Unit Root Testing in Presence of a Double Threshold Process

In this paper we propose a double threshold process that generalizes the threshold autoregressive one widely known in the literature. It is characterized by a structure with two thresholds: the first regulates the switching between two autoregressive regimes; the second threshold regulates the switching between the two regimes of the stationary innovations. A testing procedure based on a Wald statistic has been given to evaluate the presence of unit roots in the process against stationarity. The asymptotic distribution of the statistic has been derived and the size and the power of the test have been evaluated through a Monte Carlo study where the proposed test is compared to two competing unit root testing procedures. The results clearly highlight the advantage obtained from the proposed test as the asymmetry of the generating process increases.     

A Combinatorial Approach to Piecewise Linear Time Series Analysis

Over recent years, several nonlinear time series models have been proposed in the literature. One model that has found a large number of successful applications is the threshold autoregressive model (TAR). The TAR model is a piecewise linear process whose central idea is to change the parameters of a linear autoregressive model according to the value of an observable variable, called the threshold variable. If this variable is a lagged value of the time series, the model is called a self-exciting threshold autoregressive (SETAR) model. In this article, we propose a heuristic to estimate a more general SETAR model, where the thresholds are multivariate. We formulate the task of finding multivariate thresholds as a combinatorial optimization problem. We develop an algorithm based on a greedy randomized adaptive search procedure (GRASP) to solve the problem. GRASP is an iterative randomized sampling technique that has been shown to quickly produce good quality solutions for a wide variety of optimization problems. The proposed model performs well on both simulated and real data.     

Weak convergence of the sequential empirical processes of residuals in TAR models

This paper studies the weak convergence of the sequential empirical process K n of the residuals in the threshold autoregressive(TAR)model of order p.Under some mild conditions,it is shown that K n converges weakly to a Kiefer process plus a random variable which converges to a multivariate normal.This differs from that given by Bai(1994)for a stationary autoregressive and moving average(ARMA)model.     

On the approximation of continuous time threshold ARMA processes

Threshold autoregressive (AR) and autoregressive moving average (ARMA) processes with continuous time parameter have been discussed in several recent papers by Brockwellet al. (1991,Statist. Sinica,1, 401–410), Tong and Yeung (1991,Statist. Sinica,1, 411–430), Brockwell and Hyndman (1992,International Journal Forecasting,8, 157–173) and Brockwell (1994,J. Statist. Plann. Inference,39, 291–304). A threshold ARMA process with boundary width 2>0 is easy to define in terms of the unique strong solution of a stochastic differential equation whose coefficients are piecewise linear and Lipschitz. The positive boundary-width is a convenient mathematical device to smooth out the coefficient changes at the boundary and hence to ensure the existence and uniqueness of the strong solution of the stochastic differential equation from which the process is derived. In this paper we give a direct definition of a threshold ARMA processes with =0 in the important case when only the autoregressive coefficients change with the level of the process. (This of course includes all threshold AR processes with constant scale parameter.) The idea is to express the distributions of the process in terms of the weak solution of a certain stochastic differential equation. It is shown that the joint distributions of this solution with =0 are the weak limits as 0 of the distributions of the solution with >0. The sense in which the approximating sequence of processes used by Brockwell and Hyndman (1992,International Journal Forecasting,8, 157–173) converges to this weak solution is also investigated. Some numerical examples illustrate the value of the latter approximation in comparison with the more direct representation of the process obtained from the Cameron-Martin-Girsanov formula. It is used in particular to fit continuous-time threshold models to the sunspot and Canadian lynx series.Research partially supported by National Science Foundation Research Grants DMS 9105745 and 9243648.     

周期相关时间序列与周期自回归模型

介绍了周期相关时间序列和周期自回归模型,并研究了周期自回归时间序列的稳定性及周期性,得到了它为周期相关时间序列的一个充要条件,推广了文献[1]的结论.     

Global maxima of real-valued functions

In this paper, sufficient conditions for a local maximum to be global are discussed. The result is that a real continuously differentiable functionf, defined on a subset ofR ⁿ, under fairly weak conditions, is unimodal iff has a strict local maximum at any stationary point.The author would like to thank the referee for detailed and constructive comments.     

Weak Stationarity: Eliminating the gap between necessary and sufficient conditions

Starting from known necessary extremality conditions in terms of strict subdifferentials and normals the notion of weak stationarity is introduced. It is defined in terms of initial space elements. The necessary conditions become necessary and sufficient (for stationarity).     

