This action might not be possible to undo. Are you sure you want to continue?

BooksAudiobooksComicsSheet Music### Categories

### Categories

### Categories

Editors' Picks Books

Hand-picked favorites from

our editors

our editors

Editors' Picks Audiobooks

Hand-picked favorites from

our editors

our editors

Editors' Picks Comics

Hand-picked favorites from

our editors

our editors

Editors' Picks Sheet Music

Hand-picked favorites from

our editors

our editors

Top Books

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Audiobooks

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Comics

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Sheet Music

What's trending, bestsellers,

award-winners & more

award-winners & more

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

Department of Economics Fall 2001 TA Roberto Perrelli

**Introduction to ARCH & GARCH models
**

Recent developments in ﬁnancial econometrics suggest the use of nonlinear time series structures to model the attitude of investors toward risk and expected return. For example, Bera and Higgins (1993, p.315) remarked that “a major contribution of the ARCH literature is the ﬁnding that apparent changes in the volatility of economic time series may be predictable and result from a speciﬁc type of nonlinear dependence rather than exogenous structural changes in variables.” Campbell, Lo, and MacKinlay (1997, p.481) argued that “it is both logically inconsistent and statistically ineﬃcient to use volatility measures that are based on the assumption of constant volatility over some period when the resulting series moves through time.” In the case of ﬁnancial data, for example, large and small errors tend to occur in clusters, i.e., large returns are followed by more large returns, and small returns by more small returns. This suggests that returns are serially correlated. When dealing with nonlinearities, Campbell, Lo, and MacKinlay (1997) make the distinction between: • Linear Time Series: shocks are assumed to be uncorrelated but not necessarily identically independent distributed (iid). • Nonlinear Time Series: shocks are assumed to be iid, but there is a nonlinear function relating the observed time series {Xt }∞ t=0 and the ∞ underlying shocks, {εt }t=0 .

1

.) + εt h(εt−1 .. ... xt−1 . εt−2 . εt−2 ..)|Ψt−1 }2 ] = V ar[εt h(εt−1 . most of the time series studies concentrate in one form or another.. xt−2 . εt−2 . .) E [Xt |Ψt−1 ] = g (εt−1 .They suggest the following structure to describe a nonlinear process: Xt = g (εt−1 . According to the authors.. it is nonlinear in mean but linear in variance.)}2 (1) where the function g (·) corresponds to the conditional mean of Xt .. Bera and Higgins (1993) explained that “the ARCH model characterizes the distribution of the stochastic error εt conditional on the realized values of the set of variables Ψt−1 = {yt−1 . Given such motivations. 2 . εt−2 .) V ar[Xt |Ψt−1 ] = E [{(Xt − E [Xt ])|Ψt−1 }2 ] = E [{εt h(εt−1 . and the function h(·) is the coeﬃcient of proportionality between the innovation in X t and the shock εt .. . • Engle’s (1982) ARCH Model: Xt = εt αε2 t−1 .. .)|Ψt−1 ] = {h(εt−1 . yt−2 . Engle (1982) proposed the following model to capture serial correlation in volatility: 2 σ 2 = ω + α(L)ηt (2) 2 where α(L) is the polynomial lag operator. εt−2 . The process is nonlinear in variance but linear in mean. .. The function g (·) = 0 and the function h = αε2 t−1 . εt−2 .}. they mention • Nonlinear Moving Average Model: Xt = εt + αε2 t−1 . σt −1 ) is the innovation in the asset return. .. Here the function 2 g = αεt−1 and the function h = 1. and ηt |Ψt−1 ∼ N (0. Thus. .. The general form above leads to a natural division in Nonlinear Time Series literature in two branches: • Models Nonlinear in Mean: g (·) is nonlinear.. As examples. • Models Nonlinear in Variance: h(·)2 is nonlinear.

