Backtesting Expected Shortfall
(2021) NEKN01 20211Department of Economics
- Abstract
- This paper examines the “back-testability” of Expected Shortfall and if different backtesting
methods produce similar results, with the aim of illuminating whether various markets have
any effect on the methods’ ability to perform. In order to answer this question, the three
backtesting methods – Test 1 and Test 2 by Acerbi & Szekely, and the Approximative
Quantile Test by Emmer et al. – are implemented on three different markets: first, Sweden’s
OMXS30 Index; second, Hong Kong’s Hang Seng Index; third, India’s S&P BSE Sensex
Index. This study, then, splits the sample period between an overall period (2002-2021) and a
sub-sample/the global financial crisis (2007-2009), to elucidate the effect of market stress.
The findings of this... (More) - This paper examines the “back-testability” of Expected Shortfall and if different backtesting
methods produce similar results, with the aim of illuminating whether various markets have
any effect on the methods’ ability to perform. In order to answer this question, the three
backtesting methods – Test 1 and Test 2 by Acerbi & Szekely, and the Approximative
Quantile Test by Emmer et al. – are implemented on three different markets: first, Sweden’s
OMXS30 Index; second, Hong Kong’s Hang Seng Index; third, India’s S&P BSE Sensex
Index. This study, then, splits the sample period between an overall period (2002-2021) and a
sub-sample/the global financial crisis (2007-2009), to elucidate the effect of market stress.
The findings of this study reveal that it is in fact possible to backtest Expected Shortfall,
regardless of index. However, as the backtests’ do not provide any information to why and
how their robustness is affected when carried out over different markets and by market stress,
practitioners/regulators should take extra care when choosing a backtest that “correctly
estimate” Expected Shortfall. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9050006
- author
- Nordenlöw, Karl LU and Halvarson, Tor LU
- supervisor
- organization
- alternative title
- Implementing Backtesting Methods on Different Markets
- course
- NEKN01 20211
- year
- 2021
- type
- H1 - Master's Degree (One Year)
- subject
- keywords
- Value-at-Risk, Expected Shortfall, Backtest, Risk, Market stress, Elicitability, Coherency, Conditional test, Unconditional test, Approximative Quantile test
- language
- English
- id
- 9050006
- date added to LUP
- 2021-07-05 13:26:03
- date last changed
- 2021-07-05 13:26:03
@misc{9050006, abstract = {{This paper examines the “back-testability” of Expected Shortfall and if different backtesting methods produce similar results, with the aim of illuminating whether various markets have any effect on the methods’ ability to perform. In order to answer this question, the three backtesting methods – Test 1 and Test 2 by Acerbi & Szekely, and the Approximative Quantile Test by Emmer et al. – are implemented on three different markets: first, Sweden’s OMXS30 Index; second, Hong Kong’s Hang Seng Index; third, India’s S&P BSE Sensex Index. This study, then, splits the sample period between an overall period (2002-2021) and a sub-sample/the global financial crisis (2007-2009), to elucidate the effect of market stress. The findings of this study reveal that it is in fact possible to backtest Expected Shortfall, regardless of index. However, as the backtests’ do not provide any information to why and how their robustness is affected when carried out over different markets and by market stress, practitioners/regulators should take extra care when choosing a backtest that “correctly estimate” Expected Shortfall.}}, author = {{Nordenlöw, Karl and Halvarson, Tor}}, language = {{eng}}, note = {{Student Paper}}, title = {{Backtesting Expected Shortfall}}, year = {{2021}}, }