Comparison of value at risk models and expected shortfall models for selected mineral commodities
| dc.contributor.author | Letsoalo, Katlego Masilu | |
| dc.date.accessioned | 2021-05-06T09:31:50Z | |
| dc.date.available | 2021-05-06T09:31:50Z | |
| dc.date.issued | 2020 | |
| dc.description | A research report submitted to the Faculty of Engineering and the Built Environment, University of the Witwatersrand, in partial fulfilment of the requirements for the degree of Master of Science in Engineering, 2020 | en_ZA |
| dc.description.abstract | Risk management is a critical component of modern-day finance with the banking sector having led the developments over the years. Following the impacts of the Global Financial Crisis of mid-2008 on various markets, investors are more concerned with their overall risk exposure including mining companies. Literature presents Value at Risk (VaR) and Expected Shortfall (ES) methods as the most common mechanism used to determine risk exposure. These methods have not been applied extensively in the mining sector despite their popularity in the finance sector. This research study explores the theoretical concept of VaR as a method of risk measurement including the computational considerations and some of the drawbacks of these models. Several studies criticise the ability of VaR to capture risk in a portfolio particularly during a period of risk. Given the drawbacks of VaR, ES is discussed as an alternative method including a comparison of key parameters between the two methods. This research study investigated the most optimal risk exposure evaluation methods for mineral commodities with a focus on coal and gold mining companies listed in South African given the commodities risk exposures and the available Mineral Resources. To apply the risk measurement methods a less volatile period was chosen as a time horizon, data from 2013 to 2019. The study calculated VaR using parametric model through the variance–covariance method, semi-parametric model through Monte Carlo Simulation and non-parametric models using Historical simulation methods. The alternative risk measure was calculated using ES. The outcomes of the VaR models are compared to the ES model to determine the risk measure that captures the possible losses with the highest degree of confidence. The accuracy of the models was tested through a process of backtesting that is discussed through the body of work. The backtesting results show that the ES method performs better than all the VaR methods at different confidence levels and recommends that a 95% confidence level should be used. The comparison of the methods further highlights how the methods perform on volatile companies in comparison to slightly steady companies. The outcomes of VaR and ES methods varies across each company, commodity and confidence level. It was found that the relationship between Monte Carlo Simulation, variance-covariance and historical simulation varies at different confidence levels and companies despite these methods being VaR methods. The possible loss estimates from the historical simulation methods immerge higher than the variance-covariance in some companies while the opposite applied in others. The losses estimated by the ES models were also higher than the VaR in all the companies analysed. The research study recommends that the ES method should be used to determine the possible loses in mining companies as the results of this method performed better than VaR | en_ZA |
| dc.description.librarian | CK2021 | en_ZA |
| dc.faculty | Faculty of Engineering and the Built Environment | en_ZA |
| dc.identifier.uri | https://hdl.handle.net/10539/31130 | |
| dc.language.iso | en | en_ZA |
| dc.school | School of Mining Engineering | en_ZA |
| dc.title | Comparison of value at risk models and expected shortfall models for selected mineral commodities | en_ZA |
| dc.type | Thesis | en_ZA |
Files
License bundle
1 - 1 of 1
No Thumbnail Available
- Name:
- license.txt
- Size:
- 1.71 KB
- Format:
- Item-specific license agreed upon to submission
- Description: