Dynamic changepoint detection in count time series: a particle filter approach

We study Bayesian dynamic models for detecting changepoints in count time series that present structural breaks. As the inferential approach, we develop a parameter learning version of the algorithm proposed by Chopin [Chopin N. Dynamic detection of changepoints in long time series. Annals of the Institute of Statistical Mathematics 2007;59:349–366.], called the Chopin filter with parameter learning, which allows us to estimate the static parameters in the model. In this extension, the static parameters are addressed by using the kernel smoothing approximations proposed by Liu and West [Liu J, West M. Combined parameters and state estimation in simulation-based filtering. In: Doucet A, de Freitas N, Gordon N, editors. Sequential Monte Carlo methods in practice. New York: Springer-Verlag; 2001]. The proposed methodology is then applied to both simulated and real data sets and the time series models include distributions that allow for overdispersion and/or zero inflation. Since our procedure is general, robust and naturally adaptive because the particle filter approach does not require restrictive specifications to ensure its validity and effectiveness, we believe it is a valuable alternative for dealing with the problem of detecting changepoints in count time series. The proposed methodology is also suitable for count time series with no changepoints and for independent count data.     

Multivariate Real-Time Signal Extraction by a Robust Adaptive Regression Filter

We propose a new regression-based filter for extracting signals online from multivariate high frequency time series. It separates relevant signals of several variables from noise and (multivariate) outliers.

Unlike parallel univariate filters, the new procedure takes into account the local covariance structure between the single time series components. It is based on high-breakdown estimates, which makes it robust against (patches of) outliers in one or several of the components as well as against outliers with respect to the multivariate covariance structure. Moreover, the trade-off problem between bias and variance for the optimal choice of the window width is approached by choosing the size of the window adaptively, depending on the current data situation.

Furthermore, we present an advanced algorithm of our filtering procedure that includes the replacement of missing observations in real time. Thus, the new procedure can be applied in online-monitoring practice. Applications to physiological time series from intensive care show the practical effect of the proposed filtering technique.     

The sensitivity of robust unit root tests

The power properties of the rank-based Dickey–Fuller (DF) unit root test of Granger and Hallman [C. Granger and J. Hallman, Nonlinear transformations of integrated time series, J. Time Ser. Anal. 12 (1991), pp. 207–218] and the range unit root tests of Aparicio et al. [F. Aparicio, A. Escribano, and A. Siplos, Range unit root (RUR) tests: Robust against non-linearities, error distributions, structural breaks and outliers, J. Time Ser. Anal. 27 (2006), pp. 545–576] are considered when applied to near-integrated time series processes with differing initial conditions. The results obtained show the empirical powers of the tests to be generally robust to smaller deviations of the initial condition of the time series from its underlying deterministic component, particularly for more highly stationary processes. However, dramatic decreases in power are observed when either the mean or variance of the deviation of the initial condition is increased. The robustness of the rank- and range-based unit root tests and their higher power results relative to the seminal DF test have both been noted previously in the econometrics literature. These results are questioned by the findings of the present paper.     

Discontinuities in robust nonparametric regression with α-mixing dependence

The main idea of the paper is to introduce a robust regression estimation method under an α-mixing dependence assumption, staying free of any parametric model restrictions while also allowing for some sudden changes in the unknown regression function. The sudden changes in the model may correspond to discontinuity points (jumps) or higher order breaks (jumps in corresponding derivatives) as well. We firstly derive some important statistical properties for local polynomial M-smoother estimates and we will propose a statistical test to decide whether some given point of interest is significantly important for a change to occur or not. As the asymptotic distribution of the test statistic depends on quantities which are left unknown we also introduce a bootstrap algorithm which can be used to mimic the target distribution of interest. All necessary proofs are provided together with some experimental results from a simulation study and a real data example.     

