六先蒋、公開數(shù)據(jù)的Stata實操
以上論文解析都沒有使用數(shù)據(jù)復(fù)現(xiàn)，下面將使用一篇論文的公開數(shù)據(jù)與代碼簡單演示一下樣本選擇模型與處理效應(yīng)模型在Stata中的操作嚷掠。
需要提前說明的是：
- 這篇論文出現(xiàn)的問題本質(zhì)上是自選擇偏差酬凳，作者誤認(rèn)為是樣本選擇偏差互纯，并錯誤地使用了樣本選擇模型尚蝌。
- 由于在Stata的
etregress命令下迎变，第一步選擇方程的被解釋變量D只能是0-1型的虛擬變量，如果使用連續(xù)型變量飘言，Stata將會報錯衣形。 - 在
D為連續(xù)型變量的情況下，為了使得Stata不報錯姿鸿，選擇方程使用D是否取值的虛擬變量谆吴，第二步方程直接引入D作為核心解釋變量倒源，這樣第二步回歸模型中將會同時出現(xiàn)D以及D是否取值的虛擬變量，這種情況下可能產(chǎn)生多重共線性問題纪铺。 - 為了避免出現(xiàn)多重共線性問題相速，公開代碼使用
heckman命令而非etregress命令碟渺，因為heckman命令下第一步回歸的被解釋變量不會自動帶入第二步回歸鲜锚，這或許也是論文作者選擇heckman而非etregress的動機。但需要說明的是苫拍，這樣的做法本質(zhì)上是不嚴(yán)謹(jǐn)?shù)奈叻保煌哪Ｐ瓦m用于不同的問題，模型混用可能導(dǎo)致估計結(jié)果的嚴(yán)重偏誤绒极，雖然最后得到的結(jié)果可能符合預(yù)期骏令，但不嚴(yán)謹(jǐn)?shù)膶嵶C設(shè)計很難說服讀者接受其結(jié)論。 - 為了避免模型混用導(dǎo)致的估計偏誤垄提，推文中的代碼將使用手工兩步法進行處理效應(yīng)模型的估計榔袋。
這篇論文是杜勇等（2021）發(fā)表在《中國工業(yè)經(jīng)濟》的《共同機構(gòu)所有權(quán)與企業(yè)盈余管理》，研究主題是探討共同機構(gòu)持股對企業(yè)盈余管理的影響铡俐。
[10] 杜勇, 孫帆, 鄧旭. 共同機構(gòu)所有權(quán)與企業(yè)盈余管理[J]. 中國工業(yè)經(jīng)濟, 2021(06): 155-173.
由于上文已對兩個命令進行了足夠詳細(xì)的講解凰兑，并且對相關(guān)問題進行了提前說明，因此下面直接放出代碼與結(jié)果审丘，代碼與結(jié)果的具體解讀可以參考前文吏够。
**# 【數(shù)據(jù)來源】
*- 杜勇等（2021），參見在《中國工業(yè)經(jīng)濟》網(wǎng)站（http://ciejournal.ajcass.org/Magazine/show/?id=77795）
**# 【參考文獻】
*- 杜勇, 孫帆, 鄧旭. 共同機構(gòu)所有權(quán)與企業(yè)盈余管理[J]. 中國工業(yè)經(jīng)濟, 2021(06): 155-173.
