Analyzing the South African Equity Market Volatility and Economic Policy Uncertainty During COVID-19

Abstract

This study examines the dynamics of equity market volatility and economic policy uncertainty (EPU) in South Africa during the COVID-19 pandemic. Using daily return data for sectoral indices and the JSE All Share Index (ALSI) from 1 January 2020 to 31 March 2022, the analysis explores both market-wide and sector-specific volatility responses. Univariate GARCH-family models (GARCH (1,1), E-GARCH, and T-GARCH) are employed to capture volatility clustering, persistence, and asymmetry across sectors. The results show that volatility was highly persistent during the pandemic, with sectoral differences in sensitivity to shocks: Consumer Staples and Financials were particularly reactive to recent news, while Health Care and Basic Materials were more stable. Asymmetric models confirm that market sentiment was predominantly driven by negative news, except in the Energy sector, where positive recovery signals played a stronger role. Correlation analysis further indicates that most sectors were moderately correlated with the ALSI, while Energy and Health Care behaved more independently. In contrast, both the ALSI and sector returns exhibited weak and negative correlations with the South African EPU index, suggesting that uncertainty did not translate directly into equity market declines. Overall, the findings highlight the importance of sectoral heterogeneity in volatility dynamics and suggest that during extreme market events, investors can mitigate downside risk by reallocating portfolios toward more resilient sectors.

1. Introduction

The COVID-19 pandemic triggered the deepest global recession since the Second World War (World Bank Group, 2020). Declared a pandemic on 11 March 2020, the virus had spread to 110 countries with over 118,000 reported cases (World Health Organization, 2020). According to H. K. Baker et al. (2020), COVID-19 impacted financial markets more severely than any global health crisis in recent decades, including the Spanish flu. Governments worldwide imposed state-of-emergencies, strict lockdowns, travel restrictions, and social distancing measures (Sarkodie & Owusu, 2020a, 2020b), disrupting economic activity, trade, and capital flows. In South Africa, already facing economic challenges, the pandemic exacerbated uncertainty and vulnerability, with GDP contracting by 7.10% in 2020 (Erero & Makananisa, 2020).

These developments highlight the importance of understanding the relationship between economic policy uncertainty (EPU) and equity market volatility. Policy uncertainty affects investment, consumption, and saving decisions, which in turn influence asset prices (Pastor & Veronesi, 2012, 2013; Gómez-Méndez & Hansen, 2021; Salisu et al., 2023). High levels of uncertainty may discourage investment, while unpredictable policy responses can amplify volatility, creating a feedback loop that further destabilizes markets (Ghani & Ghani, 2024).

Although numerous studies have examined EPU’s effect on stock markets (Balcilar et al., 2013; Brogaard & Detzel, 2015; Johnson & Lee, 2014; H. Liu et al., 2020; L. Liu & Zhang, 2015; Amengual & Xiu, 2018), most focus on aggregate market indices and developed economies. Very few studies consider sector-specific volatility in emerging markets, leaving a critical gap in understanding how different industries respond to policy uncertainty during crises. Similarly, research using country-level EPU measures for South Africa is scarce, limiting insights for domestic investors and policymakers.

This study addresses these gaps by examining sectoral volatility in the South African equity market during the COVID-19 pandemic using a South African EPU index. Unlike global indices such as the WPUI, this measure incorporates news coverage, expert opinions, and business barriers, providing a locally relevant assessment of policy uncertainty.

This study makes two primary contributions. First, it improves understanding of the relationship between JSE sectoral volatility and South African EPU, offering insights into market behavior under uncertainty. Second, it provides practical implications for investors by identifying sectoral risk patterns and volatility clustering during the pandemic. Specifically, this study has the following aims:

• Analyze sectoral returns and identify volatility clustering.
• Examine sector correlations with the FTSE/JSE All Share Index (ALSI).
• Investigate the relationship between the South African EPU and sectoral returns.
• Estimate conditional volatilities using univariate GARCH models.

The findings aim to support investors and policymakers in risk assessment and decision-making during periods of heightened uncertainty.

The remainder of the study is structured as follows: Section 2 reviews related theoretical and empirical studies; Section 3 presents data and methodology; Section 4 discusses the results; and Section 5 concludes the study.