Estimation and strict stationarity testing of ARCH processes based on weighted least squares

This paper proposes a unified unconstrained two-stage weighted least squares estimate (2S-WLSE) theory for both stationary and nonstationary ARCH(1) processes. Without assuming strict stationarity, we show that the unconstrained 2S-WLSE of the conditional variance slope ARCH(1) parameter is consistent and asymptotically Gaussian and has the same asymptotic variance as its unconstrained quasi-maximum likelihood counterpart. Moreover, a consistent estimate of the asymptotic variance of the 2S-WLSE is provided irrespective of the stationarity requirement. As a result, strict stationarity testing of the ARCH process is considered. A numerical illustration on simulated and real data assesses the theory in finite samples.     

Variable-Length Coding of Two-sided Asymptotically Mean Stationary Measures

We collect several observations that concern variable-length coding of two-sided infinite sequences in a probabilistic setting. Attention is paid to images and preimages of asymptotically mean stationary measures defined on subsets of these sequences. We point out sufficient conditions under which the variable-length coding and its inverse preserve asymptotic mean stationarity. Moreover, conditions for preservation of shift-invariant σ-fields and the finite-energy property are discussed, and the block entropies for stationary means of coded processes are related in some cases. Subsequently, we apply certain of these results to construct a stationary nonergodic process with a desired linguistic interpretation.     

Multiple autoregressive models with random coefficients

This paper derives conditions for the stationarity of a class of multiple autoregressive models with random coefficients. The models considered include as special cases those previously discussed by Andel (Ann. Math. Statist.42 (1971), 755–759; Math. Operationsforsch. Statist.7 (1976), 735–741).     

Kernel estimation for stationary density of Markov chains with general state space

Let {X _n } _n ≥0 be a Markov chain with stationary distributionf(x)ν(dx), ν being a σ-finite measure onE⊂R ^d . Under strict stationarity and mixing conditions we obtain the consistency and asymptotic normality for a general class of kernel estimates off(·). When the assumption of stationarity is dropped these results are extended to geometrically ergodic chains. Partially supported by CAPES. Partially supported by CNPq, PROCAD/CAPES, PRONEX/FAPDF and FINATEC/UnB.     

A scalar dynamic conditional correlation model: Structure and estimation

The dynamic conditional correlation(DCC) model has been widely used for modeling the conditional correlation of multivariate time series by Engle(2002). However, the stationarity conditions have been established only recently and the asymptotic theory of parameter estimation for the DCC model has not yet to be fully discussed. In this paper, we propose an alternative model, namely the scalar dynamic conditional correlation(SDCC) model. Sufficient and easily-checked conditions for stationarity, geometric ergodicity, andβ-mixing with exponential-decay rates are provided. We then show the strong consistency and asymptotic normality of the quasi-maximum-likelihood estimator(QMLE) of the model parameters under regular conditions.The asymptotic results are illustrated by Monte Carlo experiments. As a real-data example, the proposed SDCC model is applied to analyzing the daily returns of the FSTE(financial times and stock exchange) 100 index and FSTE 100 futures. Our model improves the performance of the DCC model in the sense that the Li-Mc Leod statistic of the SDCC model is much smaller and the hedging efficiency is higher.     

On identification of the threshold diffusion processes

We consider the problems of parameter estimation for several models of threshold ergodic diffusion processes in the asymptotics of large samples. These models are the direct continuous time analogues of the well known in time series analysis threshold autoregressive models. In such models, the trend is switching when the observed process attaints some (unknown) values and the problem is to estimate it or to test some hypotheses concerning these values. The related statistical problems correspond to the singular estimation or testing, for example, the rate of convergence of estimators is T and not ?T{\sqrt{T}} as in regular estimation problems. We study the asymptotic behavior of the maximum likelihood and Bayesian estimators and discuss the possibility of the construction of the goodness-of-fit test for such models of observation.     

Regression quantiles for unstable autoregressive models

This paper investigates regression quantiles (RQ) for unstable autoregressive models. The uniform Bahadur representation of the RQ process is obtained. The joint asymptotic distribution of the RQ process is derived in a unified manner for all types of characteristic roots on or outside the unit circle. It involves stochastic integrals in terms of a sequence of independent and identically distributed multivariate Brownian motions with correlated components. The related L-estimator is also discussed. The asymptotic distributions of the RQ and the L-estimator corresponding to the nonstationary componentwise arguments can be transformed into a function of a normal random variable and a sequence of i.i.d. univariate Brownian motions. This is different from the analysis based on the LSE in the literature. As an auxiliary theorem, a weak convergence of a randomly weighted residual empirical process to the stochastic integral of a Kiefer process is established. The results obtained in this paper provide an asymptotic theory for nonstationary time series processes, which can be used to construct robust unit root tests.     