********************************************************************************
*- Stata Version: 16 | 17
*- 定義路徑
cd "C:\Users\KEMOSABE\Desktop\heckman"
use 數(shù)據(jù).dta, clear
*- 連續(xù)變量縮尾處理
winsor2 da1 da2 da11 rem ///
coz2 coz3 coz_power coz_number ///
coz22 coz33 coz2_year coz3_year ///
director djg auditfees epcm ///
net institution size leverage ///
roa growth toptenrate independent ///
magpay boardshare invrec analyst, ///
cut(1 99) replace
*- 設(shè)置控制變量的全局暫元
global CV institution size leverage roa ///
growth toptenrate dual independent ///
magpay boardshare invrec analyst ///
opin aud
*- 將控制變量及核心解釋變量滯后一期
xtset code year
foreach i of global CV {
gen Lag`i' = L.`i'
}
gen Lagcoz1 = L.coz1
*- 設(shè)置第一階段回歸控制變量的全局暫元
global CV1 Laginstitution Lagsize Lagleverage Lagroa ///
Laggrowth Lagtoptenrate Lagdual Lagindependent ///
Lagmagpay Lagboardshare Laginvrec Laganalyst ///
Lagopin Lagaud
**# 一滩报、coz1的基準(zhǔn)回歸及穩(wěn)健性檢驗
*- （1）基準(zhǔn)回歸
qui: reg da1 coz1 $CV i.year i.industry, r
est store OLS1
*- （2）處理效應(yīng)模型（MLE）
qui: etregress da1 $CV i.year i.industry, ///
treat(coz1 = Lagcoz1 $CV i.year i.industry) nolog vce(robust)
est store et_ML1
/*
Linear regression with endogenous treatment Number of obs = 20,018
Estimator: maximum likelihood Wald chi2(31) = 1485.50
Log pseudolikelihood = 18287.002 Prob > chi2 = 0.0000
------------------------------------------------------------------------------
| Robust
| Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
da1 |
institution | .0001648 .0001052 1.57 0.117 -.0000414 .000371
size | -.0066258 .0008814 -7.52 0.000 -.0083534 -.0048982
leverage | .0237043 .0048317 4.91 0.000 .0142344 .0331742
roa | -.2253829 .0170078 -13.25 0.000 -.2587175 -.1920482
growth | .0353113 .0023365 15.11 0.000 .0307318 .0398909
toptenrate | .0001971 .0000445 4.43 0.000 .0001099 .0002844
dual | .0031043 .0015126 2.05 0.040 .0001396 .006069
independent | .0040687 .0115344 0.35 0.724 -.0185383 .0266758
magpay | .4069487 .1409151 2.89 0.004 .1307602 .6831373
boardshare | -.0090979 .0037856 -2.40 0.016 -.0165174 -.0016784
invrec | .0183075 .0051727 3.54 0.000 .0081693 .0284458
analyst | .0005457 .000082 6.65 0.000 .0003849 .0007065
opin | -.0389673 .0040999 -9.50 0.000 -.047003 -.0309316
aud | -.0038764 .0012816 -3.02 0.002 -.0063884 -.0013644
|
year |
2009 | -.0008564 .0043254 -0.20 0.843 -.0093341 .0076213
2010 | .0186483 .0049932 3.73 0.000 .0088619 .0284348
2011 | .0220373 .0042826 5.15 0.000 .0136435 .030431
2012 | -.0158729 .0037267 -4.26 0.000 -.0231771 -.0085688
2013 | -.0161189 .0036 -4.48 0.000 -.0231748 -.009063
2014 | -.0161046 .0035979 -4.48 0.000 -.0231564 -.0090528
2015 | -.0031965 .0037801 -0.85 0.398 -.0106053 .0042122
2016 | -.0027102 .0037549 -0.72 0.470 -.0100697 .0046493
2017 | -.0153086 .0036145 -4.24 0.000 -.0223928 -.0082243
2018 | -.0154719 .0036009 -4.30 0.000 -.0225295 -.0084143
2019 | -.0178076 .0035699 -4.99 0.000 -.0248046 -.0108107
|
industry |
2 | -.0297887 .0105573 -2.82 0.005 -.0504806 -.0090968
3 | -.0116159 .0109897 -1.06 0.291 -.0331553 .0099234