2. Literature Review

2.1. The Relationship Between Investment and EPU

Businesses and individuals invest with the primary objective of maximizing their return on investment, wherein the probability of failure and success is determined by the current costs, risks, and perceived barriers to entry in that particular industry (World Bank, 2005). Reducing the EPU is critical because a higher EPU severely discourages both domestic and foreign direct investment (Pastor & Veronesi, 2012). A high EPU increases the cost of doing business in an economy because it increases the equity risk premiums, which therefore increases the probability of default of the borrowers and the volatility of the cash flow (Nguyen & Phan, 2017; Kwabi et al., 2022). As a result, lenders often become concerned about the borrower’s ability to repay, which leads to higher risk premiums to offset the increased lending risk. Thus, to hedge themselves against borrowers defaulting as a result of higher EPU, creditors often impose stricter loan covenants granted to small, highly leveraged borrowers and loans where credit spreads are high. Long-term projects and expansion plans (which often increase employment and overall increase economic growth in an economy) may take a halt as interest rates or the cost of borrowing are relatively high. Therefore, the cost of doing business in such a market or country would be expensive, leading to decreased levels of FDI (Nguyen & Phan, 2017; Kwabi et al., 2022).

Emerging research has shown that EPU has a significant effect on investment decisions made by both individuals and corporations and ultimately the broader economic landscape. Most academics agree that higher levels of EPU deter business investment (Gulen & Ion, 2016; Julio & Yook, 2012). Since EPU substantially affects a firm’s investment behavior, it may lead to higher-than-expected expenditures, lower productivity, and decreased long-term investment (Hartman, 1972). When an economy or a particular market is experiencing heightened levels of EPU, entrepreneurs and businesses are often reluctant to increase investment, particularly in emerging markets (Huang et al., 2022).

According to S. R. Baker et al. (2016), a higher EPU reduces investment, employment, and expenditure by households and businesses. The EPU also hinders economic recovery because it is difficult for economic forecasters and participants to forecast with precision when and/or how the government will change its economic policies. The reason for this slowdown could be that EPU makes investors feel more risk-averse and negatively impacts investor sentiment, which can affect investment business (Liao & Mehdian, 2016).

As a result of a limited understanding of the implications of the pandemic on financial markets, many research studies seek to gain insight into the reasons behind the drastic volatility in the equity market during the peak of the COVID-19 outbreak (Ashraf, 2020; H. K. Baker et al., 2020; Gormsen & Koijen, 2020). Bialkowski et al. (2022) investigated the factors that impact the relationship between EPU and implied volatility and also empirically tested whether the link between the two affects equity market performance. According to the theoretical model of the study, fluctuations in investor opinions lead to overtrading and increased volatility in the stock market. The main findings of the research showed that there is a positive relationship between EPU and market volatility and that it varies with different levels of stock market performance. Bialkowski et al. (2022) further put forward that the relationship between EPU and market volatility gets weaker when there is a high level of opinion divergence between investors, coupled with exceptional stock market performance.

Paule-Vianez et al. (2023) explored whether the impact of the EPU on various stock types varies based on which stocks are the most or least profitable. Essentially, the objective of the study was to analyze how EPU impacts growth/value and small/large cap stock returns in bullish and bearish markets. Building on the approaches used in previous research, the study used quantile regression to assess the impact of EPU. The study employed heteroskedasticity correction and ordinary least squares to evaluate the reliability of the results. The research concluded that the economic cycle, particularly for growth and small stocks, moderates the negative sensitivity pattern of EPU on stock returns, with a stronger effect during recessions. Only the most profitable stocks are spared from the negative effects of the economic cycle on the EPU. With little impact felt in the bottom tail of the stock return distribution, this moderation becomes ineffective as stock prices increase.

Research on economic policy uncertainty (EPU) has increasingly emphasized its influence on investment decisions and portfolio management across different contexts. For example, Sun and Li (2025) investigate the relationship between EPU and household investment portfolios in China, a country characterized by a dynamic economic landscape and regulatory volatility. They find that heightened EPU leads households to adopt more conservative investment strategies, shifting towards cash and fixed-income securities while reducing exposure to equities and real estate. Moreover, the study reveals that EPU undermines portfolio diversification by increasing correlations between asset classes, thereby diminishing the effectiveness of risk mitigation strategies. These effects are particularly pronounced during periods of elevated economic volatility, while more stable environments encourage greater diversification.

