欢迎来到天天文库
浏览记录
ID:34440420
大小:96.58 KB
页数:3页
时间:2019-03-06
《OLS_Variance_Unbiased_Estimate.pdf》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、UnivariateOLS:UnbiasedEstimatorofErrorVarianceConsidertheunivariatelinearregressionmodelyt=+xt+ut;t=1;:::;T:wheretheregressorsarenon-stochastic(fixed)andthedisturbanceshavezeromeanandareuncorrelatedandhomoscedasticwithvarianceequalto2,i.e.,E[u]=0;tC[u;u]=0,t6=s,andV[u]=E[u2]=2:Wewillshowthattstt1X
2、T22s=u^tT 2t=1isanunbiasedestimatorofthedisturbancevariance2;whereu^=y y^,y^=^+x^;tttttand^and^aretheOLSestimatorsofand,respectively.TheOLSestimates^and^aregivenbyPT^=t=1P(yt y)(xt x);^=y ^x;T2t=1(xt x)whichcanalsobeexpressedasXTXT^=wtyt;^=qtyt;t=1t=1wherext x1wt=PT2;qt= xwt:t=1(xt x)TThe
3、followingpropertiesareeasilyverified:XTXTXTXTwt=0;wtxt=1;qt=1;qtxt=0;t=1t=1t=1t=1andXTXT1PT2XT212Tt=1xt xwt=PT2;qt=PT2;qtwt=PT2:t=1t=1(xt x)t=1t=1(xt x)t=1t=1(xt x)1Usingtheseproperties,weobservethatXTXTXT^=wtyt=wt(+xt+ut)=+wtut;t=1t=1t=1andXTXTXT^=qtyt=qt(+xt+ut)=+qtut:t=1t=1t=1Notethattheresi
4、dualscanbeexpressedasu^t=yt y^t=yt ^+x^t=(+xt+ut) ^+x^t;orequivalentlyhiu^t= (^ )+^ xt+ut:Hence,hi2hiu^2=(^ )+^ x+u2 2(^ )+^ xutttttwhere!!XTXT(^ )+^ xt=qsus+wsusxt:s=1s=1ItfollowsthathhiiE(^ )+^ xtut""!!##XTXT=Eqsus+wsusxtuts=1s=12=(qt+wtxt);andsoXThhiiXTE(^ )+^ xu=(q+wx)2=(1+1
5、)2=22:tttttt=1t=12Moreover,hi2E(^ )+^ xt2"!!#23XTXT=E4qsus+wsusxt5s=1s=12!232!23"!!#XTXTXTXT=E4qu5+E4wu5x2+2Equwuxsssstssssts=1s=1s=1s=1"!!!#XTXTXT2222=qs+wsxt+2qswsxts=1s=1s=1"P#1Tx2x2xx=2Tt=1t+t 2t;PT2PT2PT2t=1(xt x)t=1(xt x)t=1(xt x)andsoXThi2E(^ )+^ xtt=1"PP#Tx2Tx2Tx2=2
6、t=1t+t=1t 2;PT2PT2PT2t=1(xt x)t=1(xt x)t=1(xt x)"P#1Tx2 x2=22Tt=1t1PT2Tt=1(xt x)2=2:PT22Finally,t=1E[ut]=TandsoXT22222Eu^t=2+T 22=(T 2)t=1whichyields"#1XT22Eu^t=:T 2t=13
此文档下载收益归作者所有
点击更多查看相关文章~~