4 | -.0282726 .0115443 -2.45 0.014 -.0508991 -.0056462
5 | -.0434586 .0104201 -4.17 0.000 -.0638816 -.0230356
6 | -.0323764 .0107643 -3.01 0.003 -.0534741 -.0112787
|
1.coz1 | -.0067129 .0019325 -3.47 0.001 -.0105006 -.0029253
_cons | .2775419 .0214512 12.94 0.000 .2354982 .3195855
-------------+----------------------------------------------------------------
coz1 |
Lagcoz1 | 3.234637 .041428 78.08 0.000 3.15344 3.315835
institution | .0037066 .0029546 1.25 0.210 -.0020843 .0094976
size | .1103075 .0245284 4.50 0.000 .0622327 .1583822
leverage | -.0127087 .1239879 -0.10 0.918 -.2557205 .2303031
roa | -.5487199 .328454 -1.67 0.095 -1.192478 .0950382
growth | .083781 .037123 2.26 0.024 .0110212 .1565408
toptenrate | .0032139 .0012742 2.52 0.012 .0007165 .0057114
dual | -.0820718 .0508186 -1.61 0.106 -.1816744 .0175308
independent | -1.012848 .3659513 -2.77 0.006 -1.730099 -.2955967
magpay | 2.53631 4.014366 0.63 0.528 -5.331703 10.40432
boardshare | -1.021367 .1736045 -5.88 0.000 -1.361625 -.6811082
invrec | -.1793844 .1334323 -1.34 0.179 -.440907 .0821381
analyst | -.0033357 .0025439 -1.31 0.190 -.0083216 .0016502
opin | .1161474 .1078862 1.08 0.282 -.0953057 .3276005
aud | .0256347 .0394008 0.65 0.515 -.0515895 .1028589
|
year |
2009 | .0545774 .1108197 0.49 0.622 -.1626253 .2717801
2010 | .2064666 .1102673 1.87 0.061 -.0096534 .4225866
2011 | -.0294345 .1096026 -0.27 0.788 -.2442517 .1853827
2012 | .0080566 .1085939 0.07 0.941 -.2047835 .2208966
2013 | -.0643272 .1033437 -0.62 0.534 -.2668772 .1382227
2014 | -.0427743 .103273 -0.41 0.679 -.2451857 .1596371
2015 | -.0567423 .1105243 -0.51 0.608 -.273366 .1598813
2016 | -.0928139 .1030316 -0.90 0.368 -.2947522 .1091244
2017 | -.2477218 .106496 -2.33 0.020 -.45645 -.0389935
2018 | -.0962982 .1050334 -0.92 0.359 -.3021598 .1095634
2019 | .0256564 .0998903 0.26 0.797 -.1701251 .2214378
|
industry |
2 | -.2286483 .2099405 -1.09 0.276 -.6401241 .1828274
3 | -.3286038 .2173325 -1.51 0.131 -.7545676 .09736
4 | -.4177377 .259037 -1.61 0.107 -.9254409 .0899654
5 | .069847 .2032102 0.34 0.731 -.3284377 .4681318
6 | .0932101 .210735 0.44 0.658 -.3198229 .5062431
|
_cons | -4.316258 .5725048 -7.54 0.000 -5.438347 -3.194169
-------------+----------------------------------------------------------------
/athrho | .0413626 .0199663 2.07 0.038 .0022294 .0804958
/lnsigma | -2.451377 .0118703 -206.51 0.000 -2.474643 -2.428112
-------------+----------------------------------------------------------------
rho | .041339 .0199322 .0022294 .0803224
sigma | .0861748 .0010229 .0841931 .0882032
lambda | .0035624 .0017194 .0001924 .0069323
------------------------------------------------------------------------------
Wald test of indep. eqns. (rho = 0): chi2(1) = 4.29 Prob > chi2 = 0.0383
*/
*- （3）處理效應(yīng)模型（TwoStep）
qui: etregress da1 $CV i.year i.industry, ///
treat(coz1 = Lagcoz1 $CV i.year i.industry) twostep
est store et_2S1
/*
Linear regression with endogenous treatment Number of obs = 20018
Estimator: two-step Wald chi2(61) = 3629.46