Importantly, Sun and Li (2025) frame these behavioral shifts within the lens of behavioral finance, suggesting that fear and uncertainty drive biases that encourage overly conservative portfolio adjustments. The study also highlights the policy dimension, advocating for greater predictability and stability in economic policy to bolster investor confidence. Although their analysis is focused on household-level investment in China, the findings offer valuable parallels for understanding investor responses in emerging markets such as South Africa. In particular, the tendency for EPU to magnify volatility and weaken diversification strategies provides a useful comparative lens for examining the behavior of sectoral equity portfolios on the JSE during the COVID-19 period.

In line with these findings, Kyriazis et al. (2025) extend the discourse by examining how different dimensions of policy uncertainty—monetary, regulatory, and national security—interact to influence equity markets, using the S&P 500 as a benchmark. Their econometric analysis, which employs models such as VAR and GARCH, reveals that these uncertainties are significantly correlated with stock market volatility, with national security concerns exerting the strongest impact, particularly during geopolitical crises. Importantly, the study highlights the temporal dynamics of investor responses, noting that the effects of policy uncertainty materialize with lags and often precede major market downturns. The authors emphasize that transparency and clarity in monetary and regulatory communication are critical in mitigating the adverse effects of uncertainty on investor sentiment and market stability.

2.2. COVID-19 and Stock Markets

Vengesai (2022) examined the COVID-19 pandemic as a shock, impacting the sector returns on the JSE. By employing a Pooled Mean Group estimator on the daily sector returns to estimate the Autoregressive Distributed Lag (ARDL) model, the research study sampled the ten JSE sectors. The sample period of the study was from March 2020 to February 2022. The study found that in the short run, although most of the sectors were negatively impacted by the pandemic, the impact on the sector returns varied, as some showed no response, while others showed positive returns. The study further established a negative relationship between the returns of the Consumer Services, Industrials, Technology, Healthcare, Financials, and the Consumer Discretionary sectors and the COVID-19 pandemic, implying that these sectors were significantly impacted adversely. In contrast, the study found the Telecommunications and Resources sector had a positive relationship with the pandemic over the sampled period. The study further revealed that the energy sector was more resilient to the impact of the pandemic.

Mokoena and Nomlala (2022) investigated the relationship between the pandemic and the South African stock markets. The article aimed to investigate how the financial markets of South Africa reacted to COVID-19 and how the different market indices responded to the outbreak. The findings of the study support the findings made by Kusumahadi and Permana (2021), wherein the authors found that the pandemic had a major effect on the stock market. The study by Mokoena and Nomlala (2022) further claimed that the prolonged government lockdowns had a significant impact on stock prices and also brought more uncertainty to investors. Moreover, the study noted that the volatility measures not only peaked during the pandemic, but also became more persistent. The study also found that the pandemic has caused extended periods of high and long-term financial volatility in various markets.

In a study investigating how company performance was impacted by COVID-19, using companies from different sectors on the JSE, Muthu and Wesson (2023) noted that the pandemic negatively affected the performance of companies across the sectors. The study analyzed JSE-listed companies using monthly secondary data on a sample period of 2014–2020. The results were consistent with studies by Vasileiou (2021), Tan et al. (2022), and Kharabsheh et al. (2022). Although the studies focused on different markets in different countries and continents, the findings are that the pandemic negatively impacted stock markets.

Vasileiou (2021) conducted a study on the effectiveness of the stock markets in the United States of America (USA) during the COVID-19 pandemic. The study analyzed the performance of the stock market by employing a fundamental financial analysis approach. The study demonstrated that during the period examined, the world’s largest stock market was not operating efficiently. In the analysis, the study further showed that in some instances during the pandemic, the reaction was slow and inconsistent with the financial theory. The study also noted that the pandemic negatively impacted the performance of the US stock market.

In the analysis of the repercussions of the COVID-19 pandemic on stock market returns and volatility within the G7 countries using the TVP-VAR-SV estimation, Tan et al. (2022) found that the pandemic significantly decreased stock returns and increased volatility in the G7 countries. However, the study noted that the decrease is transient and that the stock market recovers more quickly in the later period (Tan et al., 2022). The study findings also clearly revealed that the spread of the pandemic negatively affected the stock market indices in the G7 countries. This research finding is also consistent with the findings of Ahmar and Del Val (2020) and Ashraf (2020). The study concluded that the global spread of COVID-19 has had a significant negative impact on investor sentiment, which then led to global investor panic. As a result, investor expectations and trading behavior changed, which then decreased stock returns. Fiscal stimulus and/or economic aid programs were not sufficient to mitigate this impact (Tan et al., 2022).

