6

Estimating ARCH Models by Least Squares

The simplest estimation method for ARCH models is that of ordinary least squares (OLS). This estimation procedure has the advantage of being numerically simple, but has two drawbacks: (i) the OLS estimator is not efficient and is outperformed by methods based on the likelihood or on the quasi-likelihood that will be presented in the next chapters; (ii) in order to provide asymptotically normal estimators, the method requires moments of order 8 for the observed process. An extension of the OLS method, the feasible generalized least squares (FGLS) method, suppresses the first drawback and attenuates the second by providing estimators that are asymptotically as accurate as the quasi-maximum likelihood under the assumption that moments of order 4 exist. Note that the least-squares methods are of interest in practice because they provide initial estimators for the optimization procedure that is used in the quasi-maximum likelihood method.

We begin with the unconstrained OLS and FGLS estimators. Then, in Section 6.3, we will see how to take into account positivity constraints on the parameters.

6.1 Estimation of ARCH(q) models by Ordinary Least Squares

In this section, we consider the OLS estimator of the ARCH(q) model:

(6.1) c06e001_fmt

The OLS method uses the AR representation on the squares of the observed process. No assumption is made on the law of ηt.

The true value of the vector of the parameters is denoted by θ0 = (ω0, α01, …, α0q)′ and we denote by θ a generic value of the parameter.

From (6.1) we obtain the AR(q) representation

(6.2) c06e002_fmt

where ut = 2208-2t_fmtsigma-2t_fmt = (small-eta-2t_fmt − 1)sigma-2t_fmt. The sequence (ut, 2131_fmtt)t constitutes a martingale difference when E2208-2t_fmt = sigma-2t_fmt < ∞, denoting by 2131_fmtt the σ-field generated by {2208_fmts : st}.

Assume that we observe 2208_fmt1, …, 2208_fmtn, a realization of length n of the process (2208_fmtt), and let 2208_fmt0, …, 2208_fmt1 − q be initial values. For instance, the initial values can be chosen equal to zero. Introducing the vector

c06ue001_fmt

in view of (6.2) we obtain the system

(6.3) c06e003_fmt

which can be written as

c06ue002_fmt

with the n × q matrix

c06ue003_fmt

and the n × 1 vectors

c06ue004_fmt

Assume that the matrix X′X is invertible, or equivalently that X has full column rank (we will see that this is always the case asymptotically, and thus for n large enough). The OLS estimator of ¸0 follows:

(6.4) c06e004_fmt

Under assumptions OLS1 and OLS2 below the variance of ut, exists and is constant. The OLS estimator of sigma-20_fmt = Var(ut) is

c06ue005_fmt

Remark 6.1 (OLS estimator of a GARCH model) An OLS estimator can also be defined for a GARCH (p, q) model, but the estimator is not explicit, because 2208-2t_fmt does not satisfy an AR model when p ≠ 0 (see Exercise 7.5).

To establish the consistency of the OLS estimators of θ0 and sigma-20_fmt , we must consider the following assumptions.

OLS1: (2208_fmtt) is the nonanticipative strictly stationary solution of model (6.1), and ω0 > 0.

OLS2: E2208-4t_fmt < +∞

OLS3: 2119_fmt(small-eta-2t_fmt = 1) ≠ 1.

Explicit conditions for assumptions OLS1 and OLS2 were given in Chapter 2. Assumption OLS3 that the law of ηt is nondegenerate allows us to identify the parameters. The assumption also guarantees the invertibility of X′X for n large enough.

Theorem 6.1 (Consistency of the OLS estimator of an ARCH model) Let (carret-theta_fmtn) be a sequence of estimators satisfying (6.4). Under assumptions OLS1–OLS3, almost surely

c06ue006_fmt

Proof. The proof consists of several steps.