Prob > chi2 = 0.0000
------------------------------------------------------------------------------
| Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
da1 |
institution | .0001648 .0000966 1.71 0.088 -.0000245 .0003541
size | -.0066256 .0007952 -8.33 0.000 -.0081841 -.0050671
leverage | .0237045 .0038194 6.21 0.000 .0162187 .0311904
roa | -.2253832 .0098235 -22.94 0.000 -.2446369 -.2061294
growth | .0353113 .0011144 31.69 0.000 .0331271 .0374955
toptenrate | .0001971 .000043 4.58 0.000 .0001128 .0002814
dual | .0031042 .0014823 2.09 0.036 .0001989 .0060095
independent | .0040678 .0114332 0.36 0.722 -.0183409 .0264765
magpay | .4069558 .1104819 3.68 0.000 .1904153 .6234964
boardshare | -.0090989 .0041262 -2.21 0.027 -.017186 -.0010117
invrec | .0183072 .0041006 4.46 0.000 .0102702 .0263441
analyst | .0005457 .0000836 6.53 0.000 .0003818 .0007096
opin | -.0389672 .0030457 -12.79 0.000 -.0449367 -.0329978
aud | -.0038764 .0012695 -3.05 0.002 -.0063645 -.0013882
|
year |
2009 | -.0008564 .0038108 -0.22 0.822 -.0083254 .0066127
2010 | .0186484 .003792 4.92 0.000 .0112163 .0260805
2011 | .0220373 .003709 5.94 0.000 .0147678 .0293068
2012 | -.0158729 .0035715 -4.44 0.000 -.0228729 -.008873
2013 | -.0161189 .0035103 -4.59 0.000 -.022999 -.0092388
2014 | -.0161047 .0034962 -4.61 0.000 -.022957 -.0092523
2015 | -.0031967 .0035138 -0.91 0.363 -.0100836 .0036903
2016 | -.0027104 .0035086 -0.77 0.440 -.0095872 .0041665
2017 | -.0153088 .0034914 -4.38 0.000 -.0221519 -.0084658
2018 | -.0154721 .0034815 -4.44 0.000 -.0222958 -.0086485
2019 | -.0178078 .0034585 -5.15 0.000 -.0245864 -.0110293
|
industry |
2 | -.0297886 .0066066 -4.51 0.000 -.0427372 -.0168399
3 | -.0116162 .0069175 -1.68 0.093 -.0251743 .001942
4 | -.0282727 .0074022 -3.82 0.000 -.0427807 -.0137646
5 | -.0434582 .0064804 -6.71 0.000 -.0561595 -.0307569
6 | -.0323759 .006855 -4.72 0.000 -.0458115 -.0189404
|
coz1 | -.0067176 .0021868 -3.07 0.002 -.0110037 -.0024315
_cons | .2775378 .0181209 15.32 0.000 .2420215 .3130541
-------------+----------------------------------------------------------------
coz1 |
Lagcoz1 | 3.235202 .0413025 78.33 0.000 3.154251 3.316153
institution | .0036154 .0029089 1.24 0.214 -.002086 .0093168
size | .1121405 .0239223 4.69 0.000 .0652536 .1590274
leverage | -.0208505 .1217953 -0.17 0.864 -.2595648 .2178638
roa | -.5578723 .3217238 -1.73 0.083 -1.188439 .0726948
growth | .0879266 .0303994 2.89 0.004 .0283449 .1475083
toptenrate | .0032304 .0013442 2.40 0.016 .0005958 .0058649
dual | -.079793 .050784 -1.57 0.116 -.1793278 .0197418
independent | -1.019378 .3750888 -2.72 0.007 -1.754539 -.2842178
magpay | 2.68412 3.683406 0.73 0.466 -4.535223 9.903462
boardshare | -1.018715 .1736654 -5.87 0.000 -1.359093 -.6783367
invrec | -.178663 .132241 -1.35 0.177 -.4378506 .0805247
analyst | -.0033943 .0025893 -1.31 0.190 -.0084692 .0016805
opin | .1167083 .0995624 1.17 0.241 -.0784304 .3118469
aud | .0260504 .040476 0.64 0.520 -.053281 .1053818
|
year |
2009 | .0515999 .1140643 0.45 0.651 -.171962 .2751618