Estimating and Testing Nonlinear Local Dependence Between Two Time Series

ABSTRACT

The most common measure of dependence between two time series is the cross-correlation function. This measure gives a complete characterization of dependence for two linear and jointly Gaussian time series, but it often fails for nonlinear and non-Gaussian time series models, such as the ARCH-type models used in finance. The cross-correlation function is a global measure of dependence. In this article, we apply to bivariate time series the nonlinear local measure of dependence called local Gaussian correlation. It generally works well also for nonlinear models, and it can distinguish between positive and negative local dependence. We construct confidence intervals for the local Gaussian correlation and develop a test based on this measure of dependence. Asymptotic properties are derived for the parameter estimates, for the test functional and for a block bootstrap procedure. For both simulated and financial index data, we construct confidence intervals and we compare the proposed test with one based on the ordinary correlation and with one based on the Brownian distance correlation. Financial indexes are examined over a long time period and their local joint behavior, including tail behavior, is analyzed prior to, during and after the financial crisis. Supplementary material for this article is available online.     

Robust online scale estimation in time series: A model-free approach

This paper presents variance extraction procedures for univariate time series. The volatility of a times series is monitored allowing for non-linearities, jumps and outliers in the level. The volatility is measured using the height of triangles formed by consecutive observations of the time series. This idea was proposed by Rousseeuw and Hubert [1996. Regression-free and robust estimation of scale for bivariate data. Comput. Statist. Data Anal. 21, 67–85] in the bivariate setting. This paper extends their procedure to apply for online scale estimation in time series analysis. The statistical properties of the new methods are derived and finite sample properties are given. A financial and a medical application illustrate the use of the procedures.     

Structural breaks in panel data: Large number of panels and short length time series

The detection of (structural) breaks or the so called change point problem has drawn increasing attention from the theoretical, applied economic and financial fields. Much of the existing research concentrates on the detection of change points and asymptotic properties of their estimators in panels when N, the number of panels, as well as T, the number of observations in each panel are large. In this paper we pursue a different approach, i.e., we consider the asymptotic properties when N→∞ while keeping T fixed. This situation is typically related to large (firm-level) data containing financial information about an immense number of firms/stocks across a limited number of years/quarters/months. We propose a general approach for testing for break(s) in this setup. In particular, we obtain the asymptotic behavior of test statistics. We also propose a wild bootstrap procedure that could be used to generate the critical values of the test statistics. The theoretical approach is supplemented by numerous simulations and by an empirical illustration. We demonstrate that the testing procedure works well in the framework of the four factors CAPM model. In particular, we estimate the breaks in the monthly returns of US mutual funds during the period January 2006 to February 2010 which covers the subprime crises.     

Detecting structural breaks in multivariate financial time series: evidence from hedge fund investment strategies

This paper extends the class of asset-based style factor models with multiple structural breaks to the multivariate setting. We propose a model that allows for the presence of common breaks in a system of factor models for individual hedge fund investment strategies, which share common investment characteristics. We develop a Bayesian approach to inference for the unknown number and positions of the structural breaks, based on a set of filtering recursions similar to those of the forward–backward algorithm. Furthermore, we identify relevant risk factors, common among the series of hedge funds, using a Bayesian model comparison approach. We apply our method to a set of correlated hedge fund strategies, which are mainly characterized by equity related bets. Multiple common breaks are identified, consistent with well-known market events, which reveal evidence for structural changes in the risk exposures as well as in the correlation structure of the analysed series.     

Nonparametric monitoring of financial time series by jump-preserving control charts

Since structural changes in a possibly transformed financial time series may contain important information for investors and analysts, we consider the following problem of sequential econometrics. For a given time series we aim at detecting the first change-point where a jump of size a occurs, i.e., the mean changes from, say, m ₀to m ₀+ a and returns to m ₀after a possibly short period s. To address this problem, we study a Shewhart-type control chart based on a sequential version of the sigma filter, which extends kernel smoothers by employing stochastic weights depending on the process history to detect jumps in the data more accurately than classical approaches. We study both theoretical properties and performance issues. Concerning the statistical properties, it is important to know whether the normed delay time of the considered control chart is bounded, at least asymptotically. Extending known results for linear statistics employing deterministic weighting schemes, we establish an upper bound which holds if the memory of the chart tends to infinity. The performance of the proposed control charts is studied by simulations. We confine ourselves to some special models which try to mimic important features of real time series. Our empirical results provide some evidence that jump-preserving weights are preferable under certain circumstances.     