Kharabsheh et al. (2022) conducted a study in which they examined how the COVID-19 pandemic affected the main Amman Stock Exchange (ASE) index and its sub-indices, that is, the indices of the four main sectors being financial, industrial, services, and insurance. The study employed a quarterly panel data analysis. The results of the study showed that the impact of the daily accumulation of confirmed COVID-19 cases was negative and statistically significant on the daily returns of the main market index, as well as the indices for the financial and service sectors. This result showed that the financial sector suffered the largest losses during the pandemic, followed by the industrial and service sectors, indicating a statistically weak and negative effect on the industry sector index. The only index that showed a positive impact was the insurance industry index; however, it was not statistically significant. The findings also show that COVID-19 has a negative, strong, and statistically significant impact on the returns of a company’s stock. The data indicates that larger companies and those with more capital on hand performed better during the pandemic due to their stronger capital resources (Kharabsheh et al., 2022).

In summary, these studies present diverse empirical evidence on how the COVID-19 pandemic affected financial markets by employing different methodologies and focusing on various sectors. According to the empirical evidence provided, the pandemic negatively affected the stock markets and consequently induced volatility within the markets. Several factors, including an increase in COVID-19 cases and fatalities, as well as government policies, led to complex market responses and increased policy uncertainty. Broadly, existing research highlights the intricate interactions among various factors that shaped market dynamics during the pandemic. It advocates for adaptive strategies and efficient policy formulation to guarantee stability and investor protection. However, there is less literature focusing on the SA equity market sectors and the role played by the economic policy uncertainty brought about the pandemic, highlighting the vast gap in research.

3. Data and Methodology

3.1. Data and Sources

This study uses secondary data obtained from IRESS BFA, an online platform providing fundamental research and stock market data. Daily closing prices for South African equity market indices were collected for the period 1 January 2020 to 31 March 2022, encompassing the onset of the COVID-19 pandemic, lockdowns, subsequent waves, the start of vaccine rollouts, and early stages of economic recovery. This period allows for a comprehensive assessment of market and economic responses.

Daily prices were used to calculate daily returns, which normalize the data and preserve volatility patterns during crisis periods. Returns were calculated using Microsoft Excel, enabling comparison across sectors and relative to the FTSE/JSE All Share Index (ALSI), which serves as a benchmark representing the South African market.

The Economic Policy Uncertainty (EPU) Index was sourced from the North-West University School of Business and Governance and the School of Economics, Potchefstroom Campus. This index combines three components: (i) coverage of policy uncertainty in news, (ii) opinions from a panel of prominent economists, and (iii) input from the University of Stellenbosch’s Bureau for Economic Research regarding barriers to business faced by manufacturers (Parsons & Krugell, 2022). A preliminary threshold of 50 was established: values below 49 indicate lower policy uncertainty, while values above 50 indicate higher uncertainty. As the index is published quarterly, linear interpolation was applied to generate daily EPU data for alignment with sectoral returns.

Following prior research (Adekoya & Nti, 2020; Mazur et al., 2021; Ramterath, 2024), the study examines multiple equity sectors in the South African stock market (see

Table 1. Index description.

Name of Index	Index Code
Basic materials	JI0055
Industrials	JI0050
Health Care	JI0020
Telecommunications	JI0015
Financials	JI0030
Consumer Discretionary	JI0040
Technology	JI0010
Consumer Staples	JI0045
Energy	JI0060
FTSE/JSE ALSI	J203

for sector codes). The FTSE/JSE ALSI is used as a proxy for the overall market, covering approximately 99% of companies listed on the main board by market capitalization.

All descriptive and inferential statistical analyses were conducted using EViews 14, which is particularly suitable for time series analysis and supports data interpolation, offering advantages over other software such as SPSS and STATA (Masimba & Shadreck, 2021).

3.2. Methodology

This study includes various statistical techniques such as Descriptive Statistics, Stationarity Test, ARCH effect test, GARCH (1,1), E-GARCH, and the T-GARCH.