q) has a polynomial β (L) of order “p” .. Let Y be a scalar random variable. Without loss of generality. Johnston and DiNardo (1997) brieﬂy mention the following properties of ARCH models: • ut have mean zero. and a polynomial α(L) of order “q” . To facilitate such computation. Properties and Interpretations of ARCH Models Following Bera and Higgins (1993). 2 2 2 σt = ω + β (L)σt −1 + α(L)ηt (3) It is quite obvious the similar structure of Autorregressive Moving Average (ARMA) and GARCH processes: a GARCH (p. Then. two important concepts should be introduced at this point: Deﬁnition 1 (Law of Iterated Expectations): Let Ω1 and Ω2 be two sets of random variables such that Ω1 ⊆ Ω2 .Computational problems may arise when the polynomial presents a high order. let a ARCH (1) process be represented by ut = εt α0 + α1 u2 t−1 (4) where {εt }∞ t=0 is a white noise stochastic process. then E [E [Y |Ω2 ]] = E [Y ]..the autorregressive term.the moving average term. E [Y |Ω1 ] = E [E [Y |Ω2 ]|Ω1 ]. Note (Conditionality versus Inconditionality): If Ω1 = ∅. Bollerslev (1986) proposed a Generalized Autorregressive Conditional Heteroskedasticity (GARCH) model.) = 0 3 . Proof: ut Et−1 [ut ] Et−2 Et−1 [ut ] E [ut ] = = εt α0 + α1 u2 t−1 Et−1 [εt ] α0 + α1 u2 t−1 (5) = 0 = 0 (.

Bera and Higgins (1993) mention the following reasons for the ARCH success: 4 . Besides that. Proof: 2 u2 = ε2 t t [α0 + α1 ut−1 ] 2 2 Et−1 [ut ] = σε [α0 + α1 u2 t−1 ] = 1[α0 + α1 u2 t−1 ] 2 = σt (6) • ut have unconditional variance given by σ 2 = Proof: Et−2 Et−1 [u2 t] α0 .2 • ut have conditional variance given by σt = α0 + α1 u2 t−1 . • ut have zero-autocovariances. and hence the ARCH models are so popular in this ﬁeld. Proof: Et−1 [ut ut−1 ] = ut−1 Et−1 [ut ] = 0 (8) Regarding kurtosis. given that 2 E [ε 4 1 − α1 t] = 3( )>3 2 4 σε 1 − 3α1 (9) Heavy tails are a common aspect of ﬁnancial data... + α1 ) + α1 u0 α0 = 1−α1 = σ2 (7) Therefore..) t−1 2 t 2 E0 E1 E2 (.. unconditionally the process is Homoskedastic. 1− α 1 = Et−2 [α0 + α1 u2 t−1 ] = α0 + α1 Et−2 [u2 t−1 ] 2 2 = α0 + α0 α1 + α1 ut−2 2 2 = Et−3 [α0 + α0 α1 + α1 ut−2 ] 2 = α0 + α0 α1 + α1 Et−3 [u2 t−2 ] 2 3 = α0 + α0 α1 + α0 α1 + α1 ut−3 Et−3 Et−2 Et−1 [u2 t] (. Bera and Higgins (1993) show that the process has a heavier tail than the Normal distribution...)Et−2 Et−1 [u2 t ] = α0 (1 + α1 + α1 + .

• Mizrach (1990). In the history of ARCH literature. They use the daily trading volume of stock markets as a proxy for such information arrival. which is somewhat related with the old-fashioned “adaptable expectations hypothesis” in macroeconomics. His interpretation may be summarized by the argument that “any economic variable.• ARCH models are simple and easy to handle • ARCH models take care of clustered errors • ARCH models take care of nonlinearities • ARCH models take care of changes in the econometrician’s ability to forecast In fact.g. He associates ARCH models with the errors of the economic agents’ learning processes.: • Lamoureux and Lastrapes(1990). And this inappropriate use of a calendar time scale may lead to volatility clustering since relative to the calendar time. 1986]. in general. Diebold. contemporaneous errors in expectations are linked with past errors in the same expectations. 1990. The basic menu (step-by-step) is: • Regress y on x by OLS and obtain the residuals {εt }. p. interesting interpretations of process can be found. • Stock (1998). 5 . while in practice it is measured on a ‘calendar’ time scale. evolves an on ‘operational’ time scale. and conﬁrm its signiﬁcance. Estimating and Testing ARCH Models Johnston and DiNardo (1997) suggest a very simple test for the presence of ARCH problems. 329. the last aspect was pointed by Engle (1982) as a “random coeﬃcients” problem: the power of forecast changes from one period to another. They mention that the conditional heteroskedasticity may be caused by a time dependence in the rate of information arrival to the market. In this case. the variable may evolve more quickly or slowly” (Bera and Higgins. E.