On categorical time series models with covariates

We study the problem of stationarity and ergodicity for autoregressive multinomial logistic time series models which possibly include a latent process and are defined by a GARCH-type recursive equation. We improve considerably upon the existing conditions about stationarity and ergodicity of those models. Proofs are based on theory developed for chains with complete connections. A useful coupling technique is employed for studying ergodicity of infinite order finite-state stochastic processes which generalize finite-state Markov chains. Furthermore, for the case of finite order Markov chains, we discuss ergodicity properties of a model which includes strongly exogenous but not necessarily bounded covariates.     

Autoregressive approximation in nonstandard situations: the fractionally integrated and non-invertible cases

Autoregressive models are commonly employed to analyze empirical time series. In practice, however, any autoregressive model will only be an approximation to reality and in order to achieve a reasonable approximation and allow for full generality the order of the autoregression, h say, must be allowed to go to infinity with T, the sample size. Although results are available on the estimation of autoregressive models when h increases indefinitely with T such results are usually predicated on assumptions that exclude (1) non-invertible processes and (2) fractionally integrated processes. In this paper we will investigate the consequences of fitting long autoregressions under regularity conditions that allow for these two situations and where an infinite autoregressive representation of the process need not exist. Uniform convergence rates for the sample autocovariances are derived and corresponding convergence rates for the estimates of AR(h) approximations are established. A central limit theorem for the coefficient estimates is also obtained. An extension of a result on the predictive optimality of AIC to fractional and non-invertible processes is obtained.     

非对称广义自回归条件异方差的新模型

本文提出了一个新的非对称广义自回归条件异方差的新模型,证明了该模型宽平稳及其最简模型偶数价矩存在的充要条件。     

A Hessian Regularized Nonlinear Time Series Model

We introduce a nonparametric nonlinear time series model. The novel idea is to fit a model via penalization, where the penalty term is an unbiased estimator of the integrated Hessian of the underlying function. The underlying model assumption is very general: it has Hessian almost everywhere in its domain. Numerical experiments demonstrate that our model has better predictive power: if the underlying model complies with an existing parametric/semiparametric form (e.g., a threshold autoregressive model (TAR), an additive autoregressive model (AAR), or a functional coefficient autoregressive model (FAR)), our model performs comparably; if the underlying model does not comply with any preexisting form, our model outperforms in nearly all simulations. We name our model a Hessian regularized nonlinear model for time series (HRM). We conjecture on theoretical properties and use simulations to verify. Our method can be viewed as a way to generalize splines to high dimensions (when the number of variates is more than three), under which an analogous analytical derivation cannot work due to the curse of dimensionality. Supplemental materials are provided, and will help readers reproduce all results in the article.     

Bridging between short-range and long-range dependence with mixed spatio-temporal Ornstein–Uhlenbeck processes

While short-range dependence is widely assumed in the literature for its simplicity, long-range dependence is a feature that has been observed in data from finance, hydrology, geophysics and economics. In this paper, we extend a Lévy-driven spatio-temporal Ornstein–Uhlenbeck process by randomly varying its rate parameter to model both short-range and long-range dependence. This particular set-up allows for non-separable spatio-temporal correlations which are desirable for real applications, as well as flexible spatial covariances which arise from the shapes of influence regions. Theoretical properties such as spatio-temporal stationarity and second-order moments are established. An isotropic g-class is also used to illustrate how the memory of the process is related to the probability distribution of the rate parameter. We develop a simulation algorithm for the compound Poisson case which can be used to approximate other Lévy bases. The generalized method of moments is used for inference and simulation experiments are conducted with a view towards asymptotic properties.     

Tail Behavior of a Threshold Autoregressive Stochastic Volatility Model

We consider a threshold autoregressive stochastic volatility model where the driving noises are sequences of iid regularly random variables. We prove that both the right and the left tails of the marginal distribution of the log-volatility process (α_t)_t are regularly varying with tail exponent −α with α > 0. We also determine the exact values of the coefficients in the tail behaviour of the process (α_t)_t. AMS 2000 Subject Classification. Primary—62G32, 62PO5