To calculate returns that consider compounding, log returns were used. Log returns are additive over time, and they scale linearly, making them easily comparable over different periods of different lengths. All the indices are represented as daily returns and were determined by taking the current stock price divided by the stock price of the previous day:

$$y\mathsf{¡}=ln(P_{\mathrm{t}}/P_{\mathrm{t}-1})$$

(1)

where

• y¡ = the compounded daily returns
• P_t = closing index price at day t
• P_t−1 = closing index price at day t − 1

The total returns in the assessment are inclusive of both reinvested dividends and capital gains over the specified period. To tackle one of the empirical objectives, the daily closing prices from the JSE sector indices were compounded accordingly to obtain daily mean returns. The reason behind the use of daily price data is that data with a lower frequency might not capture and may distort the changes that occur daily. However, one of the issues associated with using daily data is the distortions caused by non-trading days, typically on public holidays. To circumvent this issue, Chinzara and Aziakpono (2009) proposed the complete removal of non-trading days, which is also what this study does. Removing these days should not significantly affect the empirical findings, given the sample size. The study will also follow the same approach of removing non-trading days.

3.2.1. Descriptive Statistics

Prior to the analysis of volatility models, several descriptive statistical properties were performed because they give a concise summary of large datasets, making complex data easier to understand. These include the mean and median (as both are measures of central tendency); standard deviation, kurtosis, skewness (as they are measures of variability), and the Jarque–Bera test to confirm if the results are normally distributed will be discussed.

3.2.2. Data Stationarity Test

A time series is stationary if its mean, variance, and auto-covariance remain constant over time (Momin & Chavan, 2018). Testing for unit roots helps determine the order of integration in a dataset. Stationary series are easier to model and forecast, while non-stationary data can lead to spurious regressions with inflated R2R^2R2 and misleading t-statistics (Gujarati, 2003).

To address this, the study applies two tests:

• Augmented Dickey–Fuller (ADF) test detects the presence of a unit root.
• Kwiatkowski–Phillips–Schmidt–Shin (KPSS) test directly tests for stationarity (Sjösten, 2022).

Hypotheses:

• ADF:

  ○

  H₀: series has a unit root (non-stationary).

  ○

  H₁: series is stationary.

  Reject H₀ if p-value < 0.05.

• KPSS:

  ○

  H₀: series is stationary.

  ○

  H₁: series is non-stationary.

  Reject H₀ if test statistic > critical value at 5%.

3.2.3. ARCH Effect Test

Before estimating volatility models, it is necessary to test for ARCH effects in the residuals of the mean equation. Engle (1982) introduced the ARCH model to capture time-varying volatility in financial data, where high-volatility periods cluster together, as do low-volatility periods. This behavior violates the assumption of constant variance and justifies the use of GARCH-type models.

The study employs the ARCH Lagrange Multiplier (LM) test to check for conditional heteroskedasticity.

Hypotheses:

• H₀: No ARCH effect.
• H₁: Presence of ARCH effects.

3.2.4. Mean Equation

There is a requirement to have the specification of an appropriate mean equation before conducting a GARCH analysis. The mean equation is specified as follows:

$${R_t=\mathsf{\mu }+\epsilon }_t$$

(2)

where

• R_t = conditional mean
• μ = a constant
• ε_t = error term that is normally distributed with zero mean

3.2.5. Univariate GARCH Models

As the study seeks to measure the impact of COVID-19 on the conditional volatilities and returns of different sectors, the former will be treated as a proxy for risk. Conditional volatilities of sectors will be analyzed using univariate ARCH/GARCH models, including asymmetric GARCH (1,1), E-GARCH, and T-GARCH models. Univariate GARCH models were preferred over Multivariate GARCH models because, although the latter could better capture sectoral interdependencies, Univariate GARCH models are less complex to estimate and interpret (Virbickaite et al., 2015). The rationale behind using GARCH models is that GARCH is extremely useful in determining risk and expected returns for assets that have clustered periods of return volatility (Ali et al., 2022). Three options are offered by EViews for the diagnostic procedure: the Student’s t, the normal (Gaussian), and the generalized error distribution.

Bollerslev (1986) later produced the GARCH model to address the limitations of the ARCH model. Subsequently, several extensions to the GARCH model were created. The models include the Threshold GARCH (T-GARCH) model proposed by Zakoian (1994) and Glosten et al. (1993), and the Exponential-GARCH (E-GARCH) model proposed by Nelson (1991). E-GARCH is important because its logarithmic formulation enables the parameters’ positive limitations to be relaxed. This GARCH family of models will be used in this study. The null hypothesis will be rejected if the test statistic is significant, and a conclusion will be drawn that there is evidence of an ARCH effect in the data (the series is heteroskedastic), which will make the GARCH model more appropriate.

