One-way and two-way risk filtering using generalized dynamic factor model in Tehran Stock Exchange

Abstract According to statistics, risk estimation makes unusual predictions without focusing on the relevant factors and only focusing on a set of equations. In this study, we used a spreadsheet data set of time series and a new method for risk estimation. This estimation was based on a generalized dynamic factor model (GDFM) and daily data series obtained from different measures of Tehran Stock Exchange over a 10-year period during 2008 to 2018. we first utilized a generalized dynamic factor model proposed by Forni et al in order to determine statistic and dynamic factors. In the second step, by using MATLAB, we estimated the joint component of the study series as Tehran Stock Exchange risk. Next, using the generalized least squares (GLS) method, we examined the impact of each of the filtered risks on the index returns. The results showed that although both risks estimated through one-side and two-side filtering substantially and significantly explain the changes in the performance of the studied indices, but the risk estimated through two-side filtering using GDFM can explain the returns changes much better and more accurate than the one-side filter using the same model.