stata教程四部曲,这是第四部分:工具变量。选取的工具变量一般和自变量高度相关,和干扰项不相关,继而采取2sls、3sls进行估计
IntroductiontoStata–Lecture4:Instrumentalvariables
HayleyFisher
3March2010
Keyreferences:CameronandTrivedi(2009)chapter6,AngristandPischke(2009)chapter4,Wooldridge(2009)chapter15,Greene(2008)chapter13.
1Introduction
ForOLStogiveconsistentestimatorstheerrortermmustbeunrelatedtotheregressors–thatisE(u|x)=0.Thisisoftennotanassumptionthatcanbemade.Onecommonapproachistouseaninstrumentalvariable(IV)estimator.Theinstrumentalvariable,z,needstobecorrelatedwiththeendogenousvariablex,anduncorrelatedwiththeerrortermsoE(u|z)=0.Findingavalidandstronginstrumentisoftenveryhard–seeAngristandPischke(2009)forafulldiscussion.
ThislecturefocusesontheimplementationofIVestimationinStataandtherelatedtestsavailable.IusethedatasetfromAngristandKrueger’s(1991)articleexaminingtheeconomicreturntoschoolingusingquarterofbirthanditsinteractionwithcompulsoryschoolattendancelawsasaninstrument.Thepaperisavailableatdoc.guandang.net:
lwagei=α+βschoolingi+γXi+εi(1)
However,thereisakeyunmeasureddeterminantofwages–ability.Thismeansthatabilityissubsumedintotheerrortermεi.Sinceabilityisalsolikelylinkedtoschooling,thismeansthatE(ε|schooling)=0.SoanOLSestimateofβisinconsistent.
ThekeytoAngristandKrueger’sapproachistorecognisethatthereisarelationshipbetweenquarterofbirthandtheamountofschoolinganindividualreceives:
“...moststatesrequirestudentstoenterschoolinthecalendaryearinwhichtheyturn6.Schoolstartageisthereforeafunctionofdateofbirth.Speci cally,thosebornlateintheyearareyoungfortheirgrade.InstateswithaDecember31birthdaycuto ,childrenborninthefourthquarterenterschoolshortlybeforetheyturn6,whilethoseborninthe rstquarterenterschoolataround61
.Furthermore,becausecompulsoryschoolinglawstypicallyrequire
studentstoremaininschoolonlyuntiltheir16thbirthday,thesegroupsofstudentswillbeindi erentgrades,orthroughagivengradetoadi erentdegree,whentheyreachthelegaldropoutage.Thecombinationofschoolstart-agepoliciesandcompulsoryschoolinglawscreatesanaturalexperimentinwhichchildrenarecompelledtoattendschoolfordi erentlengthsoftime,dependingontheirbirthdays.”
AngristandPischke(2009)p.117
Thismeansthatquarterofbirthcanbeusedasaninstrumentalvariableforschooling–itiscorrelatedwithschoolingbutthereisnoreasontobelievethatitalsoa ectswages.
Stata III-7-4 XT Dynamic panel data_经管营销_专业资料。STAT的详细教程,...注:Stata自动从滞后2阶开始作为工具变量,用户需要设定最高的滞后阶 数的个数。...
stata入门4 ——图形篇 如何用stata做出漂亮的统计图 .8 .6 .6 .8 .4 ...连续变量的选项 ? ? ? bin(#) : set number of bins to # width(#):...
4 《STATA应用高级培训教程》 南开大学数量经济研究所 王群勇 STATA简介:发展...工具变量回归、广义线性回归、分位数回归(稳健回 归)、系统方程模型(SUR、联立...
1.2.2 工具变量估计 设回归模型为: y=Xβ+u (1.4) 其中,解释变量为 X...1.3.5 内生变量的显著性检验 1. 单个内生变量的显著性检验 在 stata 中,...
4 《STATA应用高级培训教程》 南开大学数量经济研究所 王群勇 STATA简介:发展...工具变量回归、广义线性回归、分位数回归(稳健回 归)、系统方程模型(SUR、联立...
最前沿的stata教程。内附有命令及使用方法。Stata上机实验 工具变量(IV) ? 什么...4 5 6 18.9 17.4 19 20 18.1 19.7 19.1 18.4 19.6 20.4 18....
对4.0的WINDOWS版本,Stata可以管理4800个记录×99个变量;对WINDOWS 95下的5....四、 Stata的矩阵运算功能 矩阵代数是多元统计分析的重要工具,Stata提供了多元...
Cragg-Donald Wald F 统计量 4. Kleibergen-Paap Wald rk F 统计量” Stata命令:ivreg2 如果存在弱工具变量该怎么办? 1. 如果有很多工具变量,有部分强工具...
stata中文教程第四讲_经济学_高等教育_教育专区。stata中文教程...11.4 13.8 12 操作如下: gen d=x1-x2 swilk d 产生配对差值的变量 d ...
热门文档