Ϣ
Information and Control
1999ꡡ285ڡVol.28No.51999



ARCHģ͵о̽

硡ʵ˵»

ժҪ Իع췽(ARCH)ģǽ·չʱģͣӳ ̵һԣʱ仯仯ҾдԼԡԣARCHģѹ㷺 ӦھĽģоУĽARCHģ͵ص㣬Ĳƺͼ ԼARCHģ͵ķչ
ؼʣARCH,췽,Լ,
ͼţTP13ױʶ룺B

RESEARCH ON ARCH MODEL
MIAO Shi LIU Hai-long PAN De-hui
(FBA of the Northeastern University, Shenyang City 110006)
Abstract: Autoregressive Conditional Heteroscedastic model is the newly developed time series model, which reflects the special characteris tics of stochastic process:the variance changes with the time changing and the variance is crowd together and fluctuated. ARCH model has been widely ap plied in modeling and research of economic field, especially of financial ma rkets. This paper introduces the characteristics of ARCH model, it's estimation and test of parameter and its development in detail.
Key words: ARCH, heteroscedastic, crowd, fluctuation

1
ARCH ( Auto Regressive Conditional Heteroscedasticity)ģʥѧѧҶ(Engle R.F)1982ģҪھдԼԼ صľʱݵĻعԤ⣮ʵ֤ARCHģйطӦȡõЧ
ͳľüģͼٶķֲ䣮ž۵ķչʵ֤ ѷһ費ԽԽо߷֣ʱݣƱ۸ ͨʡʡʵȣַʱ仯ص㣮Mandelbrot(1963) ۲쵽ྭķֲźܿβ䷽Ҳڲϱ仯˺ мֵڷı仯УȽϴı仯ԵؼĳЩʱȽСı仯ԼһЩʱBera(1992)ԪӢÿ»ʣ޵ĶծȯԼŦԼƱۺָһ֤MandelbrotĽۣʱݵķױԼԼԣֱ仯ܹھǽгĶԣζҺߵԵȵȣ һʾ۲ķĳأEngleARCHģǡǡ ׽˾ʱݵص㣮
ARCHģѹ㷺ӦھķоУȡõЧEngl e(1982), EngleKraft(1983)ARCHģͷԤͨʵȡĽDomowitzHakkio(1985)ҲARCHģгоУWeiss(1984)13ֲͬĺ۾ʱARCHģȡ˳ɹĽ90ѧ߿ʼARCHģͽиƼչ磺GARCH (Generalized A uto Regressive Conditional Heteroscedasticity), NARCH (Non-linear Auto Regressiv e Conditional Heteroscedasticity)ģͣ
2ARCHģ͵
ARCHģ͵صǼ۲ݵķأ۲ķͺ ֵĺһԤֲ£
yt|t-1N(xt,ht)
У

ttʱ̵ϢϣxtһЩԤͺֵɵ ºͦǴֵΪ0Ϊhtֲ̬tN( 0, h1)EngleҪh(*)ĺʽΪµιϵ
ht=0+12t-1++p2t-p(1)
Ϊ˱֤ht0, ģҪ00, i0, i=1,2,,p
ʽУǿԿʽߵhtΪtʱķһǸʽұߵĦ2t-1,,2t-ptʱʱҲѲҲһЩ֪ķǸtʱʱϷʽ
ǿԴARCHģ췽htĻع鷽пʱ̵ķǹȥ ʱƽp׻ƽͽгеʱʮǺϣt-m ʱʱ(mp), гĳ仯ôƽͻӦ ϵiΪ0, ػᵼtʱ췽htҲ˵tʱ гܳ(t-m)ʱı仯УĴȲĳЩʱ ϣС һЩʱϣ
Ϊ˱֤ARCHģǷƽȵģһ跽̣
1-1z-2z2--pzp=0
ĸڵλԲ֮⣬ΪiǸԣ
1+2++p1
ôtķҲǶռϵĦ2iֵ


ռhi2iΪ֪ķǸɵõ
2=0+12++p2


,(1)ʽã

Ӵ˷пɿ췽htȫ򷽲2ֲ򷽳Ц2t-1, ,2t-pļȨƽ
3ARCHģ͵Ĺƺͼ
3.1 ARCHģ͵Ĺ
ARCHģ͵ĶȻΪ(ʡ˳)

ʽС=(¡, 0, 1, , p)
ӦȻۣõ㷨Եó

Ц(i)ʾiεȵĲֵiǿɱ䲽T
3.2 ARCHģ͵ļ
ճӼ鷨(LM)ARCHģͽм飺
裺H0:1=2==p=0
õLMͳ

У f0=(21h-11-12th-1t-1)

w=(0, 1,,p)
T.
ڸŶ£x2(p)ٽֵTR2ٽֵܾH0裬ʾiһΪ0ht=0+12t-1++p2t -phtʱı仯仯ģ
4ARCHģ͵ķչ
4.1 GARCH(Generalized Auto Regressive Conditional Heteroscedasticity)
GARCHģEngleѧBollerslev1986ģARCHеhtʽмԻع鲿֣
ht=0+12t-1++p2t-p+1ht-1++qht-q
00, a1,,p, 1,,q0
ʽGARCH(p,q)ʽʵӦGARCH(1,1)֪ԣ

GARCHhtѲپp׽βԣhtԹȥƽ޼ԣͽгеʱԸǺϣGARCHģͺͱ׼ʱģARARMAֳһԣԺܶҲARCH࣮
4.2 AGARCH (Asymmetric GARCH(1,1))
GARCHģʱȫ˵ĳЩݵƫ̬ͷʣütֲֲ̬ƫ̬ͨAGARCHģ͹htļ
ht=0+1(t-1-)2+1ht-100, 1,1, Ρ0
ģУǿԿt-10ʱŶΪʱhtŶΪʱȡø ֵԴģ͸ӦԽԽͻ̵󣬼νܸã
4.3 EGARCH (Exponential GARCH)
EGARCHģNelson(1991)һģͣ
loght=+g(zt-1)+loght-1
ʽ, ԿztN(0, 1)

EGACHļǷԣǶԳƵģҲҪǸ
4.4 IGARCH (Integrated GARCH)
1+1=1ʱдGARCH(1,1)ģΪ
ht=0+(1-)2t-1+ht-10ܦˡ1
ʽУGARCHģеķѲڣIGARCHǷƽȵGARCHģͣǿ԰дɣ

Կ0=0ʱȼָȨƽģͣұи㷺Ӧã
4.5 ARCH-MGARCH-M
ʽΪ
yi=xt+ht+t
tN(0, ht)
ǿԿģʱı仯ʱı仯 ĳЩúͽʱݵо
5
ʮARCHģھгеõԽԽ㷺Ӧã̽Ҳ𲽵룮ARCHģʵطӳĳЩݵص:ʱı仯仯
߼飺 ʵŮ26꣬ʿ. оΪʱģۼӦ.
У39꣬ʿ. оΪڿۼӦ.
˵»ݣУ71꣬ڣʿʦ. оΪֲϵͳÿ.
ߵλѧ̹ѧԺ110006
ο
1źȺ. гԤ⼰ߵߣARCHģ. ϵ ͳ,1991,15(1)
2Tim Bollerslev. Generalized Autoregressive Conditional Heteroskedasticity. E conometrics 1986,31:30727
3CAROL Alexander. Volatility and Correlation Forecasting. The Handbook of R isk Management and Analysis,1996
4ͯ. ûعģͼ. ѧ,1997
5ܸ߼ѫ. Ӧͳѧ. йѧ,1989
ոڣ1999-01-22
