Spectral density bandwith choice and prewightening in the estimation of heteroskadasticity and autocorrelation consistent covariance matrices in panel data models

vshfwudo hvwlpdwlrq surfhgxuhv1 Wkhlghdlvwr dwwhq wkh vshfwudo ghqvlw| e| sdvvlqj
wkh ruljlqdo vhulhv wkurxjk d owhu vr wkdw lwv rxwsxw kdv d uhodwlyho| dw vshfwuxp1 Ddw
vshfwuxp lv pxfk hdvlhu wr hvwlpdwh dqg wkh fruuhvsrqglqj nhuqho hvwlpdwru lv ohvv vhqvleoh wr
wkhfkrlfhridedqgzlgwk1DqguhzvdqgPrqdkdq*v+4<<5, suhzklwhqlqjnhuqhohvwlpdwrulv
hhfwlyh lq uhgxflqj wkh eldv/ dqg ohdgv wr frqvlghudeo| ehwwhu vl}hv iru uhodwhg whvw vwdwlvwlfv1
Lq wkh phdqwlph/ lw lv dovr irxqg wkdw suhzklwhqlqj lqdwhv wkh yduldqfh dqg pd| ohdg wr
d odujhu phdq vtxduhg huuru +PVH, wkdq wkh nhuqho hvwlpdwru zlwkrxw suhzklwhqlqj +vhh
Dqguhzv dqg Prqdkdq 4<<5/ Qhzh| dqg Zhvw 4<<7/ s1967,1

Rxu frqwulexwlrqv lq wklv sdshu lv wkdw zh frpsduh wkh qlwh vdpsoh shuirupdqfhv ri
nhuqho0edvhg hvwlpdwruv= qrqsdudphwulf zlwk dqg zlwkrxw suhzklwhqlqj dqg sdudphwulf
phwkrg ri gh Kddq dqg Ohylq +5333, lq wkh sdqho frqwh{wv1

Lq Vhfwlrq 5/ zh ghvfuleh wkh iudphzrun lq zklfk hvwlpdwlrq ri khwhurvnhgdvwlflw| dqg
dxwrfruuhodwlrq frqvlvwhqw fryduldqfh pdwulfhv lq sdqho gdwd ri lqwhuhvw1 Lq Vhfwlrq 6/ zh
uhsruw wkh Prqwh Fduor uhvxowv1 Vhfwlrq 7 surylghv d frqfoxglqj uhpdun dqg gluhfwlrqv iru
ixuwkhuuhvhdufk1Xqohvvlqglfdwhg/dooolplwvduhwdnhqdvwkhvdpsohvl}h q $4dqg W $4=

51 PRGHO

Wr prwlydwh wkh sureohp/ zh uvw frqvlghu dq wzr0zd| huuru frpsrqhqw sdqho prgho zlwk
d srvvleo| khwhurvnhgdvwlf dqg dxwrfruuhodwhg glvwxuedqfh huuru

Y_lt ` α<sub>0</sub> n x'<sub>a</sub>β<sub>o</sub> n μ<sub>i</sub> П λ<sub>t</sub> П y<sub>it</sub>, W ` • > = ==> T_i> ι ` • > = = = > q> q> T 5 ]⁺ >       (2.1)

zkhuh Y_it is a vfdθdu> X_it is d s × • yhfWru ri h{sOdqdWru∣ Vduidblhs WkdW pd∣ frqWaiq Odjjhg
ghShqghqW yauiaeOhs Y_it~⅛ (s> k 5 ]⁺), «o is aq iqWhUfhsW/ β₀ is a s × • yhfWru ri Wkh sOrsh
SdUdPhWhUS/ μ_i is Wkh iqgiyigxaO hffhfW, λ_t is Wkh Wiph hffhfW, aqg v_it is Wkh huuru Whum. Zh
aOOrz fi{hg hffhfWs ru uaqgrp hffhfWs. WkurxjkrxW, zh assxph T_i ` c_iT iru some iqWhjhu T
aqg c_i 5 df>FO = Wkxs, zh ShumiW xqeaOaqfhg saqhO gaWa. Pruhryhu, zh aOOrz Y_it>X_it>aQ aqg
β₀ Wr ghshqg rq erWk q aqg T= (Iru qrWaWirqaO simsOifiW∣, zh sxssuhss sxfk ghshqghqfh.)
WkurxjkrxW, zh assxmh Wkh irOOrziqj frqgiWirqs rq (2.1)=

DvvχpsWιrq Dι4= {Y_it>X_ij auh sWrfkasWif surfhsshs sxfk WkaW (i) iru hafk l> {y_ij is
fryauiaqfh0sWaWirqau∣ ziWk H ( V_i t) ‘ f > H ( vf_s) ‘ σ_i 5 d f> F O aqg H ( v_iɔ 5 d f> F ]; (ii) Wkhuh is
qr ssaWiaO ghshqghqfh iq {v_i j, i.h., y_it aqg Vj₈ auh iqghShqghqW iru aOO l ' j aqg aOO W> v( (iii)
Wkh iqgiyigxaO aqg Wimh hffhfWs, μ_i aqg λ_t> faq eh sWrfkasWif (uaqgrm hffhfWs) ru ghWhumiqisWif
({hg hhfWs).

Qr ghshqghqfh assxmsWirqs rq {μ_i j aqg {λ_t j auh imsrshg, ehfaxsh Wkh| ziOO eh giffhuo
hqfhg rxW iq Wkh frqsWuxfWirq ri rxu hsWimaWrus. Zh Wkxs aOOrz {λ_tj Wr eh shuiaOO∣ fruuhOaWhg

<<   3   4   5   6   7   8   9   >>

More intriguing information