GARCH模型建模.docx
《GARCH模型建模.docx》由会员分享,可在线阅读,更多相关《GARCH模型建模.docx(18页珍藏版)》请在冰豆网上搜索。
![GARCH模型建模.docx](https://file1.bdocx.com/fileroot1/2022-10/13/56f19c0e-6240-4286-88e5-489a0bc019c7/56f19c0e-6240-4286-88e5-489a0bc019c71.gif)
GARCH模型建模
以1995年1月至2000年8月日元兑美元汇率值序列ARCH建模分析洪1427个
值)日元兑美元汇率值序列JPY
1.画出时序图
^roup
EsTirrateEquatoru.
2.
=&tirrat&VAR...
JPY
150
做单位根检验
SdSeries:
PVWorkfilP:
ARCHCASFl2:
:
Ca^l2\
fvew|Proc|objec,Proptnies]
[print
NaFreeze
SsrnpleGenr
Grap^...
Descrip:
iveSiatis:
ics&This
One-Wdy"abuldtcn...
Ccrrelogram...
Leng-runVifiinoe...
UnitRootTest.,
VarianceRaiioTest..
CDSIndependenceTies:
.,.
Ubel
1IIIIIfIIIV.
NullHypothesis.JPYhasaunitroot
Exogenous:
Constant
L^gLength'0(Autornatic-basedonSICrraxlag^23)
t-StatisticProb.'
AuqmertedDickey-Fulle-rteststatistic
4851510.5411
Testcriticalvalues:
1%level
-3434727
5%level
-2663360
10%level
-25677S8
KL1acKirnon(1996)onesidedp-values
存在单位根,对差分后后数据做分析,差分后数据是否平稳
DJPY
>rocQuickOptionsAdd-insWindowHelpSample...
GenerateSeries*,.
e12.wt1)-□x_nx
石Enid5白rri口便山二”.[倉口巾门巧
HistagramandStatsCorrelograrri...
UnitRootTeSrt...
VarianceRatioTesti,.
ExponentialSmoothing,..
Hodrick-^rescottFilter...
t-Statistic
Prob*
>c12.wf1)
bs
>70137
atProt
E^inwmvMflgtsjiiip
dQpv)afjl)a<(7]srfJJ.
确定模型AR(3)MA(3)
Autd€iXt<uriFilialCurreldLun
Sample...
GenerateSeri&s...
EstimateEquation..
tquMiont'flfnsTiori
ztViewProc
ARCMCAS口2
tstimateVA^,.L
-{e:
\30
itJetm
Spe*EiDnOptborw
DependentvariablefollowedbylistofregressorsincludingARMAandPDLterms,ORanexplicitequationlike¥c(i)+cg)*X.
QuickIOptiorsAdd-insWindow
Show».
sa\eFreez
27-1427ol
27一1d?
7ot
EmptyGroup:
EditSeres)
ScriesS:
atis:
ic;
Graup^t^tisties
OrizName
Genr5arnple
LS-Squ«r#$;NLSAndARMA}
1
DD39
0039
22127
€or
2
0051
0050
&.92H
0.052
3
0094
OOSS
f5.&72
{J001
■1
fion
t62O7
0阿
=
DD11
Q019
00D6
6
-c-joe
-O.O2U
16.439
Li.011
7
-O.OOL
-G.0G3
1U.b3h
0.021
9
-0016
0012
16902
0031
Q
QO+?
0041
19473
0021
10
Q0O4
OOOf
19491
OOM
11
0003
O1S
195S6
C0&1
1?
fl価
"溜
?
01W
r
■flA
D3O?
nriifln
acni
nn町u
?
0116
nnnt*
009?
nino
ACFACClSlatPtcb
DbptinJwrtVdridbl^.D(JPY1
MethodLeastSquares
Date12OT17Time?
1245
Sampe514?
7
Indudccobservations:
1422afterixjjustmcnts
Ccnvergenccachev€dafter3liter^ions
Variable
CoofTcierrt
StdError
rStatistic
Prot
胆⑴
0042223
0026433
1597328
01104
AR
(2)
0C52554
0.026418
1939321
0W9
AR(3)
-0na7992
0f)^425
-332993
00009
R-squared
0011711
Meandependentvsr
0.003717
AdjustedN-squared
0010319
SDdependentvsr
0963127
S.L.ofregression
O.9S0145
Akaikeinfocntenoc
27b^/1
Sumsquaredresic
1303.620
ScriKtarzcritsnon
276b662
Loaiik€hhiccd
■labt.sut
H白nnarGyumrienter.
2.759814
DurbinWatson泅
199568E
liverledARRuuk
2S+35i
26*35i
*47
AR
(1)前面的系数显著为0,于是去掉ar
(1)
JependemVariableD(JP¥)
Method:
LeastSquares
Date:
12/08/17Time:
12:
45
Sample(adjusted):
5H27
ncludcdobscrvcitiDns:
1123afteradjustments
Con^rgcnccachievedafter3Horations
Variable
Coefficient
Std.Error
tStatistic
Prob
AR[2)
AR[3)
0064145
D085000
0.02&4H
0.026407
2040895
3262091
0.0406
00012
R-squarcd
R*squarcdS.EofregressionSumsquaredresidLogIkckhoodDurbin-Watscnstat
0000935
D009238
D95fi66G
1305963-195Q0G4
1.011506
MeandependentvarS.D.dependentvarAkaikcrfocriterionSchwarzcriterionHannan-Quinncritcr
0.003717
0.963127
2754061
2762255
2757623
nvertedARRoots
.24».35i
2<35
-■40
建立模型,参数显著性检验通过,残差是否是白噪声
^=]Equat'cn:
UNfrlLED^tortTile:
ARCH匚A5E12:
:
La蛇x
[vnwJprQcjotned[Mrt[NameFreen|
.可5n制三Rfsnj*i|
Reposeration:
1ofRttiduala
Estim呂ti口nOjtput
ActjaLFrtiedLResdual►
ARMAStrucLure^.
GradientsardDewarwk►
CuvaranceMatnx
A
AR^Aierms
hACPACQStatProb
Cesffie#rhlDwgi吋nsties►
10.044Q.(H42了儆)
nnzvt细n-hn-"7/ur
Re^idualDisgnostics►
CrTeogram-Qt佃廿5匸伍_
StabilityOagn^tics►
CorreigigrarnR鶴siciu尹Is*
Label
H^tngra^YiNcnr^lifyT^t
SerialCorrddUcnLMTfel...
Heteros■■0”*»1■:
J■J・A.1^IS
:
:
r
IncludedobsErvations.142i
O-stalis<]€probabilitiesari|usledfor2ARMAiemis
Auluc^ireliJlicinParballCorrddlianACPACQSldflPiob
AcUj日FrttEGResid^il
AHMAStructure.,,
结果表明是白噪声序列,继续检验残差平方是否是自回归模型
iSaveI
trTfEZe
Cieta
Show
Felrh1jtarE
CeletefsBririSamDie
-n
JbjEl
1427-AA77
1427-1427
3Equitfon;UNTHLEDWkfile:
ARCHCASE12Xase12\lh
”lEW
^TgcjobjEdt[Piirrt|Ndtne
|fc&towtE
FGiecjisfc|
■T
Reprefertatons$ofResiduals
EilimaticnOutput
ARFMterms
ktACPACQ-Stflimb
Date:
12/08/17Time:
12:
48
Sample:
11427
Includedobservations-1423
AutocorrelationPartialCorrelationACPACQ-StatProb
Zl
i
二
102530253916320.000
□
i