(i) We have seen (Theorem 2.4) that (2208_fmtt), the unique nonanticipative stationary solution of the model, is ergodic. The process (Zt) is also ergodic because Zt is a measurable function of {2208_fmtt − t, i ≥ 0}. The ergodic theorem (see Theorem A.2) then entails that

(6.5) c06e005_fmt

The existence of the expectation is guaranteed by assumption OLS3. Note that the initial values are involved only in a fixed number of terms of the sum, and thus they do not matter for the asymptotic result. Similarly, we have

c06ue007_fmt

(ii) The invertibility of the matrix EZt − 1 Zt − 1 = EZtZt is shown by contradiction. Assume that there exists a nonzero vector c of 211D_fmtq+1 such that cEZt Ztc = 0. Thus E{c′Zt}2 = 0, and it follows that cZt = 0 a.s. Therefore, there exists a linear combination of the variables c06-ie129001_fmt which is a.s. equal to a constant. Without loss of generality, one can assume that, in this linear combination, the coefficient of 2208-2t_fmt = small-eta-2t_fmtsigma-2t_fmt is 1. Thus small-eta-2t_fmt is a.s. a measurable function of the variables 2208_fmtt − 1, …, 2208_fmtt−q. However, the solution being nonanticipative, small-eta-2t_fmt is independent of these variables. This implies that small-eta-2t_fmt is a.s. equal to a constant. This constant is necessarily equal to 1, but this leads to a contradiction with OLS3. Thus E(Zt − 1 Zt − 1) is invertible.

(iii) The innovation of 2208-2t_fmt being c06-ie129002_fmt, we have the orthogonality relations

c06ue008_fmt

that is

c06ue009_fmt

(iv) Point (ii) shows that n−1 X′X is a.s. invertible, for n large enough and that, almost surely, as n → ∞,

c06ue010_fmt

For the asymptotic normality of the OLS estimator, we need the following additional assumption.

OLS4: c06-ie130001_fmt.

Consider the (q + 1) × (q + 1) matrices

c06ue011_fmt

The invertibility of A was established in the proof of Theorem 6.1, and the invertibility of B is shown by the same argument, noting that c06-ie130002_fmt if and only if c′Zt − 1 = 0 because sigma-2t_fmt > 0 a.s. The following result establishes the asymptotic normality of the OLS estimator.

Let c06-ie130003_fmt

Theorem 6.2 (Asymptotic normality of the OLS estimator) Under assumptions OLS1–OLS4,

c06ue012_fmt

Proof. In view of (6.3), we have

c06ue013_fmt

Thus

(6.6) c06e006_fmt

Let λ 2208_fmt 211D_fmtq + 1, λ ≠ 0. The sequence (λ′Zt − 1ut, 2131_fmtt) is a square integrable ergodic stationary martingale difference, with variance

c06ue014_fmt

By the CLT (see Corollary A.1) we obtain that, for all λ ≠ 0,

c06ue015_fmt

Using the Cramér-Wold device, it follows that

(6.7) c06e007_fmt

The conclusion follows from (6.5), (6.6) and (6.7).       Box_fmt

Remark 6.2 (Estimation of the Information matrices) Consistent estimators A-carret_fmt and B-carret_fmt of the matrices A and B are obtained by replacing the theoretical moments by their empirical counterparts,

c06ue016_fmt

where c06-ie131001_fmt. The fourth order moment of the process ηt = 2208_fmttt is also consistently estimated by c06-ie131002_fmt. Finally, a consistent estimator of the asymptotic variance of the OLS estimator is defined by

c06ue017_fmt

Example 6.1 (ARCH(1)) When q = 1 the moment conditions OLS2 and OLS4 take the form c06-ie131003_fmt and c06-ie131004_fmt (see (2.54)). We have

c06ue018_fmt

with

c06ue019_fmt

The other terms of the matrix B are obtained by expanding c06-ie131005_fmt and calculating the moments of order 6 and 8 of 2208-2t_fmt.

Table 6.1 shows, for different laws of the iid process, that the moment conditions OLS2 and OLS4 impose strong constraints on the parameter space.

