Authors: Andrei Rusu
Addresses: Department of Statistics Forecasts Mathematics, Babeș-Bolyai University, Cluj-Napoca, 400591, Romania
Abstract: Risk and uncertainty are concepts found in every financial environment. In order to anticipate and prevent their losses, financial market participants use various measures to quantify risk. One frequently used measure is value at risk (VaR). This study is focused on comparing and assessing the performance for a set of parametric and semi-parametric methods of estimating VaR, highlighting some of the less investigated approaches in empirical literature, which have significant proven performance nonetheless. In order to evaluate the out-of-sample performance of this measure, we considered the evolution of stock indexes from 14 international markets, between years 2006-2016. Thus, the data covers periods characterised by stability and also periods of extreme events like the financial crisis which started in 2007-2008. The results showed that filtered historical simulation performed best on all indexes. It was also found that taking into account the asymmetric character of financial information leads to more accurate predictions.
Keywords: value at risk; VaR; APARCH; generalised Pareto distribution; GDP; back testing; financial markets.
International Journal of Financial Markets and Derivatives, 2018 Vol.6 No.4, pp.321 - 334
Available online: 21 Jan 2019 *Full-text access for editors Access for subscribers Purchase this article Comment on this article