2010 | .2019266 .109724 1.84 0.066 -.0131284 .4169816
2011 | -.0315077 .1130668 -0.28 0.781 -.2531146 .1900991
2012 | .0078514 .1080623 0.07 0.942 -.2039468 .2196496
2013 | -.0658881 .108644 -0.61 0.544 -.2788264 .1470502
2014 | -.0454649 .1080572 -0.42 0.674 -.2572531 .1663232
2015 | -.0564083 .1072322 -0.53 0.599 -.2665795 .1537629
2016 | -.092317 .1085734 -0.85 0.395 -.3051169 .1204829
2017 | -.2501913 .1096922 -2.28 0.023 -.4651841 -.0351984
2018 | -.0990409 .1078524 -0.92 0.358 -.3104276 .1123459
2019 | .0223512 .1059781 0.21 0.833 -.1853622 .2300645
|
industry |
2 | -.2381904 .204628 -1.16 0.244 -.6392539 .1628731
3 | -.3332723 .2179531 -1.53 0.126 -.7604526 .093908
4 | -.4154694 .2532426 -1.64 0.101 -.9118158 .080877
5 | .0617126 .1984843 0.31 0.756 -.3273095 .4507348
6 | .0875193 .2074342 0.42 0.673 -.3190443 .4940829
|
_cons | -4.344332 .5487066 -7.92 0.000 -5.419777 -3.268887
-------------+----------------------------------------------------------------
hazard |
lambda | .0035703 .001916 1.86 0.062 -.0001849 .0073256
-------------+----------------------------------------------------------------
rho | 0.04143
sigma | .08617484
------------------------------------------------------------------------------
*/
*- （4）手工處理（TwoStep）
qui: probit coz1 Lagcoz1 $CV i.year i.industry
est store First1
predict y1, xb
gen imr1 = normalden(y1) / normal(y1)
replace imr1 = -normalden(y1) / normal(-y1) if coz1 == 0
qui: reg da1 coz1 $CV i.year i.industry imr1, r
est store Second1
*- 結(jié)果導(dǎo)出
local mlist1 "OLS1 et_ML1 et_2S1 First1 Second1"
esttab `mlist1' using coz1的基準(zhǔn)回歸及穩(wěn)健性檢驗.rtf, replace ///
b(%6.4f) t(%6.4f) ///
scalar(N r2_a r2_p) compress nogap ///
star(* 0.1 ** 0.05 *** 0.01) ///
indicate("Year FE=*.year" "Industry FE=*.industry") ///
mtitle(`mlist1') ///
title("Baseline Regression and Robustness Test of coz1")
/*
Baseline Regression and Robustness Test of coz1
---------------------------------------------------------------------------
(1) (2) (3) (4) (5)
LSDV1 et_mle1 et_2s1 first1 second1
---------------------------------------------------------------------------
main
coz1 -0.0048*** -0.0067*** -0.0067***
(-2.9786) (-3.0719) (-3.4445)
institut~n 0.0002 0.0002 0.0002* 0.0036 0.0002
(1.5850) (1.5665) (1.7065) (1.2429) (1.5654)
size -0.0066*** -0.0066*** -0.0066*** 0.1121*** -0.0066***
(-7.9796) (-7.5170) (-8.3323) (4.6877) (-7.5054)
leverage 0.0212*** 0.0237*** 0.0237*** -0.0209 0.0237***
(4.5995) (4.9060) (6.2064) (-0.1712) (4.9027)
roa -0.2110*** -0.2254*** -0.2254*** -0.5579* -0.2254***
(-12.8244) (-13.2518) (-22.9432) (-1.7340) (-13.2449)
growth 0.0345*** 0.0353*** 0.0353*** 0.0879*** 0.0353***
(15.5953) (15.1127) (31.6865) (2.8924) (15.1130)
toptenrate 0.0002*** 0.0002*** 0.0002*** 0.0032** 0.0002***
(5.4173) (4.4277) (4.5837) (2.4032) (4.4238)
dual 0.0028* 0.0031** 0.0031** -0.0798 0.0031**
(1.9385) (2.0523) (2.0941) (-1.5712) (2.0507)
independ~t 0.0036 0.0041 0.0041 -1.0194*** 0.0041
(0.3212) (0.3527) (0.3558) (-2.7177) (0.3524)
magpay 0.3756*** 0.4069*** 0.4070*** 2.6841 0.4070***