Table 6.2 displays numerical values of the asymptotic variance, for different values of α01 and ω0 = 1, when ηt follows the normal N(0, 1).

The asymptotic accuracy of carret-theta_fmtn becomes very low near the boundary of the domain of existence of c06-ie131006_fmt. The OLS method can, however, be used for higher values of α01, because the estimator remains consistent when α01 < 3−1/2 = 0.577, and thus can provide initial values for an algorithm maximizing the likelihood.

Table 6.1 Strict stationarity and moment conditions for the ARCH(l) model when ηt follows the N(0, 1) distribution or the Student t distribution (normalized in such a way that Esmall-eta-2t_fmt = 1).

c06t001_fmt

‘no’ means that the moment condition is not satisfied.

Table 6.2 Asymptotic variance of the OLS estimator of an ARCH(l) model with ω0 = 1, when ηt ~ N(0, 1).

c06t002_fmt

6.2 Estimation of ARCH(q) Models by Feasible Generalized Least Squares

In a linear regression model when, conditionally on the exogenous variables, the errors are heteroscedastic, the FGLS estimator is asymptotically more accurate than the OLS estimator. Note that in (6.3) the errors ut are, conditionally on Zt−1, heteroscedastic with conditional variance Var(ut | Zt−1) = (κη−1)sigma-4t_fmt.

For all θ = (ω, α1, …, αq)′ let

c06ue020_fmt

The FGLS estimator is defined by

c06ue021_fmt

Theorem 63 (Asymptotic properties of the FGLS estimator) Under assumptions OLS1 – OLS3 and if α0i > 0, i = 1, …, q,

c06ue022_fmt

where c06-ie132001_fmt is positive definite.

Proof. It can be shown that J is positive definite by the argument used in Theorem 6.1.

We have

(6.8) c06e008_fmt

A Taylor expansion around θ0 yields, with sigma-2t_fmt = sigma-2t_fmt0),

(6.9) c06e009_fmt

where θ* is between carret-theta_fmtn and θ0. Note that, for all θ, c06-ie132002_fmt. It follows that

c06ue023_fmt

The first term on the right-hand side of the equality converges a.s. to J by the ergodic theorem. The second term converges a.s. to 0 because the OLS estimator is consistent and

c06ue024_fmt

for n large enough. The constant bound K is obtained by arguing that the components of carret-theta_fmtn, and thus those of θ*, are strictly positive for n large enough (because carret-theta_fmtn → θ0 a.s.). Thus, we have c06-ie133001_fmt, for i = 1, …, q, and finally c06-ie133002_fmt is bounded. We have shown that a.s.

(6.10) c06e010_fmt

For the term in braces in (6.8) we have

(6.11) c06e011_fmt

by the previous arguments, noting that c06-ie133003_fmt and

c06ue025_fmt

Thus, we have shown that c06-ie133004_fmt, a.s.

Using (6.11), (6.8) and (6.10), we have

c06ue026_fmt

where Rn → 0, a.s. A new expansion around θ0 gives

(6.12) c06e012_fmt

where θ** is between θ* and θ0. It follows that

(6.13) c06e013_fmt

The CLT applied to the ergodie and square integrable stationary martingale difference c06-ie134001_fmt shows that Sn converges in distribution to a Gaussian vector with zero mean and variance

c06ue027_fmt

(see Corollary A.l). Moreover,

c06ue028_fmt

The two terms in braces tend to 0 a.s. by the ergodic theorem. Moreover, the terms c06-ie134002_fmt and c06-ie134003_fmt are bounded in probability, as well as J− 1 + Rn. It follows that Sn2 tends to 0 in probability. Finally, by arguments already used and because θ* is between carret-theta_fmtn and θ0

c06ue029_fmt

in probability. Using (6.12), we have shown the convergence in law of the theorem.     Box_fmt

The moment condition required for the asymptotic normality of the FGLS estimator is E(2208-4t_fmt) < ∞. For the OLS estimator we had the more restrictive condition c06-ie134004_fmt. Moreover, when this eighth-order moment exists, the following result shows that the OLS estimator is asymptotically less accurate than the FGLS estimator.

