Infinite Fourth Moment Research Articles

Log determinant of large correlation matrices under infinite fourth moment

Log determinant of large correlation matrices under infinite fourth moment

Large sample autocovariance matrices of linear processes with heavy tails

We provide asymptotic theory for certain functions of the sample autocovariance matrices of a high-dimensional time series with infinite fourth moment. The time series exhibits linear dependence across the coordinates and through time. Assuming that the dimension increases with the sample size, we provide theory for the eigenvectors of the sample autocovariance matrices and find explicit approximations of a simple structure, whose finite sample quality is illustrated for simulated data. We also obtain the limits of the normalized eigenvalues of functions of the sample autocovariance matrices in terms of cluster Poisson point processes. In turn, we derive the distributional limits of the largest eigenvalues and functionals acting on them. In our proofs, we use large deviation techniques for heavy-tailed processes, point process techniques motivated by extreme value theory, and related continuous mapping arguments.

Wavelet variances for heavy‐tailed time series

AbstractThis paper focuses on wavelet analysis of variability for heavy‐tailed time series. Under the assumption that time‐series values have finite second but infinite fourth moments, stable asymptotics are derived for wavelet variances across different time scales. These stable asymptotics have a slower rate of convergence than the square root of the sample size and are markedly different from conventional normal asymptotics. Furthermore, the asymptotic results apply even when the time series exhibits long‐range dependence. Wavelet variances and stable asymptotics are then used to analyze three streamflows: one in Arizona, one in Connecticut, and one in Illinois. These analyses provide a deeper understanding of streamflow variability at different time scales (e.g., extreme variation at short time scales that are characteristic of heavy rainfall, presence of seasonal variations, and, in one case, some quasi‐biennial fluctuations). Furthermore, this paper includes evaluations of local characteristic time scales, a discussion of tail heaviness, computation of Hurst exponents, and some future directions of research.

Skewness and kurtosis analysis for non-Gaussian distributions

In this paper we address a number of pitfalls regarding the use of kurtosis as a measure of deviations from the Gaussian. We treat kurtosis in both its standard definition and that which arises in q-statistics, namely q-kurtosis. We have recently shown that the relation proposed by Cristelli et al. (2012) between skewness and kurtosis can only be verified for relatively small data sets, independently of the type of statistics chosen; however it fails for sufficiently large data sets, if the fourth moment of the distribution is finite. For infinite fourth moments, kurtosis is not defined as the size of the data set tends to infinity. For distributions with finite fourth moments, the size, N, of the data set for which the standard kurtosis saturates to a fixed value, depends on the deviation of the original distribution from the Gaussian. Nevertheless, using kurtosis as a criterion for deciding which distribution deviates further from the Gaussian can be misleading for small data sets, even for finite fourth moment distributions. Going over to q-statistics, we find that although the value of q-kurtosis is finite in the range of 0<q<3, this quantity is not useful for comparing different non-Gaussian distributed data sets, unless the appropriate q value, which truly characterizes the data set of interest, is chosen. Finally, we propose a method to determine the correct q value and thereby to compute the q-kurtosis of q-Gaussian distributed data sets.

Almost sure convergence of the largest and smallest eigenvalues of high-dimensional sample correlation matrices

In this paper, we show that the largest and smallest eigenvalues of a sample correlation matrix stemming from n independent observations of a p-dimensional time series with iid components converge almost surely to (1+γ)2 and (1−γ)2, respectively, as n→∞, if p∕n→γ∈(0,1] and the truncated variance of the entry distribution is “almost slowly varying”, a condition we describe via moment properties of self-normalized sums. Moreover, the empirical spectral distributions of these sample correlation matrices converge weakly, with probability 1, to the Marčenko–Pastur law, which extends a result in Bai and Zhou (2008). We compare the behavior of the eigenvalues of the sample covariance and sample correlation matrices and argue that the latter seems more robust, in particular in the case of infinite fourth moment. We briefly address some practical issues for the estimation of extreme eigenvalues in a simulation study.In our proofs we use the method of moments combined with a Path-Shortening Algorithm, which efficiently uses the structure of sample correlation matrices, to calculate precise bounds for matrix norms. We believe that this new approach could be of further use in random matrix theory.

A CLT for martingale transforms with infinite variance

We provide a CLT for martingale transforms that holds even when the second moments are infinite. Compared to an analogous result in Hall and Yao (2003) we impose minimal assumptions and utilize the Principle of Conditioning to verify a modified version of Lindeberg’s condition. When the variance is infinite, the rate of convergence, which we allow to be matrix valued, is slower than n and depends on the rate of divergence of the truncated second moments. In many cases it can be consistently estimated. A major application concerns the characterization of the rate and the limiting distribution of the Gaussian QMLE in the case of GARCH type models with infinite fourth moments for the innovation process. The results are particularly useful in the case of the EGARCH(1,1) model as we show that the usual limit theory is still valid without any further parameter restrictions when we relax the assumption for finite fourth moments of the innovation process.