GARCH (1,1)

$$h_t^2=\omega +{\alpha }_1{\epsilon }_{t-1}^2+{\beta }_1{h}_{t-1}^2$$

(3)

wherein the parameters include conditional variance squared at time (t) $h_t^2$, residual ε, constant $\omega$, the ARCH coefficient ${\alpha }_1,$ and the GARCH coefficient ${\beta }_1$.

Both α and β are empirical parameters determined by maximum likelihood estimation. Volatility is captured by (α + β). A series is said to have long-term memory if (α + β) > 1, indicating that volatility in that specific sector is very persistent. On the other hand, if (α + β) < 1, the series exhibits short- to medium-term memory, meaning that volatility only lasts for a short while in the future.

E-GARCH

$$\mathrm{l}\mathrm{n}\left(h_t^2\right)=\omega +{\alpha }_1{\displaystyle \frac{\left|{\epsilon }_{t-1}\right|+{\gamma }_1{\epsilon }_{t-1}}{h_{t-1}}}+{\beta }_1\mathrm{l}\mathrm{n}\left(h_{t-1}^2\right)$$

(4)

The parameters $h_t^2$, ε, $\omega$, ${\alpha }_1,$ and ${\beta }_1$ are interpreted the same as in the GARCH (1,1) model, with the inclusion of the leverage parameter γ. E-GARCH models volatility as the log of the variance, which makes it superior to other GARCH models because it removes the need to impose non-negativity constraints on the model parameters artificially (Nelson, 1991). Thus, an E-GARCH model was estimated to deal with the issue of non-negativity restrictions and capture the asymmetric impact in the return series.

T-GARCH

$$h_t=\omega +{\alpha }_1{h}_{t-1}\left(\left|{\epsilon }_{t-1}\right|-{\lambda }_1{\epsilon }_{t-1}\right)+{\beta }_1{h}_{t-1}$$

(5)

where ${\lambda }_1$ is the leverage parameter and $\mid {\lambda }_1\mid \le 1$. When there is bad or good news in financial markets, stocks tend to reprice rapidly, and volatility increases or decreases depending on the news. As a result of these realities, financial econometricians have proposed ways to capture the influence of this news by putting forward the T-GARCH model. The T-GARCH model was proposed by Glosten et al. (1993), extending from the GARCH model to capture the asymmetries concerning negative and positive shocks. The T-GARCH determines whether there is a statistically significant difference when the shocks are negative by adding a dummy variable to the variance. Thus, the T-GARCH model will aim to capture the asymmetric impacts of positive and negative shocks on volatility across different sectors and the broader equity market.

Time-series regression models are perfect for examining how JSE volatility varies over time and how it is influenced by various variables, such as economic policy uncertainty. A time-series model will be used to assess how JSE volatility is influenced by economic policy uncertainty at a particular time.

$$y_t=\omega +{\mathsf{\beta }}_1{x}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{t}}$$

(6)

where $y_t$ is the dependent variable at time (t), $\omega$ is the constant parameter, β₁ represents the coefficient showing the sensitivity of the independent variable, and $x_{\mathrm{t}}$ is the independent variable at time and the error term is ε_t.

4. Results and Discussion

This section is divided into the following sub-sections: Section 4.1 A graphical representation and analysis of the returns of the sectors, depicting the trends. Section 4.2. Descriptive statistics will be discussed to provide a summary and an understanding of the dataset. Section 4.3. An assessment of the correlation between the sector index and the market index. Section 4.4. An analysis of the correlation between the SA equity market and policy uncertainty. Section 4.5. The stationarity of the data will be detected by making use of the ADF and KPSS tests. Section 4.6. The conditional volatilities of the sectors and the market by performing ARCH-LM tests and GARCH models.

4.1. Graphical Representation of the Daily Returns of the Sector

The graphs in Figure 1 showing the daily returns from 1 January 2020 to 31 March 2022, which include the nine sectors and the JSE ALSI, shed light on considerably high volatility experienced during the COVID-19 period. The trend of returns of each sector is similar to the overall market, i.e., each sector experienced sharp declines and increases in volatility. Moreover, each graph under Figure 1 exhibits evidence of volatility clustering, indicating that volatility in the present will impact volatility in the future, and every return series appears to be mean-reverting, indicating stationarity. This demonstrates that our volatility estimation using the GARCH models is reliable.

4.2. Descriptive Statistics

Table 2. Descriptive statistics of the sectors during the COVID-19 pandemic.