Theorem 6.4 (Asymptotic OLS versus FGLS variances) Under assumptions 0LS1–0LS4, the matrix

c06ue030_fmt

is positive semi-definite.

Proof. Let c06-ie135001_fmt. Then

c06ue031_fmt

is positive semi-definite, and the result follows.         Box_fmt

We will see In Chapter 7 that the asymptotic variance of the FGLS estimator coincides with that of the quasi-maximum likelihood estimator (but the asymptotic normality of the latter is obtained without moment conditions). This result explains why quasi-maximum likelihood is preferred to OLS (and even to FGLS) for the estimation of ARCH (and GARCH) models. Note, however, that the OLS estimator often provides a good initial value for the optimization algorithm required for the quasi-maximum likelihood method.

6.3 Estimation by Constrained Ordinary Least Squares

Negative components are not precluded in the OLS estimator carret-theta_fmtn defined by (6.4) (see Exercise 6.3). When the estimate has negative components, predictions of the volatility can be negative. In order to avoid this problem, we consider the constrained OLS estimator defined by

c06ue032_fmt

The existence of c06-ie135002_fmt is guaranteed by the continuity of the function Qn and the fact that

c06ue033_fmt

as 2016_fmtθ2016_fmt → ∞ and θ ≥ 0, whenever X has nonzero columns. Note that the latter condition is satisfied at least for n large enough (see Exercise 6.5).

6.3.1 Properties of the Constrained OLS Estimator

The following theorem gives a condition for equality between the constrained and unconstrained estimators. The theorem is stated in the ARCH case but is true In a much more general framework.

Theorem 6.5 (Equality between constrained and unconstrained OLS) If X is of rank q + 1, the constrained and unconstrained estimators coincide, c06-ie135003_fmt, if and only if c06-ie135004_fmt.

Proof. Since carret-theta_fmtn and c06-ie135002_fmt are obtained by minimizing the same function Qn(·), and since c06-ie135002_fmt minimizes this function on a smaller set, we have c06-ie135005_fmt Moreover, c06-ie135006_fmt, and we have c06-ie135007_fmt, for all θ 2208_fmt [0, + ∞)q + 1.

Suppose that the unconstrained estimation carret-theta_fmtn belongs to [0, + ∞)q + 1. In this case Qn(carret-theta_fmtn)= Qn(c06-ie135002_fmt) Because the unconstrained solution is unique, c06-ie135002_fmt = carret-theta_fmtn.

The converse Is trivial.       Box_fmt

We now give a way to obtain the constrained estimator from the unconstrained estimator.

Theorem 6.6 (Constrained OLS as a projection of OLS) If X has rank q + 1, the constrained estimator c06-ie135002_fmt is the orthogonal projection of carret-theta_fmtn on [0, + ∞)q + 1 with respect to the metric X′X, that is,

(6.14) c06e014_fmt

Proof. If we denote by P the orthogonal projector on the columns of X, and M = InP,

we have

c06ue034_fmt

using properties of projections, Pythagoras’s theorem and PY = Xcarret-theta_fmtn. The constrained estimation c06-ie135002_fmt thus solves (6.14). Note that, since X has full column rank, a norm is well defined by c06-ie136001_fmt. The characterization (6.14) is equivalent to

(6.15) c06e015_fmt

Since [0, +∞)q+1 is convex, c06-ie135002_fmt exists, is unique and Is the X′X-orthogonal projection of carret-theta_fmtn on [0, + ∞)q+1. This projection is characterized by

(6.16) c06e016_fmt

(see Exercise 6.9). This characterization shows that, when c06-ie136002_fmt, the constrained estimation c06-ie135002_fmt must lie at the boundary of [0, + ∞)q+1. Otherwise it suffices to take θ 2208_fmt [0, + ∞)q+1 between c06-ie135002_fmt and carret-theta_fmtn to obtain a scalar product equal to − 1.       Box_fmt

The characterization (6.15) allows us to easily obtain the strong consistency of the constrained estimator.

