Title Sisteminės rizikos užkrato plitimo tyrimas modeliuojant valiutų kursų dažnių jungiamumą /
Translation of Title Systemic risk contagion modelling in foreign exchange market using frequency connectedness.
Authors Jomantas, Jokūbas
Full Text Download
Pages 56
Keywords [eng] volatility ; connectedness ; frequency ; spectral analysis ; systemic risk
Abstract [eng] Systemic risk contagion modelling became an important topic after the global financial crisis of 2008. There‘s a number of definitions for systemic risk contagion, although there is no one unifying one that is agreed upon. Nevertheless, there exist a large number of methods that allow the modelling of systemic risk. In this work systemic risk is measured by modelling connectedness between different foreign exchange rates that arise due to heterogeneous frequency responses to shocks. The paper focuses on connectedness in short-, medium-, and long-term financial cycles. For this I use the methodology first proposed by Diebold and Yilmaz, and later extended by Baruník and Křehlík to work with frequency responses, which is based on VAR forecast error variance decompositions. Different from other papers on this subject, this paper suggests to only use bad volatility, which is approximated using high frequency negative return data. The purpose of this paper is to show, that using only the bad volatility for measuring connectedness, we can get a model that lets us properly assess the current economic situation. To get a proper approximation of volatility I use high frequency data (measured every 5 minutes) and realised semivariance modelling proposed by Barndorff-Nielsen et. al. For comparison, volatility achieved with GARCH(1,1) from low frequency daily data is used. In the paper I show, that using volatility from low frequency data introduces background noise which makes it impossible to decompose connectedness into different frequency bands. Using high frequency data we get a model that allows us to properly analyse the various timeframes of financial instability over the last 16 years by using connectedness distributions in different frequency bands. In the paper I show, that during crises of global proportions, the largest portion of connectedness is measured in the low frequency band, while the connectedness in the higher frequency bands sharply drops. Based on this, we can assume that during global financial crises, connectedness is created at lower frequencies and the shocks are more persistent and are being transmitted over longer periods of time.
Dissertation Institution Kauno technologijos universitetas.
Type Master thesis
Language Lithuanian
Publication date 2020