Empirical likelihood and quantile methods for time series. Efficiency, robustness, optimality, and prediction (Q1722900)

The book is devoted to some questions of statistical inference for time series models. The primary aim is to discuss the nonparametric methods such as quantile method and empirical likelihood method. Chapter 1 (Introduction) provides some basics of stationary time series, also explains interpolation and extrapolation problems which are some extensions of the concept of prediction for a stationary process. Chapter 2 deals with the minimum contrast estimation of time series parameters. The authors introduce a new class of contrast functions and investigate asymptotic properties of corresponding minimum contrast estimators. Quantile method of parameter estimation of stationary time series, together with its asymptotic properties and applications to the hypothesis testing problem for quantiles, is studied in Chapter 3. Chapter 4 discusses the empirical likelihood method in the frequency domain, considering the Whittle likelihood as an estimating function. The asymptotic properties of the empirical likelihood ratio statistics are investigated. In Chapter 5, the authors extend the empirical likelihood method to the self-weighted version, which allows to include possibly infinite variance time series, and derive the pivotal limit distributions of the corresponding statistics. \par The discussed methods and theoretical results are illustrated with numerical simulations. Most of presented results are obtained by the authors. The book can be useful for researches who are interested in time series analysis and statistical inference.