		Basic	Consumer	Consumer
	ALSI	Materials	Discretionary	Staples	Energy	Financials	Healthcare	Industrials	Techn	Telecoms
Mean	0.00	0.001	0.000	0.000	0.003	0.000	0.000	−0.000	−0.000	0.001
Median	0.001	0.001	0.001	0.001	0.000	0.001	0.000	0.000	0.000	0.001
Maximum	0.073	0.121	0.074	0.072	0.327	0.075	0.057	0.076	0.187	0.105
Minimum	−0.102	−0.156	−0.093	−0.100	−0.302	−0.131	−0.111	−0.097	−0.130	−0.117
Std. Dev.	0.015	0.023	0.018	0.014	0.043	0.020	0.018	0.018	0.026	0.024
Skewness	−1.071	−0.608	−0.219	−0.359	0.750	−1.029	−0.317	−0.417	0.120	−0.378
Kurtosis	11.797	10.064	5.954	10.175	15.982	10.959	6.626	7.221	9.159	7.439
Jarque–Bera	1926.460	1207.283	209.5140	1221.831	4013.218	1582.478	318.4211	435.2044	893.0064	476.3786
Probability	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
Observations	562	562	562	562	562	562	562	562	562	562

provides a summary of the descriptive statistics of the daily returns of all the sectors during the chosen COVID-19 sample period. It can be deduced from Table 2 that during the COVID-19 period, there were positive mean returns for all sectors, excluding the Industrials and Technology sectors, indicating that the South African stock market withstood the economic storm caused by the pandemic and that the sectors maintained positive returns over time. Additionally, the positive mean indicates that the daily prices of the indices increased over the observed period. On the contrary, although very small, the negative mean returns for the industrials and technology sectors point out that, on average, the two sectors had negative returns, which led to an overall loss again. This generally means the sectors were not performing well due to the effects of the pandemic and that they did not adapt to the shock that resulted from the pandemic.

When a sector’s median return is positive, it indicates that over 50% of its companies or data points had positive returns during the period under analysis. Table 2 exhibits that over 50% of companies in all the sectors had positive returns during the COVID-19 pandemic period.

Table 2 reveals that the energy sector was the most volatile, with a value of 0.043, while the consumer staples sector had the lowest standard deviation of 0.014. Thus, it can be deduced that the energy sector is likely to have experienced larger fluctuations in stock prices relative to the other sectors. The high standard deviation can be a result of travel restrictions and lockdowns that were imposed in different countries, which decreased the demand for energy products, particularly oil. Conversely, the low standard deviation for consumer staples implies that the sector was less volatile and relatively stable, which may be because companies in this sector trade in essential goods and services, including food and drinks, among others.

From Table 2, it can be deduced that the energy and technology sectors were the only two indices that exhibited positive skewness. This means that the returns are not symmetric, and instead, the sectors have small losses and few large gains. The positive skewness may mean that energy prices and stocks had periodic, notable recoveries despite overall poor performance, resulting in a few large gains. Additionally, some investors may have taken speculative bets on a recovery as oil prices plummeted. When these recoveries happened, they produced significant returns in an otherwise struggling industry, which further skewed the distribution positively. Similarly, cloud computing, remote work, and e-commerce tech companies benefited from the pandemic’s acceleration of digital technology use. As a result, several tech stocks saw significant gains, which caused the overall returns of the industry to skew positively. All the other sectors, including the JSE ALSI, deviated from zero and showed negative skewness during the COVID-19 pandemic. Financial returns with negative skewness are more likely to have extreme negative returns, which may indicate more downside risk.

Kurtosis is a statistical metric that measures the tailedness of a distribution, as this is often where the outliers occur (DeCarlo, 1997). The kurtosis values for the JSE and all the sectors were observed to be leptokurtic, meaning that they did not follow a normal distribution but rather were fat-tailed during the pandemic. This means that the returns of the indices had more pronounced peaks or fluctuations. Finally, the Jarque–Bera test is a goodness-of-fit test that determines if the skewness and kurtosis of sample data are comparable to those of a normal distribution (Koizumi et al., 2009). In contrast to the EMH, the high values of the Jarque–Bera statistic for all sectors offered additional evidence that the null hypothesis of normality could be rejected for all indices.

4.3. Correlation Between Each Sector and the Market

Portfolio diversification is important for every investor to cushion negative returns, and correlation is a metric often used when building diverse portfolios (H. K. Baker et al., 2020). Thus, to spread risk, investors often use correlation to identify sectors or stocks that have negative or low correlations with each other so that when one stock is underperforming, the other stocks are not impacted. Figure 2 reveals the relationship between different equity sectors and the JSE ALSI from 1 January 2020 to 31 March 2022. It can be deduced from Figure 2 that all sectors have a positive correlation with the JSE ALSI, being a proxy of the entire equity market. Therefore, this implies that the performance of the sectors has a significant influence on the performance of the ALSI.

