Stock market volatility during times of crisis: a comparative analysis of the conditional volatilities of JSE stock indices during the 2007/08 global financial crisis and COVID-19
- Authors: Wang, Zixiao
- Date: 2022-04-06
- Subjects: Stock exchanges , Johannesburg Stock Exchange , Global Financial Crisis, 2008-2009 , COVID-19 (Disease) Economic aspects , Economic forecasting , Stock exchanges and current events , GARCH model
- Language: English
- Type: Academic theses , Master's theses , text
- Identifier: http://hdl.handle.net/10962/284603 , vital:56078
- Description: This research analyses the comparative behaviour of stock market volatility during two crises. The goal of this research is to determine whether assumed cyclical and defensive sectors have either retained or revealed their expected properties during both the Global Financial Crisis (GFC) and COVID-19 by analysing sectoral volatility amid these two crises. Understanding how volatility changes amid crises helps to determine whether the volatility assumptions of diversified investment portfolios for both defensive and cyclical sectors still held given the different causes of each crisis. In turn, this knowledge can assist with risk management and portfolio allocation in stock market investments. The study can also contribute towards the enhancement of financial markets’ resistance against systemic risks through portfolio diversification, and aid government decision-making targeted at tackling the weaknesses of different economic sectors especially in times of overall economic weakness. This research makes use of the GARCH model to analyse a group of daily time series that consists of eleven sectoral indices and one benchmark index, all based on the South African stock markets. These observed series are categorised into two full sample periods, one designated to the Global Financial Crisis (January 2006 to May 2009) and the other for COVID-19 (January 2018 to May 2021). These are further divided into two sets of sub-sample periods, each made up of a pre-crisis and during-crisis. Furthermore, the dummy variables representing the occurrence of structural breaks are inserted into the full sample periods’ conditional variance equations. This is aimed at capturing the asymmetrical impact of the crises themselves on all observed series. Based on the movement of volatility persistency from pre-crisis to during-crisis for both crises, the results show that, firstly, Health Care and Consumer Goods are considered defensive Sectors. Secondly, Banks, Basic Materials, Chemicals, Telecommunications, and Financials are considered cyclical Sectors. Thirdly, Automobiles & Parts, Consumer Services, and Technology are considered indeterminable Sectors due to the inconsistent behaviour of these sectors’ volatility persistency throughout the sub-sample periods of both crises. Overall, according to the average volatility persistency, the observed series for COVID-19’s full sample period are generally less volatile than those of the GFC. However, the sub-sample periods suggest that the observed series for both pre-crisis and during-crisis periods of COVID-19 are more volatile than those same sub-samples of the Global Financial Crisis. Being able to analyse the characteristics of stock market sectors is crucial for risk management and optimal portfolio allocation of stock market investments. This can be achieved through portfolio diversification by investing in a variety of stocks, both cyclical and defensive, and adjusted over time based the needs of stock market investors. Diversified portfolios do not only serve the interests of individual investors, but can also enhance the financial markets’ overall resistance against systemic risks. , Thesis (MCom) -- Faculty of Commerce, Economics and Economic History, 2022
- Full Text:
- Date Issued: 2022-04-06
Wavelet Theory: for Economic & Financial Cycles
- Authors: Mlambo, Farai Fredric
- Date: 2019-12
- Subjects: Wavelets (Mathematics) , Finance -- Mathematical models , Economic forecasting
- Language: English
- Type: Doctoral theses , text
- Identifier: http://hdl.handle.net/10948/49930 , vital:41861
- Description: Cycles - their nature in existence, their implications on human-kind and the study thereof have sparked some important philosophical debates since the very pre-historic days. Notable contributions by famous, genius philosophers, mathematicians, historians and economists such as Pareto, Deulofeu, Danielewski, Kuznets, Kondratiev, Elliot and many others in itself shows how cycles and their study have been deemed important, through the history and process of scientific and philosophical inquiry. Particularly, the explication of Business, Economic and Financial cycles have seen some significant research and policy attention. Nevertheless, most of the methodologies employed in this space are either purely empirical in nature, time series based or the so-called Regime-Switching Markov model popularized in Economics by James Hamilton. In this work, we develop a Statistical, non-linear model fit based on circle geometry which is applicable for the dating of cycles. This study proposes a scalable, smooth and differentiable quarter-circular wavelet basis for the smoothing and dating of business, economic and financial cycles. The dating then necessitates the forecasting of the cyclical patterns in the evolution of business, economic and financial time series. The practical significance of dating and forecasting business and financial cycles cannot be over-emphasized. The use of wavelet decomposition in explaining cycles can be seen as an critical contribution of spectral methods of statistical modelling to finance and economic policy at large. Being a relatively new method, wavelet analysis has seen some great contribution in geophysical modelling. This study endeavours to widen the use and application of frequency-time decomposition to the economic and financial space. Wavelets are localized in both time and frequency, such that there is no loss of the time resolution. The importance of time resolution in dating of cycles is another motivation behind using wavelets. Moreover, the preservation of time resolution in wavelet analysis is a fundamental strength employed in the dating of cycles. , Thesis (DPhil) -- Faculty of Science, Mathematical Statistics, 2019
- Full Text:
- Date Issued: 2019-12