Invariance principles and almost sure central limit theorem for the error variance estimator in linear models

ABSTRACTIn this article, we establish the invariance principles for the error variance estimator in linear models, which weaken the conditions for the errors in Chen (1980) from finite fourth moment to infinite fourth moment. Furthermore, the almost sure central limit theorem of the error variance estimator is also obtained under the assumption for the errors of infinite fourth moment.

Estimation and tests for TGTACH$\bm{(1, 1)}$ models with heavy-tailed errors: A uniform framework

This paper studies the estimation and tests of heavy-tailed TGARCH$(1, 1)$ models based on quasi-maximum likelihood estimation (QMLE) and percentile-$t$ subsample bootstrap method, where heavy-tail means that the distribution of the square errors are in the domain of attraction of a stable law with infinite fourth moment. Under regular conditions, we first establish that the QMLE of ARCH and GARCH coefficients are consistent and their asymptotic distribution is non-Gaussian but is some stable law with exponent $\kappa\in (1, 2)$ despite of whether the heavy tailed TGARCH model is strictly stationary or not. However, the QMLE of the location parameter is consistent only when the model is strictly stationary. Then we propose tests for strict stationarity and symmetry for heavy-tailed TGARCH$(1, 1)$ models based the above asymptotic results and percentile-$t$ subsample bootstrap method to overcome the difficulty that the scale parameter and asymptotic distribution depend on the unknown exponent. These tests can be used in the universal parameter space. Finally, we conduct a simulation study through Monte Carlo methods to illustrate the performance of the proposed methods with finite sample sizes and then give an empirical analysis for 5-year China Treasury Bond Futures to illustrate the application of our methods.

A CLT for Martingale Transforms with Infinite Variance

We provide a CLT for martingale transforms that holds even when the second moments are infinite. Compared to an analogous result in Hall and Yao [Econometrica 71 (2003) 285-317] we impose minimal assumptions and utilize the Principle of Conditioning to verify a modified version of the Lindeberg's condition. When the variance is infinite, the rate of convergence, which we allow to be matrix valued, is slower than \sqrt{n} and depends on the rate of divergence of the truncated second moments. In many cases it can be consistently estimated. A major application concerns the characterization of the rate and the limiting distribution of the Gaussian QMLE in the case of GARCH type models with infinite fourth moments for the innovation process. The results are particularly useful in the case of the EGARCH(1,1) model as we show that the usual limit theory is still valid without any further parameter restrictions when we relax the assumption for finite fourth moments of the innovation process.

Asymptotic theory for the sample covariance matrix of a heavy-tailed multivariate time series

In this paper we give an asymptotic theory for the eigenvalues of the sample covariance matrix of a multivariate time series. The time series constitutes a linear process across time and between components. The input noise of the linear process has regularly varying tails with index α∈(0,4); in particular, the time series has infinite fourth moment. We derive the limiting behavior for the largest eigenvalues of the sample covariance matrix and show point process convergence of the normalized eigenvalues. The limiting process has an explicit form involving points of a Poisson process and eigenvalues of a non-negative definite matrix. Based on this convergence we derive limit theory for a host of other continuous functionals of the eigenvalues, including the joint convergence of the largest eigenvalues, the joint convergence of the largest eigenvalue and the trace of the sample covariance matrix, and the ratio of the largest eigenvalue to their sum.

TARGETING ESTIMATION OF CCC-GARCH MODELS WITH INFINITE FOURTH MOMENTS

As an alternative to quasi-maximum likelihood, targeting estimation is a much applied estimation method for univariate and multivariate GARCH models. In terms of variance targeting estimation, recent research has pointed out that at least finite fourth moments of the data generating process is required, if one wants to perform inference in GARCH models by relying on asymptotic normality of the estimator. Such moment conditions may not be satisfied in practice for financial returns, highlighting a potential drawback of variance targeting estimation. In this paper, we consider the large-sample properties of the variance targeting estimator for the multivariate extended constant conditional correlation GARCH model when the distribution of the data generating process has infinite fourth moments. Using nonstandard limit theory, we derive new results for the estimator stating that, under suitable conditions, its limiting distribution is multivariate stable. The rate of consistency of the estimator is slower than$\sqrt T$and depends on the tail shape of the data generating process. A simulation study illustrates the derived properties of the targeting estimator.

Robust Estimation and Inference for Heavy Tailed Nonlinear GARCH

We develop new tail-trimmed QML estimators for nonlinear GARCH models with possibly heavy tailed errors. Tail-trimming allows both identification of the true parameter and asymptotic normality. In heavy tailed cases the rate of convergence is below but arbitrarily close to root-n, the highest possible amongst M-estimators for GARCH with errors that have an infinite fourth moment, and faster than QML. We present a consistent estimator of the covariance matrix that permits classic inference without knowledge of the rate of convergence. Finally, a simulation study shows our estimators trump existing ones for sharpness and approximate normality, and we apply them to financial returns data.