Fractional integration versus level shifts: the case of realized asset correlations

Long memory has been widely documented for realized financial market volatility. As a novelty, we consider daily realized asset correlations and we investigate whether the observed persistence is (i) due to true long memory (i.e. fractional integration) or (ii) artificially generated by some structural break processes. These two phenomena are difficult to be distinguished in practice. Our empirical results strongly indicate that the hyperbolic decay of the autocorrelation functions of pair-wise realized correlation series is indeed not driven by a truly fractionally integrated process. This finding is robust against user specific parameter choices in the applied test statistic and holds for all 15 considered time series. As a next step, we apply simple models with deterministic level shifts. When selecting the number of breaks, estimating the breakpoints and the corresponding structural break models we find a substantial degree of co-movement between the realized correlation series hinting at co-breaking. The estimated structural break models are interpreted in the light of the historic economic and financial development.     

Bootstrap procedures for variance breaks test in time series with a changing trend

In this article, we consider the problem of testing for variance breaks in time series in the presence of a changing trend. In performing the test, we employ the cumulative sum of squares (CUSSQ) test introduced by Inclán and Tiao (1994, J.?Amer.?Statist.?Assoc., 89, 913 ? 923). It is shown that CUSSQ test is not robust in the case of broken trend and its asymptotic distribution does not convergence to the sup of a standard Brownian bridge. As a remedy, a bootstrap approximation method is designed to alleviate the size distortions of test statistic while preserving its high power. Via a bootstrap functional central limit theorem, the consistency of these bootstrap procedures is established under general assumptions. Simulation results are provided for illustration and an empirical example of application to a set of high frequency real data is given.     

Integer autoregressive models with structural breaks

Even though integer-valued time series are common in practice, the methods for their analysis have been developed only in recent past. Several models for stationary processes with discrete marginal distributions have been proposed in the literature. Such processes assume the parameters of the model to remain constant throughout the time period. However, this need not be true in practice. In this paper, we introduce non-stationary integer-valued autoregressive (INAR) models with structural breaks to model a situation, where the parameters of the INAR process do not remain constant over time. Such models are useful while modelling count data time series with structural breaks. The Bayesian and Markov Chain Monte Carlo (MCMC) procedures for the estimation of the parameters and break points of such models are discussed. We illustrate the model and estimation procedure with the help of a simulation study. The proposed model is applied to the two real biometrical data sets.     

2005年汇改后人民币汇率多次结构突变检验分析——基于循序检验方法

采用最新的多次结构突变循序检验方法,对2005年7月21日汇改后人民币汇率时间序列趋势项是否具有多次结构突变进行研究,并在多次结构突变检验结果的基础上对消除趋势后的人民币汇率数据进行分析,结果发现：人民币汇率时间序列是围绕着4个结构断点的分段趋势平稳的;人民币汇率服从分段趋势平稳的结论对汇率政策有效性、汇率与其他经济总量关系研究及汇率预测具有重要意义。     

Monitoring parameter change in time series models

In this paper, we develop a monitoring procedure for an early detection of parameter changes in time series models. We design the monitoring procedure in general time series models and apply it to the changes for the autocovariances of linear processes, GARCH parameters, and underlying distributions. Simulation results are provided for illustration.     

