In this book, we explain the development of a new filtering method to estimate the hidden states of random variables for multiple non-stationary time series data. This method is particularly helpful in analyzing small-sample non-stationary macro-economic time series. The method is based on the frequency-domain application of the separating information maximum likelihood (SIML) method, which was proposed by Kunitomo, Sato, and Kurisu (Springer, 2018) for financial high-frequency time series. We solve the filtering problem of hidden random variables of trend-cycle, seasonal, and measurement-error components and propose a method to handle macro-economic time series. The asymptotic theory based on the frequency-domain analysis for non-stationary time series is developed with illustrative applications, including properties of the method of Muller and Watson (2018), and analyses of macro-economic data in Japan.

Vast research has been carried out on the use of statistical time series analysis for macro-economic time series. One important feature of the series, which is different from standard statistical time series analysis, is that the observed time series is an apparent mixture of non-stationary and stationary components. We apply the SIML method for estimating the non-stationary errors-in-variables models. As well, we discuss the asymptotic and finite sample properties of the estimation of unknown parameters in the statistical models. Finally, we utilize their results to solve the filtering problem of hidden random variables and to show that they lead to new a way to handle macro-economic time series.

Presents a systematic treatment of a new filtering method for analyzing economic time series Discusses a robust filtering method of trend cycles, seasonality, and measurement error for economic time series data Includes applications of macro-time series data in Japan

Inhalt

Introduction.- Macro Examples and Non-Stationary Errors-in-Variables Model.- The SIML Filtering Method.- Comparing Estimation Methods of Non-stationary Errors-in Variables Models.- Frequency Regression and Smoothing for Noisy Non-stationary Multivariate Time Series.