Theorem 6.7 (Consistency of the constrained OLS estimator) Under the assumptions of Theorem 6.1, almost surely,

c06ue035_fmt

Proof. Since θ0 2208_fmt [0, +∞)q+1, in view of (6.15) we have

c06ue036_fmt

It follows that, using the triangle inequality,

c06ue037_fmt

Since, in view of Theorem 6.1, carret-theta_fmtn → θ0 a.s. and X′X/n converges a.s. to a positive definite matrix, It follows that c06-ie136003_fmt and thus that c06-ie136004_fmt a.s. Using Exercise 6.12, the conclusion follows.       Box_fmt

6.3.2 Computation of the Constrained OLS Estimator

We now give an explicit way to obtain the constrained estimator. We have already seen that if all the components of the unconstrained estimator θn are positive, we have c06-ie135002_fmt = carret-theta_fmtn. Now suppose that one component of carret-theta_fmtn is negative, for instance the last one. Let

c06ue038_fmt

and

c06ue039_fmt

Note that c06-ie137001_fmt in general (see Exercise 6.11).

Theorem 6.8 (Explicit form of the constrained estimator) Assume that X has rank q + 1 and carret-alpha_fmtq < 0. Then

c06ue040_fmt

Proof. Let c06-ie137002_fmt be the projector on the columns of X(1) and let M(1) = I − P. We have

c06ue041_fmt

Because c06-ie137003_fmt, with eq+1 = (0, …, 0, 1)′, we have c06-ie137004_fmt. This can be written as

c06ue042_fmt

or alternatively

c06ue043_fmt

Thus Y′M(1)X(2) < 0. It follows that for all θ = (θ(1)′, θ(2))′ such that θ(2) 2208_fmt [0, ∞),

c06ue044_fmt

In view of (6.16), we have c06-ie135002_fmt= carret-theta_fmtn because carret-theta_fmtn 2208_fmt [0, +∞)q+1.    Box_fmt

6.4 Bibliographical Notes

The OLS method was proposed by Engle (1982) for ARCH models. The asymptotic properties of the OLS estimator were established by Weiss (1984, 1986), in the ARMA-GARCH framework, under eighth-order moments assumptions. Pantula (1989) also studied the asymptotic properties of the OLS method in the AR(l)-ARCH(q) case, and he gave an explicit form for the asymptotic variance. The FGLS method was developed, in the ARCH case, by Bose and Mukherjee (2003) (see also Gouriéroux, 1997). The convexity results used for the study of the constrained estimator can be found, for instance, in Moulin and Fogelman-Soulié (1979).

6.5 Exercises

6.1 (Estimating the ARCH(q) for q = 1, 2, …)

Describe how to use the Durbin algorithm (B.7)–(B.9) to estimate an ARCH(q) model by OLS.

6.2 (Explicit expression for the OLS estimator of an ARCH process)

With the notation of Section 6.1, show that, when X has rank q, the estimator carret-theta_fmt = (X′X)−1 X′Y is the unique solution of the minimization problem

c06ue045_fmt

6.3 (OLS estimator with negative values)

Give a numerical example (with, for instance, n = 2) showing that the unconstrained OLS estimator of the ARCH(q) parameters (with, for instance, q = 1) can take negative values.

6.4 (Unconstrained and constrained OLS estimator of an ARCH(2) process)

Consider the ARCH(2) model

c06ue046_fmt

Let c06-ie138001_fmt be the unconstrained OLS estimator of θ = (ω, α1, α2)′. Is it possible to have

1. carret-alpha_fmt1 < 0?