Maximum studentized score tests for the detection of outliers in time series regression models

Efficient score tests exist among others, for testing the presence of additive and/or innovative outliers that are the result of the shifted mean of the error process under the regression model. A sample influence function of autocorrelation-based diagnostic technique also exists for the detection of outliers that are the result of the shifted autocorrelations. The later diagnostic technique is however not useful if the outlying observation does not affect the autocorrelation structure but is generated due to an inflation in the variance of the error process under the regression model. In this paper, we develop a unified maximum studentized type test which is applicable for testing the additive and innovative outliers as well as variance shifted outliers that may or may not affect the autocorrelation structure of the outlier free time series observations. Since the computation of the p-values for the maximum studentized type test is not easy in general, we propose a Satterthwaite type approximation based on suitable doubly non-central F-distributions for finding such p-values [F.E. Satterthwaite, An approximate distribution of estimates of variance components, Biometrics 2 (1946), pp. 110–114]. The approximations are evaluated through a simulation study, for example, for the detection of additive and innovative outliers as well as variance shifted outliers that do not affect the autocorrelation structure of the outlier free time series observations. Some simulation results on model misspecification effects on outlier detection are also provided.     

Generalized fixed‐T panel unit root tests

Panel data unit root tests, which can be applied to data that do not have many time series observations, are based on very restrictive error and deterministic component specification assumptions. In this paper, we develop a new, doubly modified estimator, based on which we propose a panel unit root test that allows for multiple structural breaks, linear and nonlinear trends, heteroscedasticity, serial correlation, and error cross‐section heterogeneity, when the number of time series observations is finite. The test has the additional perk that it is invariant to the initial condition.     

A new unit root test with two structural breaks in level and slope at unknown time

In this paper, we propose a new augmented Dickey–Fuller-type test for unit roots which accounts for two structural breaks. We consider two different specifications: (a) two breaks in the level of a trending data series and (b) two breaks in the level and slope of a trending data series. The breaks whose time of occurrence is assumed to be unknown are modeled as innovational outliers and thus take effect gradually. Using Monte Carlo simulations, we show that our proposed test has correct size, stable power, and identifies the structural breaks accurately.     

A new nonlinear unit root test with Fourier function

Abstract

Traditional unit root tests display a tendency to be nonstationary in the case of structural breaks and nonlinearity. To eliminate this problem this paper proposes a new flexible Fourier form nonlinear unit root test. This test eliminates this problem to add structural breaks and nonlinearity together to the test procedure. In this test procedure, structural breaks are modeled by means of a Fourier function and nonlinear adjustment is modeled by means of an exponential smooth threshold autoregressive (ESTAR) model. The simulation results indicate that the proposed unit root test is more powerful than the Kruse and KSS tests.     

Continuous-time threshold autoregressions with jumps: Properties,estimation, and application to electricity markets

Continuous-time autoregressive processes have been applied successfully in many fields and are particularly advantageous in the modeling of irregularly spaced or high-frequency time series data. A convenient nonlinear extension of this model are continuous-time threshold autoregressions (CTAR). CTAR allow for greater flexibility in model parameters and can represent a regime switching behavior. However, so far only Gaussian CTAR processes have been defined, so that this model class could not be used for data with jumps, as frequently observed in financial applications. Hence, as a novelty, we construct CTAR processes with jumps in this paper. Existence of a unique weak solution and weak consistency of an Euler approximation scheme is proven. As a closed form expression of the likelihood is not available, we use kernel-based particle filtering for estimation. We fit our model to the Physical Electricity Index and show that it describes the data better than other comparable approaches.     

Testing for Threshold Diffusion

The threshold diffusion model assumes a piecewise linear drift term and a piecewise smooth diffusion term, which constitutes a rich model for analyzing nonlinear continuous-time processes. We consider the problem of testing for threshold nonlinearity in the drift term. We do this by developing a quasi-likelihood test derived under the working assumption of a constant diffusion term, which circumvents the problem of generally unknown functional form for the diffusion term. The test is first developed for testing for one threshold at which the drift term breaks into two linear functions. We show that under some mild regularity conditions, the asymptotic null distribution of the proposed test statistic is given by the distribution of certain functional of some centered Gaussian process. We develop a computationally efficient method for calibrating the p-value of the test statistic by bootstrapping its asymptotic null distribution. The local power function is also derived, which establishes the consistency of the proposed test. The test is then extended to testing for multiple thresholds. We demonstrate the efficacy of the proposed test by simulations. Using the proposed test, we examine the evidence of nonlinearity in the term structure of a long time series of U.S. interest rates.