Benchmarking, temporal distribution, and reconciliation methods for time series. (Q2494599)

Before their official release by statistical agencies, time series data are subject to several adjustments intended to increase efficiency, reduce bias, replace missing values, correct errors, facilitate analysis and fulfill cross-sectional additivity constraints. This book is a very detailed course on statistical methods for such adjustments, focused on benchmarking, interpolation, temporal distribution, calendarization, and reconciliation. The goal of the authors is to point out the properties and limitations of existing procedures ranging from ad hoc to regression-based methods, to signal extraction based models. They also bring together the scattered literature on these topics and present them using a consistent notation and a unified view. The book consists of fourteen chapters and five appendices. It starts with a general introduction of the topics discussed. Chapter 2 is intended as a general introduction to time series analysis. Chapters 3, 4 and 5 discuss the additive and multiplicative regression-based model developed by \textit{P.A. Cholette} and \textit{E.B. Dagum} [Int. Stat. Rev. 62, No.3, 365-377 (1994; Zbl 0831.62062)]. Chapter 6 deals with the classical Denton method for benchmarking and interpolation based on the principle of movement preservation by \textit{F.T. Denton} [J. Am. Stat. Assoc. 66, 99-102 (1971; Zbl 0216.22801)]. Chapter 7 focuses on interpolation, temporal distribution and extrapolation. The presented methods include ad hoc procedures, regression-model based, ARIMA model based, the procedures of \textit{G.C. Chow} and \textit{A.-L. Lin} [Rev. Econ. Stat. 53, 372--375 (1971); J. Am. Stat. Assoc. 71, 719--721 (1976; Zbl 0342.62018)], and its dynamic extensions. Chapter 8 provides an overview of benchmarking methods which are based on signal extraction. In chapter 9, the problem of calendarization is examined. Chapter 10 discusses a unified regression-based framework for signal extraction, benchmarking and interpolation developed by \textit{E.B. Dagum, P.A. Cholette} and \textit{Z.-G. Chen} [Int. Stat. Rev. 66, No. 3, 245-269 (1998; Zbl 0952.62080)]. The last chapters 11 to 14 deal with adjustments for systems of time series, namely reconciliation or balancing. These chapters provide analytical solutions which enable implementation for large systems of series.The appendices A, B and C contain proofs of important theorems and a number of matrix algebra results. Appendices D and E discuss various seasonal and trading-day regressors for regression-based models. Each chapter of the book is self-contained and illustrates the methods discussed with a large number of real data examples. The book can be recommended as a very useful textbook for students as well as a reference book for researchers and practitioners in the field.