|
2023年3月27日，计量经济学服务中心&数量经济学2023年因果推断专题讲座之DID专题讲座第一课于19:00-21:200顺利开讲。 往期部分Stata寒暑假班课程学员（2017--2023年）通过相关报名，参加了我们的讲座，下面就讲座内容做一介绍。 首先我们先来看下讲座大纲目录： 一、DID专题讲座内容1、简介/引言/理论部分 2、DID资源/推文 3、传统DID模型估计命令及交叠DID估计命令 4、DID模型 - 1、交错双重差分法 (staggered DiD)
- 2、广义双重差分法 (generalized DiD)
- 5、三重差分法 (triple differences)
5、传统DID专题 6、多期DID/渐进DID 7、PSM-DID、PSM-DDD 8、交叠DID 9、DID异质处理效应估计量
DID专题讲座第一课于3月27日(周一)19:00一直持续到21:20分,主要讲解了DID理论知识及操作应用。 双重差分法（Difference-In-Difference, DID）由于其清晰直观 ，易于操作，并且实际操作难度较低，上手简单等特点而广为应用，目前已经成为应用最广的因果推断方法（几乎没有之一），在政策评估中受到国内外学者青睐。 双重差分法，又称为倍分法，倍差法，或者差异中差异，用DID或者DD来表示。目前DID模型有很多类，可以分为标准DID，多期DID，交错DID，广义DID，队列双重差分法、模糊双重差分法、混合截面DID、PSM-DID、PSM-DDD以及其他双重差分法。 讲座从180多年前的约翰--斯诺的霍乱假说讲起，引入了DID双重差分的一个设计，同时为大家介绍了因果推断中的基础理论知识，包括随机实验以及自然实验，因果推断的基本前提假设，通过SUVTA以及前侧后侧事实比较等引入DID模型设计。 本次讲座主要为大家介绍了单一时点DID以及多时点DID的区别，同时针对传统的DID模型，讲解了最低工资法案的具体应用，最后通过普林斯顿大学Panel101数据为例，为大家介绍了传统DID模型中的7个常见命令的具体操作应用。 一图读懂：7个常规DID命令+13个最新DID命令 据悉DID专题讲座第2课也将于今晚3月29日（周三）19:00-21:00再次开讲，后续DID专题讲座新增课程也将近期通知。 面向往期学员推出SCM专题讲座也将于近期开讲，敬请请关注。 本次讲座部分板书截图如下，你也可以通过本文后面提供的文档进行操作学习。 
1、路径设置
*-------------------------------------------------------------------------------
. *-定义工作路径等常规设置
. global path 'E:\2023年3月DID专题讲座\Stata-202303\A17-diff'
. global D '$path\data' // 范例数据
. global R '$path\refs' // 参考文献
. global Out '$path\Out' // 结果：图形和表格
. adopath + '$path\adofiles' // 外部命令
. cd '$D'
*-------------------------------------------------------------------------------
.
2、导入数据查看数据. use 'Panel101.dta', clear
.
. ed
. desc
Contains data from Panel101.dta
Observations: 70
Variables: 9 3 Jan 2020 11:28
----------------------------------------------------------------------------------------------
Variable Storage Display Value
name type format label Variable label
----------------------------------------------------------------------------------------------
country long %14.0g country Country
year int %8.0g Year
y double %10.0g Outcome Y
y_bin float %9.0g Binary outcome Y
x1 float %9.0g Predictor x1
x2 float %9.0g Predictor x2
x3 float %9.0g Predictor x3
opinion float %18.0g agree Categorical variable
op float %9.0g
----------------------------------------------------------------------------------------------
Sorted by: country year
3、DID模型变量设置. * Create a dummy variable to indicate the time when the treatment started.
. * Lets assume that treatment started in 1994.
. * In this case, years before 1994 will have a value of 0 and 1994+ a 1.
. * If you already have this skip this step.设置虚拟变量，政策执行时间为1994年
. gen time = (year>=1994) & !missing(year)
.
. * Create a dummy variable to identify the group exposed to the treatment.
. * In this example lets assumed that countries with code 5,6, and 7 were treated (=1).
. * Countries 1-4 were not treated (=0). If you already have this skip this step生成地区的
> 虚拟变量
. gen treated = (country>4) & !missing(country)
.
. * Create an interaction between time and treated. We will call this interaction `did' 产
> 生交互项
. gen did = time*treated
4、第1个命令传统reg估计.