2. carret-alpha_fmt1 < 0 and carret-alpha_fmt2 < 0?

3. carret-omega_fmt < 0, carret-alpha_fmt1 < 0 and carret-alpha_fmt2 < 0?

Let c06-ie138002_fmt be the OLS constrained estimator with c06-ie138003_fmt and c06-ie138004_fmt. Consider the following numerical example with n = 3 observations and two initial values: c06-ie138005_fmt, c06-ie138006_fmt. Compute carret-theta_fmt and carret-theta_fmtC for these observations.

6.5 (The columns of the matrix X are nonzero)

Show that if ω0 > 0, the matrix X cannot have a column equal to zero for n large enough.

6.6 (Estimating an AR(l) with ARCH(q) errors)

Consider the model

c06ue047_fmt

where (2208_fmtt) is the strictly stationary solution of model (6.1) under the condition c06-ie139001_fmt. Show that the OLS estimator of φ is consistent and asymptotically normal. Is the assumption c06-ie139001_fmt necessary in the case of iid errors?

6.7 (Inversion of a block matrix)

For a matrix partitioned as c06-ie139002_fmt, show that the inverse (when it exists) is of the form

c06ue048_fmt

where c06-ie139003_fmt

6.8 (Does the OLS asymptotic variance depend on ω0?)

1. Show that for an ARCH(q) model c06-ie139004_fmt is proportional to c06-ie139005_fmt (when it exists).

2. Using Exercise 6.7, show that, for an ARCE(q) model, the asymptotic variance of the OLS estimator of the α0i does not depend on ω0.

3. Show that the asymptotic variance of the OLS estimator of ω0 is proportional to c06-ie139006_fmt.

6.9 (Properties of the projections on closed convex sets)

Let E be an Hilbert space, with a scalar product 2329_fmt·, ·232A_fmt and a norm 2016_fmt · 2016_fmt. When C 2282_fmt E and x 2208_fmt E,it is said that x* 2208_fmt C is a best approximation of x on C if 2016_fmtxx* 2016_fmt = miny2208_fmtc 2016_fmtxy 2016_fmt.

1. Show that if C is closed and convex, x* exists and is unique. This point is then called the projection of x on C.

2. Show that x* satisfies the so-called variational inequalities:

(6.17) c06e017_fmt

and prove that x* is the unique point of C satisfying these inequalities.

6.10 (Properties of the projections on closed convex cones)

Recall that a subset K of the vectorial space E is a cone if, for all x 2208_fmt K, and for all λ ≥ 0, we have λx 2208_fmt K. Let K be a closed convex cone of the Hilbert space E.

1. Show that the projection x* of x on K (see Exercise 6.9) is characterized by

(6.18) c06e018_fmt

2. Show that x* satisfies

   (a) 2200_fmtx 2208_fmt E, 2200_fmtλ ≥ 0, (λx)* = λx*.

   (b) 2200_fmtx 2208_fmt E, 2016_fmtx2016_fmt2 = 2016_fmtx*2016_fmt2 + 2016_fmtxx*2016_fmt2, thus 2016_fmtx*2016_fmt2016_fmtx2016_fmt.

6.11 (OLS estimation of a subvector of parameters)

Consider the linear model Y = Xθ + U with the usual assumptions. Let M2 be the matrix of the orthogonal projection on the orthogonal subspace of X(2), where X = (X(1), X(2)). Show that the OLS estimator of θ(1) (where θ = (θ(l)′, θ(2)′)′, with obvious notation) is c06-ie139007_fmt

6.12 (A matrix result used in the proof of Theorem 6.7)

Let (Jn) be a sequence of symmetric k × k matrices converging to a positive definite matrix J. Let (Xn) be a sequence of vectors in 211D_fmtk such that XnJnXn → 0. Show that Xn → 0.

6.13 (Example of constrained estimator calculus)

Take the example of Exercise 6.3 and compute the constrained estimator.

..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset
3.147.48.78