(2.9158) (2.8879) (3.6835) (0.7287) (2.8857)
boardshare -0.0024 -0.0091** -0.0091** -1.0187*** -0.0091**
(-0.6691) (-2.4033) (-2.2052) (-5.8660) (-2.4013)
invrec 0.0230*** 0.0183*** 0.0183*** -0.1787 0.0183***
(4.6611) (3.5393) (4.4646) (-1.3510) (3.5365)
analyst 0.0006*** 0.0005*** 0.0005*** -0.0034 0.0005***
(7.1090) (6.6524) (6.5269) (-1.3109) (6.6451)
opin -0.0409*** -0.0390*** -0.0390*** 0.1167 -0.0390***
(-10.2467) (-9.5044) (-12.7942) (1.1722) (-9.4991)
aud -0.0043*** -0.0039*** -0.0039*** 0.0261 -0.0039***
(-3.5288) (-3.0246) (-3.0534) (0.6436) (-3.0220)
1.coz1 -0.0067***
(-3.4737)
Lagcoz1 3.2352***
(78.3295)
imr1 0.0036*
(1.8696)
_cons 0.2681*** 0.2775*** 0.2775*** -4.3443*** 0.2775***
(13.4216) (12.9383) (15.3159) (-7.9174) (12.9241)
---------------------------------------------------------------------------
coz1
Lagcoz1 3.2346*** 3.2352***
(78.0784) (78.3295)
institut~n 0.0037 0.0036
(1.2545) (1.2429)
size 0.1103*** 0.1121***
(4.4971) (4.6877)
leverage -0.0127 -0.0209
(-0.1025) (-0.1712)
roa -0.5487* -0.5579*
(-1.6706) (-1.7340)
growth 0.0838** 0.0879***
(2.2568) (2.8924)
toptenrate 0.0032** 0.0032**
(2.5222) (2.4032)
dual -0.0821 -0.0798
(-1.6150) (-1.5712)
independ~t -1.0128*** -1.0194***
(-2.7677) (-2.7177)
magpay 2.5363 2.6841
(0.6318) (0.7287)
boardshare -1.0214*** -1.0187***
(-5.8833) (-5.8660)
invrec -0.1794 -0.1787
(-1.3444) (-1.3510)
analyst -0.0033 -0.0034
(-1.3113) (-1.3109)
opin 0.1161 0.1167
(1.0766) (1.1722)
aud 0.0256 0.0261
(0.6506) (0.6436)
_cons -4.3163*** -4.3443***
(-7.5393) (-7.9174)
---------------------------------------------------------------------------
/
athrho 0.0414**
(2.0716)
lnsigma -2.4514***
(-2.1e+02)
---------------------------------------------------------------------------
hazard
lambda 0.0036*
(1.8635)
Year FE Yes Yes Yes Yes Yes
Industry~E Yes Yes Yes Yes Yes
---------------------------------------------------------------------------
N 22591 20018 20018 20018 20018
r2_a 0.1387 0.1441
r2_p 0.6912
---------------------------------------------------------------------------
t statistics in parentheses
* p<0.1, ** p<0.05, *** p<0.01
*/
**# 二锅知、coz2的基準(zhǔn)回歸及穩(wěn)健性檢驗
*- （1）基準(zhǔn)回歸
qui: reg da1 coz2 $CV i.year i.industry, r
est store OLS2
*- （2）樣本選擇模型（TwoStep）
qui: heckman da1 coz2 $CV i.year i.industry, ///
select( coz2 = $CV1 ) twostep
est store he_2S2
/*
Heckman selection model -- two-step estimates Number of obs = 20,018
(regression model with sample selection) Selected = 2,593
Nonselected = 17,425
Wald chi2(31) = 559.43
Prob > chi2 = 0.0000
--------------------------------------------------------------------------------
| Coef. Std. Err. z P>|z| [95% Conf. Interval]
---------------+----------------------------------------------------------------
da1 |
coz2 | -.0208381 .0114122 -1.83 0.068 -.0432055 .0015293
institution | -.0001099 .0002544 -0.43 0.666 -.0006085 .0003886
size | .0065137 .0025792 2.53 0.012 .0014586 .0115689
leverage | .0287133 .0115701 2.48 0.013 .0060363 .0513904
roa | -.1696233 .0264688 -6.41 0.000 -.2215012 -.1177455