. * Estimating the DID estimator随后将这三个变量作为解释变量，y作为被解释变量进行回归：
. reg y time treated did, r
Linear regression Number of obs = 70
F(3, 66) = 2.17
Prob > F = 0.0998
R-squared = 0.0827
Root MSE = 3.0e+09
------------------------------------------------------------------------------
| Robust
y | Coefficient std. err. t P>|t| [95% conf. interval]
-------------+----------------------------------------------------------------
time | 2.29e+09 9.00e+08 2.54 0.013 4.92e+08 4.09e+09
treated | 1.78e+09 1.05e+09 1.70 0.094 -3.11e+08 3.86e+09
did | -2.52e+09 1.45e+09 -1.73 0.088 -5.42e+09 3.81e+08
_cons | 3.58e+08 7.61e+08 0.47 0.640 -1.16e+09 1.88e+09
------------------------------------------------------------------------------
. est store ols
.
end of do-file
. reg y time##treated, r
Linear regression Number of obs = 70
F(3, 66) = 2.17
Prob > F = 0.0998
R-squared = 0.0827
Root MSE = 3.0e+09
------------------------------------------------------------------------------
| Robust
y | Coefficient std. err. t P>|t| [95% conf. interval]
-------------+----------------------------------------------------------------
1.time | 2.29e+09 9.00e+08 2.54 0.013 4.92e+08 4.09e+09
1.treated | 1.78e+09 1.05e+09 1.70 0.094 -3.11e+08 3.86e+09
|
time#treated |
1 1 | -2.52e+09 1.45e+09 -1.73 0.088 -5.42e+09 3.81e+08
|
_cons | 3.58e+08 7.61e+08 0.47 0.640 -1.16e+09 1.88e+09
------------------------------------------------------------------------------
5、第2个命令diff操作. diff y, t(treated) p(time)
DIFFERENCE-IN-DIFFERENCES ESTIMATION RESULTS
Number of observations in the DIFF-IN-DIFF: 70
Before After
Control: 16 24 40
Treated: 12 18 30
28 42
--------------------------------------------------------
Outcome var. | y | S. Err. | |t| | P>|t|
----------------+---------+---------+---------+---------
Before | | | |
Control | 3.6e+08| | |
Treated | 2.1e+09| | |
Diff (T-C) | 1.8e+09| 1.1e+09| 1.58 | 0.120
After | | | |
Control | 2.6e+09| | |
Treated | 1.9e+09| | |
Diff (T-C) | -7.4e+08| 9.2e+08| 0.81 | 0.422
| | | |
Diff-in-Diff | -2.5e+09| 1.5e+09| 1.73 | 0.088*
--------------------------------------------------------
R-square: 0.08
* Means and Standard Errors are estimated by linear regression
**Inference: *** p<0.01; ** p<0.05; * p<0.1
.
end of do-file
. do 'C:\Users\Metrics\AppData\Local\Temp\STD3ac0_000000.tmp'
. diff y, t(treated) p(time)
DIFFERENCE-IN-DIFFERENCES ESTIMATION RESULTS
Number of observations in the DIFF-IN-DIFF: 70
Before After
Control: 16 24 40
Treated: 12 18 30
28 42
--------------------------------------------------------
Outcome var. | y | S. Err. | |t| | P>|t|
----------------+---------+---------+---------+---------
Before | | | |
Control | 3.6e+08| | |
Treated | 2.1e+09| | |
Diff (T-C) | 1.8e+09| 1.1e+09| 1.58 | 0.120
After | | | |
Control | 2.6e+09| | |
Treated | 1.9e+09| | |
Diff (T-C) | -7.4e+08| 9.2e+08| 0.81 | 0.422
| | | |
Diff-in-Diff | -2.5e+09| 1.5e+09| 1.73 | 0.088*
--------------------------------------------------------
R-square: 0.08
* Means and Standard Errors are estimated by linear regression
**Inference: *** p<0.01; ** p<0.05; * p<0.1
. est store diff
6、第3个命令xtreg. xtset country year
Panel variable: country (strongly balanced)
Time variable: year, 1990 to 1999
Delta: 1 unit
. xtreg y did i.year,r fe
Fixed-effects (within) regression Number of obs = 70
Group variable: country Number of groups = 7
R-squared: Obs per group:
Within = 0.2661 min = 10
Between = 0.0116 avg = 10.0
Overall = 0.1568 max = 10
F(6,6) = .
corr(u_i, Xb) = -0.2753 Prob > F = .
