I. Introduction
Extracting economic characteristics of economic data is a useful way to understand social conditions. For example, there is a study that examines the factors of stock and bond returns and finds general fluctuations in the stock and bond markets [1]. Economic data are time-series data, and there are several methods for filtering such time-series data, including Hodrick-Prescott (HP) filtering [2] and the exponential smoothing filter [3]. Furthermore, Yamada showed that the two smoothing methods can be regarded as a type of graph spectral filter [4].