Robust M-Estimation for Heavy Tailed Nonlinear AR-GARCH

We develop new tail-trimmed M-estimation methods for heavy tailed Nonlinear AR-GARCH models. Tail-trimming allows both identification of the true parameter and asymptotic normality for nonlinear models with asymmetric errors. In heavy tailed cases the rate of convergence is infinitesimally close to the highest possible amongst M-estimators for a particular loss function, hence super- root(n)-convergence can be achieved in nonlinear AR models with infinite variance errors, and arbitrarily near root(n)-convergence for GARCH with errors that have an infinite fourth moment. We present a consistent estimator of the covariance matrix that permits classic inference without knowledge of the rate of convergence, and explore asymptotic covariance and bootstrap mean-squared-error methods for selecting trimming parameters. A simulation study shows the estimator trumps existing ones for AR and GARCH models based on sharpness, approximate normality, rate of convergence, and test accuracy. We then use the estimator to study asset returns data.

On estimation of the variances for critical branching processes with immigration

The conditional least-squares estimators of the variances are studied for a critical branching process with immigration that allows the offspring distributions to have infinite fourth moments. We derive different forms of limiting distributions for these estimators when the offspring distributions have regularly varying tails with index α. In particular, in the case in which 2 &amp;lt; α &amp;lt; 8/3, the normalizing factor of the estimator for the offspring variance is smaller than √n, which is different from that of Winnicki (1991).

On estimation of the variances for critical branching processes with immigration

The conditional least-squares estimators of the variances are studied for a critical branching process with immigration that allows the offspring distributions to have infinite fourth moments. We derive different forms of limiting distributions for these estimators when the offspring distributions have regularly varying tails with index α. In particular, in the case in which 2 &lt; α &lt; 8/3, the normalizing factor of the estimator for the offspring variance is smaller than √n, which is different from that of Winnicki (1991).

Break detection in the covariance structure of multivariate time series models

In this paper, we introduce an asymptotic test procedure to assess the stability of volatilities and cross-volatilites of linear and nonlinear multivariate time series models. The test is very flexible as it can be applied, for example, to many of the multivariate GARCH models established in the literature, and also works well in the case of high dimensionality of the underlying data. Since it is nonparametric, the procedure avoids the difficulties associated with parametric model selection, model fitting and parameter estimation. We provide the theoretical foundation for the test and demonstrate its applicability via a simulation study and an analysis of financial data. Extensions to multiple changes and the case of infinite fourth moments are also discussed.

A quadratic ARCH(∞) model with long memory and Lévy stable behavior of squares

We introduce a modification of the linear ARCH (LARCH) model (Giraitis, Robinson, and Surgailis (2000)) - a special case of Sentana's (1995) quadratic ARCH (QARCH) model - for which the conditional variance is a sum of a positive constant and the square of an inhomogeneous linear combination of past observations. Necessary and sufficient conditions for the existence of a stationary solution with finite variance are obtained. We give conditions under which the stationary solution with infinite fourth moment can exhibit long memory, the leverage effect, and a Lévy-stable limit behavior of partial sums of squares.

A quadratic ARCH(∞) model with long memory and Lévy stable behavior of squares

We introduce a modification of the linear ARCH (LARCH) model (Giraitis, Robinson, and Surgailis (2000)) - a special case of Sentana's (1995) quadratic ARCH (QARCH) model - for which the conditional variance is a sum of a positive constant and the square of an inhomogeneous linear combination of past observations. Necessary and sufficient conditions for the existence of a stationary solution with finite variance are obtained. We give conditions under which the stationary solution with infinite fourth moment can exhibit long memory, the leverage effect, and a Lévy-stable limit behavior of partial sums of squares.

Sample autocovariances of long-memory time series

We find the asymptotic distribution of the sample autocovariances of long-memory processes in cases of finite and infinite fourth moment. Depending on the interplay of assumptions on moments and the intensity of dependence, there are three types of convergence rates and limit distributions. In particular, a normal approximation with the standard rate does not always hold in practically relevant cases.

Innovations algorithm asymptotics for periodically stationary time series with heavy tails

The innovations algorithm can be used to obtain parameter estimates for periodically stationary time series models. In this paper we compute the asymptotic distribution for these estimates in the case where the underlying noise sequence has infinite fourth moment but finite second moment. In this case, the sample covariances on which the innovations algorithm are based are known to be asymptotically stable. The asymptotic results developed here are useful to determine which model parameters are significant. In the process, we also compute the asymptotic distributions of least squares estimates of parameters in an autoregressive model.