GDPit = β0 + β1lnFDI1it + β2TD2it + β3lnLF3it + β4INF4it + β5HDI5it + εit
where GDPit denotes GDP growth (annual %), lnFDI1it denotes logarithm of Chinese net
overseas investment received by 10 Asian countries (value: US MN) TD2it denotes trade
(% of GDP), lnLF3it denotes logarithm of total labor force, INF4it denotes inflation
(annual %), lnHDI5it denotes Human Development Index, β0 is a constant, β1–β5 are the
estimated coefficients, ε is the residual term, i refers to the selected 10 Asian countries, andt is the time span.
以上是数据的描述
首先,因为净对外投资数据有负数,我在FDI的所有数据加了一个常量 30 (因负数最小值为27)
接着开始测试多重线性问题
. correlate fdi1 trade labor1 inf hdi
(obs=100)
| fdi1 trade labor1 inf hdi
-------------+---------------------------------------------
fdi1 | 1.0000
trade | 0.4140 1.0000
labor1 | 0.2430 -0.2634 1.0000
inf | -0.0121 -0.1704 0.2720 1.0000
hdi | 0.0607 0.6601 -0.4094 -0.3094 1.0000
绝对值小于 0.80
接着测试VIF
. regress gdp fdi1 trade labor1 inf hdi
Source | SS df MS Number of obs = 100
-------------+---------------------------------- F(5, 94) = 16.90
Model | 434.429276 5 86.8858551 Prob > F = 0.0000
Residual | 483.188008 94 5.14029795 R-squared = 0.4734
-------------+---------------------------------- Adj R-squared = 0.4454
Total | 917.617283 99 9.26886145 Root MSE = 2.2672
------------------------------------------------------------------------------
gdp | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
fdi1 | .5669492 .2095794 2.71 0.008 .1508245 .983074
trade | .0064563 .003921 1.65 0.103 -.001329 .0142415
labor1 | .1695892 .1512321 1.12 0.265 -.1306857 .469864
inf | .0895231 .0389604 2.30 0.024 .0121664 .1668799
hdi | -14.9037 2.802432 -5.32 0.000 -20.46799 -9.339405
_cons | 8.228503 3.324058 2.48 0.015 1.628508 14.8285
------------------------------------------------------------------------------
. vif
Variable | VIF 1/VIF
-------------+----------------------
trade | 2.44 0.409606
hdi | 2.17 0.459944
fdi1 | 1.51 0.660868
labor1 | 1.40 0.714556
inf | 1.15 0.871323
-------------+----------------------
Mean VIF | 1.74
小于10
说明没有多重共线性问题
接着单位根检测
xtunitroot fisher gdp, pperron lags(0) pvalue = 0.0000
xtunitroot fisher fdi1, pperron lags(0) pvalue = 0.0243
xtunitroot fisher fdi1, pperron lags(1) pvalue = 0.0133
xtunitroot fisher trade, pperron lags(0) pvalue = 0.1168
xtunitroot fisher trade, pperron lags(1) pvalue =0.0985
xtunitroot fisher inf, pperron lags(0) pvalue = 0.0000
labor1也是0
故,面板是稳定的,没有单位根问题
接着固定效应和随机效应检测
. xtreg gdp fdi1 trade labor1 inf hdi, fe
Fixed-effects (within) regression Number of obs = 100
Group variable: country Number of groups = 10
R-sq: Obs per group:
within = 0.1073 min = 10
between = 0.2537 avg = 10.0
overall = 0.1521 max = 10
F(5,85) = 2.04
corr(u_i, Xb) = -0.9627 Prob > F = 0.0806
------------------------------------------------------------------------------
gdp | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
fdi1 | .2573467 .3029278 0.85 0.398 -.3449549 .8596482
trade | .0119573 .0161928 0.74 0.462 -.0202382 .0441528
labor1 | 4.775972 8.187781 0.58 0.561 -11.50353 21.05547
inf | .1151159 .0394299 2.92 0.004 .0367187 .1935131
hdi | 4.866999 24.73528 0.20 0.844 -44.31337 54.04737
_cons | -79.54596 120.6869 -0.66 0.512 -319.5038 160.4119
-------------+----------------------------------------------------------------
sigma_u | 7.8919162
sigma_e | 2.0812669
rho | .93497356 (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0: F(9, 85) = 2.95 Prob > F = 0.0043
P小于0.01 说明固定效应显著
.
. xtreg gdp fdi1 trade labor1 inf hdi, re
Random-effects GLS regression Number of obs = 100
Group variable: country Number of groups = 10
R-sq: Obs per group:
within = 0.0686 min = 10
between = 0.7987 avg = 10.0
overall = 0.4677 max = 10
Wald chi2(5) = 23.52
corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0003
------------------------------------------------------------------------------
gdp | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
fdi1 | .5490596 .2369088 2.32 0.020 .0847268 1.013392
trade | .0048386 .0063132 0.77 0.443 -.007535 .0172122
labor1 | .2558804 .3039163 0.84 0.400 -.3397845 .8515453
inf | .0920676 .0371967 2.48 0.013 .0191634 .1649717
hdi | -11.7938 5.143388 -2.29 0.022 -21.87466 -1.712947
_cons | 4.951306 6.588381 0.75 0.452 -7.961683 17.8643
-------------+----------------------------------------------------------------
sigma_u | 1.3798903
sigma_e | 2.0812669
rho | .3053508 (fraction of variance due to u_i)
------------------------------------------------------------------------------
. xttest0
Breusch and Pagan Lagrangian multiplier test for random effects
gdp[country,t] = Xb + u[country] + e[country,t]
Estimated results:
| Var sd = sqrt(Var)
---------+-----------------------------
gdp | 9.268861 3.04448
e | 4.331672 2.081267
u | 1.904097 1.37989
Test: Var(u) = 0
chibar2(01) = 5.01
Prob > chibar2 = 0.0126
.
P小于0.05 说明随机效应显著
接着豪斯曼检测
. hausman fe_result re_result
---- Coefficients ----
| (b) (B) (b-B) sqrt(diag(V_b-V_B))
| fe_result re_result Difference S.E.
-------------+----------------------------------------------------------------
fdi1 | .2573467 .5490596 -.2917129 .1887841
trade | .0119573 .0048386 .0071187 .0149114
labor1 | 4.775972 .2558804 4.520092 8.182139
inf | .1151159 .0920676 .0230483 .0130814
hdi | 4.866999 -11.7938 16.6608 24.19462
------------------------------------------------------------------------------
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(5) = (b-B)'[(V_b-V_B)^(-1)](b-B)
= 4.44
Prob>chi2 = 0.4875
.