a straight-forward method of estimation (correction) is provided by Greene (1997). + α t−p + error . where dβ is the least squares coeﬃcient vector in the regression [ εt st ] = wx ˆ t rt + error rt (12) ˜ is given by (w The asymptotic covariance matrix for β ˆw ˆ )−1 . then compute ˜1 εt 1 rt = [ f + 2( α )] ft+1 t 2 1/2 (11) st = 1 ft − 2 α ˜1 εt+1 [ ft+1 ft+1 − 1] → ˜ = − Compute the estimate β β + dβ . α ˆp . • Test the joint signiﬁcance of α ˆ1 .. 6 . α ˆ1 ] = − α. Denote [α ˆ0 . where z ˆ is the regressor vector in this regression. → using the whole sample (T).• Compute the OLS regression ε2 ˆ0 + α ˆ1 ε2 ˆp ε2 t = α t−1 + . where dα is the least squares coeﬃcient vector in the regression [( ε2 ε2 1 t−1 t ) − 1] = z ˆ ˆ ) + error 0( ) + z 1( ft ft ft (10) The asymptotic covariance matrix for α ˜ is 2(ˆ zz ˆ)−1 . . In case that any of the coeﬃcients are signiﬁcant. • Recompute ft using α ˜ . • Regress y on x using least squares to obtain β 2 • Regress ε2 t on a constant and εt−1 to obtain the estimates of α0 and α1 . Then compute the asymptotically eﬃ→ − cient estimate α ˆ... where w ˆ is the regressor vector on the equation above.. • Compute ft = α ˆ0 + α ˆ1 ε2 t−1 . α ˜ = α + dα .. It consists in a four-step FGLS: ˆ and ε vectors.

Department of Finance. 307-366.. [6] Greene. and DiNardo. (1997). and MacKinlay. (1986). (1997). No. 77-85. Princeton. G. “Estimating Continuous-Time Processes Subject to Time Deformation. T. W. 51-56. D. (1990). 7 .. University of Pennsylvania. “ARCH Models: Properties.” Journal of Economic Surveys. 31. Econometric Methods. A.. The Warthon School.. Econometric Analysis. [9] Mizrach. C. New Jersey: Princeton University Press. [2] Bollerslev. W.” Econometric Reviews. (1988). J. [7] Johnston. X. (1990). J. 5.” Journal of the American Statistical Association (JASA). 4. 307-327. (1993). “Modelling the persistence of Conditional Variances: A Comment. “Learning and Conditional Heteroskedasticity in Asset Returns. New Jersey: Prentice-Hall.” Journal of Finance. Fourth Edition. [5] Engle.” Mimeo. (1986). “Generalized Autorregressive Conditional Heteroskedasticity. A. “Heteroskedasticity in Stock Return Data: Volume versus GARCH Eﬀects. L. J.” Journal of Econometrics. 221-229.. J. 50. Y. and Higgins. A. 83. Lo. Econometrica. 7. 987-1008. F. and Lastrapes. New York: McGraw-Hill. W. B. [10] Stock. C. R. [4] Diebold.“Autorregressive Conditional Heteroskedasticity with Estimates of United Kingdom Inﬂation”. [3] Campbell. Estimation and Testing. (1982).References [1] Bera. Third Edition. [8] Lamoureux. The Econometrics of Financial Markets. K. 45. Vol. (1997). M. H.

ARCH Model Econometrics

ARCH Model Econometrics

A Computationally Efficient Algorithm to Solve Generalized Method of Moments Estimating Equations Based on Secant Procedure

by AI Coordinator - CSC Journals

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue listening from where you left off, or restart the preview.

scribd