(Std. err. adjusted for 7 clusters in country)
------------------------------------------------------------------------------
| Robust
y | Coefficient std. err. t P>|t| [95% conf. interval]
-------------+----------------------------------------------------------------
did | -2.52e+09 1.09e+09 -2.32 0.059 -5.18e+09 1.38e+08
|
year |
1991 | 6.17e+08 1.41e+09 0.44 0.676 -2.82e+09 4.06e+09
1992 | 6.47e+08 8.04e+08 0.80 0.452 -1.32e+09 2.61e+09
1993 | 3.20e+09 1.56e+09 2.05 0.086 -6.10e+08 7.01e+09
1994 | 4.74e+09 1.66e+09 2.85 0.029 6.76e+08 8.80e+09
1995 | 2.62e+09 1.90e+09 1.38 0.218 -2.04e+09 7.28e+09
1996 | 3.49e+09 1.06e+09 3.29 0.017 8.94e+08 6.09e+09
1997 | 4.80e+09 1.14e+09 4.22 0.006 2.02e+09 7.58e+09
1998 | 2.07e+09 1.31e+09 1.58 0.164 -1.13e+09 5.28e+09
1999 | 2.70e+09 1.96e+09 1.38 0.216 -2.08e+09 7.49e+09
|
_cons | 3871104 7.97e+08 0.00 0.996 -1.95e+09 1.96e+09
-------------+----------------------------------------------------------------
sigma_u | 1.615e+09
sigma_e | 2.693e+09
rho | .26458903 (fraction of variance due to u_i)
------------------------------------------------------------------------------
. est sto xtreg
7、第4个命令areg. areg y did i.year ,absorb(country ) r
Linear regression, absorbing indicators Number of obs = 70
Absorbed variable: country No. of categories = 7
F(10, 53) = 3.87
Prob > F = 0.0006
R-squared = 0.3874
Adj R-squared = 0.2024
Root MSE = 2.693e+09
------------------------------------------------------------------------------
| Robust
y | Coefficient std. err. t P>|t| [95% conf. interval]
-------------+----------------------------------------------------------------
did | -2.52e+09 1.29e+09 -1.96 0.055 -5.10e+09 5.91e+07
|
year |
1991 | 6.17e+08 1.33e+09 0.46 0.644 -2.05e+09 3.28e+09
1992 | 6.47e+08 8.73e+08 0.74 0.462 -1.10e+09 2.40e+09
1993 | 3.20e+09 1.20e+09 2.67 0.010 7.92e+08 5.60e+09
1994 | 4.74e+09 1.23e+09 3.84 0.000 2.26e+09 7.21e+09
1995 | 2.62e+09 1.83e+09 1.43 0.158 -1.05e+09 6.29e+09
1996 | 3.49e+09 1.18e+09 2.96 0.005 1.13e+09 5.86e+09
1997 | 4.80e+09 1.16e+09 4.15 0.000 2.48e+09 7.12e+09
1998 | 2.07e+09 1.55e+09 1.33 0.188 -1.04e+09 5.19e+09
1999 | 2.70e+09 1.75e+09 1.54 0.129 -8.15e+08 6.22e+09
|
_cons | 3871104 7.85e+08 0.00 0.996 -1.57e+09 1.58e+09
------------------------------------------------------------------------------
. est sto areg
.
end of do-file
8、第5个命令reghdfe. . reghdfe y did,absorb(country year )
(MWFE estimator converged in 2 iterations)
HDFE Linear regression Number of obs = 70
Absorbing 2 HDFE groups F( 1, 53) = 3.60
Prob > F = 0.0632
R-squared = 0.3874
Adj R-squared = 0.2024
Within R-sq. = 0.0636
Root MSE = 2.693e+09
------------------------------------------------------------------------------
y | Coefficient Std. err. t P>|t| [95% conf. interval]
-------------+----------------------------------------------------------------
did | -2.52e+09 1.33e+09 -1.90 0.063 -5.18e+09 1.43e+08
_cons | 2.49e+09 4.69e+08 5.31 0.000 1.55e+09 3.43e+09
------------------------------------------------------------------------------
Absorbed degrees of freedom:
-----------------------------------------------------+
Absorbed FE | Categories - Redundant = Num. Coefs |
-------------+---------------------------------------|
country | 7 0 7 |
year | 10 1 9 |
-----------------------------------------------------+
.
9、其他2个命令：didregress+ xtdidregressDID最新应用必读文章： 2023年第3期《数量经济技术经济研究》目录及计量方法汇总表（DID、DDD等） 《数量经济技术经济研究》上2篇最新DID论文：交错DID及小样本双重差分RI-DID DID前沿论文推荐 |《中国工业经济》：交错DID及异质性—稳健DID（附代码实现） csdid2：多时期DID的异质性稳健估计量 一图读懂：7个常规DID命令+13个最新DID命令 推荐：《中国工业经济》2023年第2期-稳健DID估计量+Goodman Bacon分解 permute：DID安慰剂检验随机抽样500/1000次--2023年《中国工业经济》最新应用 2023新版_DID进展汇总：命令、书单、论文、文章资源汇总 理解DID出了什么问题?双向固定效应模型TWFE与异质性处理效应drdid和csdid DRDID--双重稳健估计量 Bacon分解bacondecomp新+旧版本操作及ddtiming命令（三合一）操作应用 免费公开课：交叠DID偏误及Bacon分解+案例应用 微观计量最新进展及最新DID规范动作、文献、命令等资源推荐 资源推荐：直接动态展示培根分解、事件研究和交错处理、断点回归、随机化推理 csdid:多时期DID的异质性稳健估计量 理解DID出了什么问题?双向固定效应模型TWFE与异质性处理效应drdid和csdid 书籍推荐：《因果推断：混音带》（内涵高级DID、合成控制法、机器学习和因果推理等资源课件） 双重差分法（DID）平行趋势及安慰剂检验方法案例合集 csdid:多时期DID的异质性稳健估计量 综合控制、合成DID最新书单、命令包汇总 微观计量最新进展及最新DID规范动作、文献、命令等资源推荐
|
