求助！stata门限面板回归模型xthreg命令结果中 _cat#c.表格中的回归系数该怎么解释

8735

收藏 2019-04-18

求助！stata门限面板回归模型xthreg命令结果中 _cat#c表格中的回归系数代表什么，该怎么解释，其中的系数分别应该和门槛变量的哪一个区间对应

程序如下：
. xthreg ei ,rx(ur) qx(gdp) thnum(2) bs(100,100) trim(0.05,0.05) grid(400)

结果如下：
Estimating  the  threshold  parameters: 1st ......  2nd ......  Done
Boostrap for single threshold
.................................................. + 50
.................................................. +  100
Boostrap for double threshold model:
.................................................. + 50
.................................................. +  100

Threshold estimator (level = 95):

model    Threshold       Lower       Upper

Th-1 73856.0000 73002.0000 74016.0000
Th-21 73856.0000 73002.0000 74016.0000
Th-22 38572.0000 38130.5000 38909.0000

Threshold effect test (bootstrap = 100 100):

Threshold       RSS       MSE    Fstat Prob Crit10 Crit5 Crit1

Single 18.3264    0.0679    74.77  0.0100  29.9380  38.1389  48.8694
Double 16.2775    0.0603    33.98  0.0200  24.3504  27.9518  39.1289

Fixed-effects (within) regression             Number of obs    =    280
Group variable: province1                      Number of groups =       28

R-sq:  within  = 0.4878                      Obs per group: min =       10
between = 0.3558                                     avg =    10.0
overall = 0.3711                                     max =       10

F(3,249)          =    79.04
corr(u_i, Xb)  = -0.4822                      Prob > F          = 0.0000

ei    Coef. Std. Err.    t P>t    [95% Conf. Interval]

_cat#c.ur
0    .0175802 .0056805    3.09 0.002    .0063923 .0287681
1    .0232043 .0050716    4.58 0.000    .0132156    .033193
2    .032115 .0049242    6.52 0.000    .0224167 .0418133

_cons .2418701    .28917    0.84 0.404 -.3276608 .8114011

sigma_u .46929019
sigma_e .25567885
rho .77111155 (fraction of variance due to u_i)

F test that all u_i=0:    F(27, 249) = 25.08          Prob > F = 0.0000

扫码加我拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

全部回复

lvchunwen

2019-4-18 01:35:50

如果对变量都取对数
程序：
xthreg lnei,rx(lnur) qx(lngdp) thnum(2) bs(100,100) trim(0.05,0.05) grid(400)

结果：
Estimating  the  threshold  parameters: 1st ......  2nd ......  Done
Boostrap for single threshold
.................................................. + 50
.................................................. +  100
Boostrap for double threshold model:
.................................................. + 50
.................................................. +  100

Threshold estimator (level = 95):

model    Threshold       Lower       Upper

Th-1    11.1442    11.1329    11.1711
Th-21    11.1324    11.1268    11.1380
Th-22    10.5603    10.5487    10.5690

Threshold effect test (bootstrap = 100 100):

Threshold       RSS       MSE    Fstat Prob Crit10 Crit5 Crit1

Single    8.4641    0.0313    47.03  0.0200  28.9665  37.5807  47.9113
Double    7.1172    0.0264    51.10  0.0000  18.9818  21.9562  28.1769

Fixed-effects (within) regression             Number of obs    =    280
Group variable: province1                      Number of groups =       28

R-sq:  within  = 0.4430                      Obs per group: min =       10
between = 0.2121                                     avg =    10.0
overall = 0.2596                                     max =       10

F(3,249)          =    66.01
corr(u_i, Xb)  = -0.2442                      Prob > F          = 0.0000

lnei    Coef. Std. Err.    t P>t    [95% Conf. Interval]

_cat#c.lnur
0    .4475518 .1683134    2.66 0.008    .1160524 .7790513
1    .5048399 .1632545    3.09 0.002    .1833041 .8263758
2    .584672 .1610306    3.63 0.000    .2675164 .9018276

_cons -1.651741 .6566617 -2.52 0.013 -2.945061 -.358422

sigma_u .34679913
sigma_e .17023074
rho .80583675 (fraction of variance due to u_i)

F test that all u_i=0:    F(27, 249) = 38.76          Prob > F = 0.0000

问题一：标注的是_cat#c.lnur  是不是说明回归系数都应该从lnur这个解释变量入手来解释，那如果考虑门限变量lngdp对被解释变量lnei的影响，应该看哪个系数，怎么解释呢？

问题二：对变量取对数之后的模型给出的门槛值是不是需要计算一下转化成原来的样子，就是根据lngdp的门槛值算出gdp的门槛值？