Institutional Repository of Vilnius University
Backtesting of expected shortfall using quantile representation on baltic equity index example /
Abstract
dc:description41 pages (excluding summaries and annexes), 9 figures (17 subfigures), 10 tables, 32 unique references, 52 equations. The main aim of this academic work is to consider the extension of industry-standard value-at-risk backtesting approaches (``exceedance-based approaches") to the backtesting of expected shortfall as a superior risk measure. The research is motivated by the upcoming implementation of the Fundamental Review of the Trading Book regulation, which sanctions the use of expected shortfall instead of popular value-at-risk. The paper consists of four main parts: the analysis of literature, the theoretical part, the empirical part, the conclusions and recommendations. The literature analysis reviews the main approaches to the risk measures' estimation, common types of dynamic volatility models, and backtesting approaches for both value-at-risk and expected shortfall. Having conducted the literature analysis, the theoretical part of the thesis provides the approach of transforming expected shortfall into its quantile form, which is suitable for exceedance-based backtesting. The approach is described for four different estimation approaches, namely the analytical delta-normal method, analytical Generalised Pareto Distribution-based method, numerical Monte Carlo method and non-parametric historical simulation method. Additionally, the adjustment to the exceedance-based approaches is proposed, which can help accommodate the lack of stability in the expected shortfall quantile estimation. In the empirical part, an illustrative example on the Baltic equity index is provided for all four approaches considered and a separate example for the non-stable quantile case. Three confidence levels were used throughout, corresponding to the common confidence levels for risk models - 90%, 95% and 97.5%, where the latter is the regulatory-defined requirement. The work is concluded with a discussion on the statistical power of tests and the trade-off with the straightforwardness of application and interpretation.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Liakhovchenko, Andrii,
Subjects
dc:subject × 1Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- eng
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD239010975&prefLang=en_US