sk | .053383 .0361196    1.48 0.152 -.0210067 .1277726
      lab | .2145648 .0838741    2.56 0.017    .0418229 .3873067
      gov | .0539975 .0756708    0.71 0.482 -.1018495 .2098445
      urb |  -.0503701 .0244163 -2.06 0.050 -.1006564 -.0000837
      free | .0060708 .0247289    0.25 0.808 -.0448593    .057001
      inf | -4.27646 1.666315 -2.57 0.017 -7.708301 -.8446198
      t1 | 6.75664 1.567241    4.31 0.000    3.528848 9.984433
      t2 |  -.9951898 .1997647 -4.98 0.000 -1.406613 -.5837668
      t3 | .0496738 .0102516    4.85 0.000    .0285601 .0707874
      t4 |  -.0008149 .0001792 -4.55 0.000 -.0011839 -.0004459
   _cons |  -47.32324 8.917071 -5.31 0.000 -65.68829 -28.95819
-------------+----------------------------------------------------------------
   sigma_u |  5.9182493
   sigma_e |  6.1940427
      rho |  .47724209 (fraction of variance due to u_i)
------------------------------------------------------------------------------

12楼

平衡木的心 发表于 2015-7-13 14:11:05

楼主您好，您这篇帖子给了我很棒的启发，有一个比较初级的问题想要请教您，谢谢！
在控制时间趋势的时候设的t1 t2 t3 t4分别是时间趋势、二、三、四次项，那么二、三、四次项的含义是什么呢？假设时间趋势数列为1 2 3 4 ……那么t2 t3 t4数列是否就是t1的数学乘方呢？
如果想要利用自变量的原始数据和回归方程来计算得到因变量的取值是否就是把各变量取值与系数相乘（包括t1 t2 t3 t4)呢？
多谢！