growth | .0406549 .0028657 14.19 0.000 .0350382 .0462716
toptenrate | .0005991 .0001313 4.56 0.000 .0003418 .0008563
dual | .004107 .0047744 0.86 0.390 -.0052507 .0134647
independent | -.1094472 .0341261 -3.21 0.001 -.1763332 -.0425612
magpay | .7141059 .3357257 2.13 0.033 .0560956 1.372116
boardshare | -.1155465 .030857 -3.74 0.000 -.1760251 -.055068
invrec | -.0200946 .0138371 -1.45 0.146 -.0472147 .0070255
analyst | -3.14e-06 .0002186 -0.01 0.989 -.0004316 .0004253
opin | -.0348136 .0083493 -4.17 0.000 -.051178 -.0184492
aud | .0026224 .0036027 0.73 0.467 -.0044387 .0096835
|
year |
2009 | -.0133637 .0087203 -1.53 0.125 -.0304551 .0037278
2010 | -.0023983 .0085558 -0.28 0.779 -.0191672 .0143707
2011 | .007618 .0084863 0.90 0.369 -.0090149 .0242508
2012 | -.018136 .0082114 -2.21 0.027 -.03423 -.002042
2013 | -.0343626 .0082714 -4.15 0.000 -.0505743 -.0181509
2014 | -.0268081 .0082289 -3.26 0.001 -.0429364 -.0106798
2015 | .0043649 .0082808 0.53 0.598 -.0118652 .020595
2016 | -.0049588 .0082864 -0.60 0.550 -.0212 .0112823
2017 | -.0219129 .0084003 -2.61 0.009 -.0383771 -.0054486
2018 | -.0298953 .008332 -3.59 0.000 -.0462257 -.0135649
2019 | -.028234 .0081615 -3.46 0.001 -.0442303 -.0122377
|
industry |
2 | -.0652461 .0191914 -3.40 0.001 -.1028605 -.0276318
3 | -.0279979 .0207284 -1.35 0.177 -.0686249 .012629
4 | -.0100713 .0240308 -0.42 0.675 -.0571708 .0370283
5 | -.0641525 .0188057 -3.41 0.001 -.101011 -.027294
6 | -.0458935 .0192651 -2.38 0.017 -.0836525 -.0081345
|
_cons | -.1057979 .077478 -1.37 0.172 -.257652 .0460562
---------------+----------------------------------------------------------------
coz2 |
Laginstitution | .000229 .0017509 0.13 0.896 -.0032026 .0036607
Lagsize | .1220394 .0129369 9.43 0.000 .0966835 .1473953
Lagleverage | .3218294 .0709585 4.54 0.000 .1827532 .4609055
Lagroa | -.3267603 .204901 -1.59 0.111 -.7283589 .0748383
Laggrowth | -.0533554 .0214025 -2.49 0.013 -.0953036 -.0114072
Lagtoptenrate | .0049017 .0008002 6.13 0.000 .0033334 .0064701
Lagdual | -.1151072 .030836 -3.73 0.000 -.1755447 -.0546697
Lagindependent | -1.470211 .2309171 -6.37 0.000 -1.9228 -1.017622
Lagmagpay | 1.025986 2.201411 0.47 0.641 -3.2887 5.340672
Lagboardshare | -1.582939 .1017037 -15.56 0.000 -1.782274 -1.383604
Laginvrec | -.6018562 .0724373 -8.31 0.000 -.7438308 -.4598817
Laganalyst | -.0040897 .0015752 -2.60 0.009 -.0071769 -.0010024
Lagopin | .0822428 .0619452 1.33 0.184 -.0391676 .2036531
Lagaud | .0487422 .0246017 1.98 0.048 .0005237 .0969607
_cons | -3.507714 .2810605 -12.48 0.000 -4.058582 -2.956845
---------------+----------------------------------------------------------------
/mills |
lambda | .0948201 .0173606 5.46 0.000 .0607939 .1288462
---------------+----------------------------------------------------------------
rho | 0.84225
sigma | .11257936
--------------------------------------------------------------------------------
*/
*- （3）處理效應(yīng)模型 - 手工處理（TwoStep）
qui: probit coz1 Lagcoz1 $CV i.year i.industry
est store First2
predict y2, xb
