CMStatistics 2019: Start Registration
View Submission - CFE
A0968
Title: The non-linear relationship between macro-financial state variables and market volatility Authors:  Johan Jakobsen - Nordea (Denmark) [presenting]
Daniel Borup - Aarhus University (Denmark)
Bezirgen Veliyev - Aarhus University (Denmark)
Abstract: The relationship between the level of financial market volatility and macro-financial state variables is investigated. To this end, we propose a parametric alternative to the GARCH model, labelled the ST-GARCH model, in which the level (intercept) is allowed to be time-varying as a smooth function of those state variables. This model entertains non-linear relationship(s) between the state variable and volatility. We establish the asymptotic theory of the quasi-maximum-likelihood estimator and develop a testing framework for selecting the number of transitions. Our broad application to the equity, foreign exchange, fixed income, credit, and real estate markets document clear level shifts in volatility. As such, our ST-GARCH realizes significant in-sample gains by capturing those (non-linear) level shifts using macro-financial state variables. This also translates to notable out-of-sample gains.