第11章のStataコード
第11章 操作変数法
11.2 操作変数法の実施：2段階最小二乗法
11.2.1 2段階最小二乗法
import delimited "cigarette.csv", case(preserve) clear
regress bw cigpacks Source | SS df MS Number of obs = 1,000
-------------+---------------------------------- F(1, 998) = 1632.56
Model | 83763793.8 1 83763793.8 Prob > F = 0.0000
Residual | 51205762.7 998 51308.3795 R-squared = 0.6206
-------------+---------------------------------- Adj R-squared = 0.6202
Total | 134969557 999 135104.661 Root MSE = 226.51
------------------------------------------------------------------------------
bw | Coefficient Std. err. t P>|t| [95% conf. interval]
-------------+----------------------------------------------------------------
cigpacks | -62.40191 1.544414 -40.40 0.000 -65.43258 -59.37124
_cons | 3423.228 17.02299 201.09 0.000 3389.823 3456.633
------------------------------------------------------------------------------quietly regress cigpacks price
predict fitted, xb
regress bw fitted Source | SS df MS Number of obs = 1,000
-------------+---------------------------------- F(1, 998) = 686.04
Model | 54983766.9 1 54983766.9 Prob > F = 0.0000
Residual | 79985789.6 998 80146.0818 R-squared = 0.4074
-------------+---------------------------------- Adj R-squared = 0.4068
Total | 134969557 999 135104.661 Root MSE = 283.1
------------------------------------------------------------------------------
bw | Coefficient Std. err. t P>|t| [95% conf. interval]
-------------+----------------------------------------------------------------
fitted | -53.26758 2.0337 -26.19 0.000 -57.2584 -49.27676
_cons | 3331.894 22.21839 149.96 0.000 3288.294 3375.494
------------------------------------------------------------------------------Rによるデータ演習
import delimited "proximity.csv", case(preserve) clear
list in 1/4 +---------------------------------------------------------+
| lwage educ exper black south smsa nearc4 |
|---------------------------------------------------------|
1. | 6.306275 7 16 1 0 1 0 |
2. | 6.175867 12 9 0 0 1 0 |
3. | 6.580639 12 16 0 0 1 0 |
4. | 5.521461 11 10 0 0 1 1 |
+---------------------------------------------------------+regress lwage educ exper black south smsa
estimates store OLS Source | SS df MS Number of obs = 3,010
-------------+---------------------------------- F(5, 3004) = 232.21
Model | 165.205668 5 33.0411336 Prob > F = 0.0000
Residual | 427.435978 3,004 .142288941 R-squared = 0.2788
-------------+---------------------------------- Adj R-squared = 0.2776
Total | 592.641646 3,009 .196956346 Root MSE = .37721
------------------------------------------------------------------------------
lwage | Coefficient Std. err. t P>|t| [95% conf. interval]
-------------+----------------------------------------------------------------
educ | .073807 .0035336 20.89 0.000 .0668784 .0807356
exper | .0393134 .0021955 17.91 0.000 .0350085 .0436183
black | -.1882225 .0177678 -10.59 0.000 -.2230607 -.1533843
south | -.1290528 .0152285 -8.47 0.000 -.1589122 -.0991935
smsa | .1647411 .0156919 10.50 0.000 .1339732 .195509
_cons | 4.913331 .0631212 77.84 0.000 4.789566 5.037096
------------------------------------------------------------------------------ivregress 2sls lwage exper black south smsa (educ = nearc4)
estimates store TSLSInstrumental variables 2SLS regression Number of obs = 3,010
Wald chi2(5) = 673.47
Prob > chi2 = 0.0000
R-squared = 0.2140
Root MSE = .3934
------------------------------------------------------------------------------
lwage | Coefficient Std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
educ | .1318498 .0495237 2.66 0.008 .0347852 .2289144
exper | .0622698 .0196665 3.17 0.002 .0237241 .1008156
black | -.1296012 .0532094 -2.44 0.015 -.2338897 -.0253128
south | -.1092522 .0231533 -4.72 0.000 -.1546318 -.0638725
smsa | .1348259 .0302606 4.46 0.000 .0755162 .1941356
_cons | 3.939822 .8309333 4.74 0.000 2.311222 5.568421
------------------------------------------------------------------------------
Instrumented: educ
Instruments: exper black south smsa nearc4etable, estimates(OLS TSLS) showstars showstarsnote column(estimates) title(Dpendent Var. = Log of Wage)Dpendent Var. = Log of Wage
--------------------------------------------
OLS TSLS
--------------------------------------------
educ 0.074 ** 0.132 **
(0.004) (0.050)
exper 0.039 ** 0.062 **
(0.002) (0.020)
black -0.188 ** -0.130 *
(0.018) (0.053)
south -0.129 ** -0.109 **
(0.015) (0.023)
smsa 0.165 ** 0.135 **
(0.016) (0.030)
Intercept 4.913 ** 3.940 **
(0.063) (0.831)
Number of observations 3010 3010
--------------------------------------------
** p<.01, * p<.05