gen imr2 = normalden(y2) / normal(y2)
replace imr2 = -normalden(y2) / normal(-y2) if coz2 == 0
qui: reg da1 coz2 $CV i.year i.industry imr2, r
est store Second2
*- 結(jié)果導(dǎo)出
local mlist2 "OLS2 he_2S2 First2 Second2"
esttab `mlist2' using coz2的基準(zhǔn)回歸及穩(wěn)健性檢驗.rtf, replace ///
b(%6.4f) t(%6.4f) ///
scalar(N r2_a r2_p) compress nogap ///
star(* 0.1 ** 0.05 *** 0.01) ///
indicate("Year FE=*.year" "Industry FE=*.industry") ///
mtitle(`mlist2') ///
title("Baseline Regression and Robustness Test of coz2")
/*
Baseline Regression and Robustness Test of coz2
--------------------------------------------------------------
(1) (2) (3) (4)
OLS2 he_2S2 First2 Second2
--------------------------------------------------------------
main
coz2 -0.0090*** -0.0208* -0.0109***
(-3.5991) (-1.8260) (-3.8498)
institut~n 0.0002 -0.0001 0.0036 0.0002
(1.5711) (-0.4321) (1.2429) (1.5472)
size -0.0066*** 0.0065** 0.1121*** -0.0066***
(-7.9283) (2.5255) (4.6877) (-7.4912)
leverage 0.0213*** 0.0287** -0.0209 0.0237***
(4.6120) (2.4817) (-0.1712) (4.9087)
roa -0.2110*** -0.1696*** -0.5579* -0.2253***
(-12.8242) (-6.4084) (-1.7340) (-13.2397)
growth 0.0344*** 0.0407*** 0.0879*** 0.0353***
(15.5849) (14.1866) (2.8924) (15.1052)
toptenrate 0.0002*** 0.0006*** 0.0032** 0.0002***
(5.4342) (4.5641) (2.4032) (4.4256)
dual 0.0027* 0.0041 -0.0798 0.0031**
(1.9266) (0.8602) (-1.5712) (2.0504)
independ~t 0.0034 -0.1094*** -1.0194*** 0.0040
(0.3042) (-3.2071) (-2.7177) (0.3470)
magpay 0.3763*** 0.7141** 2.6841 0.4066***
(2.9216) (2.1271) (0.7287) (2.8833)
boardshare -0.0026 -0.1155*** -1.0187*** -0.0092**
(-0.7150) (-3.7446) (-5.8660) (-2.4146)
invrec 0.0229*** -0.0201 -0.1787 0.0182***
(4.6381) (-1.4522) (-1.3510) (3.5220)
analyst 0.0006*** -0.0000 -0.0034 0.0005***
(7.0944) (-0.0144) (-1.3109) (6.6466)
opin -0.0408*** -0.0348*** 0.1167 -0.0389***
(-10.2360) (-4.1696) (1.1722) (-9.4915)
aud -0.0043*** 0.0026 0.0261 -0.0039***
(-3.5238) (0.7279) (0.6436) (-3.0258)
Lagcoz1 3.2352***
(78.3295)
imr2 0.0031*
(1.6981)
_cons 0.2673*** -0.1058 -4.3443*** 0.2773***
(13.3690) (-1.3655) (-7.9174) (12.9075)
--------------------------------------------------------------
coz2
Laginsti~n 0.0002
(0.1308)
Lagsize 0.1220***
(9.4334)
Laglever~e 0.3218***
(4.5355)
Lagroa -0.3268
(-1.5947)
Laggrowth -0.0534**
(-2.4929)
Lagtopte~e 0.0049***
(6.1256)
Lagdual -0.1151***
(-3.7329)
Lagindep~t -1.4702***
(-6.3668)
Lagmagpay 1.0260
(0.4661)
Lagboard~e -1.5829***
(-15.5642)
Laginvrec -0.6019***
(-8.3086)
Laganalyst -0.0041***
(-2.5964)
Lagopin 0.0822
(1.3277)
Lagaud 0.0487**
(1.9813)
_cons -3.5077***
(-12.4803)
--------------------------------------------------------------
/mills
lambda 0.0948***
(5.4618)
Year FE Yes Yes Yes Yes
Industry~E Yes Yes Yes Yes
--------------------------------------------------------------
N 22591 20018 20018 20018
r2_a 0.1388 0.1442
r2_p 0.6912
--------------------------------------------------------------
t statistics in parentheses
* p<0.1, ** p<0.05, *** p<0.01
*/
的基準(zhǔn)回歸及穩(wěn)健性檢驗的第三段代碼見公眾號下載附件（公眾號后臺回復(fù)
heckman即可獲取下載鏈接）。
