Time Series Chain Graphical Models in the Inference of Economic Data: A Case Study from S&P 500

Abstract

Main purpose of this study is the investigation of the relationships between economic and financial variables. This subject is well documented in the literature for both emerging and developed markets but, the contribution of this study to the literature is that the direction of the relationships is investigated by using a different method. In this study, the time series chain graphical model is utilized to examine the relationship between selected economic and financial variables over time. Time Series Chain Graphical Model enables to explore the conditional dependence among variables that are repeatedly measured at different time points. Our research validates the accuracy of the proposed model by segmenting the data by year. Additionally, graphical models are employed for precision and autocorrelation matrix analysis. We use the USA dataset, which can be found in the study of Gloyal and Welch (2021), there exist 16 variables that exhibit occasional conditional dependence and infrequent temporal dependence. This analysis, which is important in showing policy makers whether there is a relationship between variables, can also be applied to Turkish data at later stages.