Notably, consumer staples, industrials, and telecommunications have a moderately positive correlation with the JSE ALSI. These sectors have an impact on the market, but not as much as the sectors that are more strongly correlated (basic materials, consumer discretionary, and financials). In contrast, the energy sector exhibits the lowest correlation with the JSE ALSI, suggesting that the sector performs largely independently of the overall market. Thus, it could be a strong option for portfolio diversification because it could not be as impacted by changes in the market as a whole.

Figure 2 also exhibits a low correlation between the health care sector and the JSE ALSI, which indicates that the former often moved independently of broader market movements during the pandemic. Regulation changes and public health crises may be some of the driving forces behind the healthcare sector. Since health care services and goods are necessities or non-cyclical, they are often in constant demand regardless of the state of the economy. Thus, the sector often performs better during economic downturns. Similarly, the technology sector also exhibits a low correlation with the overall market, which suggests that the sector is heavily influenced by industry-specific dynamics like innovation cycles, technology adoption, and consumer demand for tech products, and less by the broader market dynamics.

The correlation analysis is insightful, showing that the energy sector had the lowest correlation with the overall market, making it a useful diversification tool. Most sectors showed a weak negative correlation with policy uncertainty, contradicting global studies. In order to create a diversified portfolio, investors must take these correlations into consideration. With their low correlations to the ALSI, sectors like energy can aid in lowering the overall portfolio risk. On the other hand, although they might provide less diversity, sectors with strong correlations to the ALSI can also be used to support market-aligned strategies.

4.4. SA Policy Uncertainty and SA Equity Market Correlation

Table 3. Correlation coefficients between the SA policy uncertainty and SA equity market.

Sector	SA EPU
Basic material	−0.025149
Technology	−0.013227
Health care	−0.040865
Financials	−0.041454
Telecoms	−0.040437
Consumer staples	−0.029015
Consumer discretionary	−0.013562
Energy	0.018880
Industrials	−0.081127
JSE ALSI	−0.051096

exhibits the correlation coefficients of different equity sectors and the SA policy uncertainty index. With the exception of the energy sector, the returns of all sectors and the broader market have a weak and negative correlation with the uncertainty index. Therefore, SA’s equity market as a whole is negatively impacted by policy uncertainty; however, the impact is minimal. These results are in contradiction with the findings from the study by Bialkowski et al. (2022), which found a positive relationship between EPU and market volatility. The weak but positive correlation between the SA policy uncertainty and the energy sector means that an increase in policy uncertainty might bolster the sector’s returns. This might be through the government’s interventions in the form of subsidies to oil prices, for instance.

4.5. Index Stationarity

Table 4. Sector stationarity tests using daily returns.

	ADF	KPSS
	At Level	At Level
Basic Materials	−24.09621	0.058225
Consumer Discretionary	−23.06556	0.265727
Health Care	−25.01165	0.147341
Telecommunications	−24.63220	0.254901
Financials	−22.85420	0.409503
Technology	−22.95540	0.254901
Consumer Staples	−15.10000	0.050104
Energy	−26.97711	0.147329
Industrials	−24.03337	0.267713
JSE ALSI	−24.13508	0.082527

Notes: The ADF 5% critical value = −2.866455 and the KPSS 5% critical value = 0.463000.

exhibits the results from the ADF and KPSS tests of the indices. The ADF test indicates that every observed series or sector does not contain a unit root or is stationary at the 5% significance level during the COVID-19 sample period; thus, the null hypothesis of the presence of a unit root is rejected at the level terms. The KPSS test, similar to the ADF test, suggests that all the observed sectors are stationary at the 5% significance level. As a result, all observed series have the same order of integration and are stationary at the level.

4.5.1. Findings from the ARCH-LM Tests

Prior to estimating a volatility model in forecasting, it is necessary to determine whether volatility clusters are present in the dataset. In order to make sure the dataset is suitable for the model, Engle (1982) stressed the importance of looking for volatility clusters (AbdElaal, 2011). The mean equation given in Equation (2) was estimated for each sector. Subsequently, the results were tested for ARCH effects using the OLS method to see if volatility had been sufficiently captured. The alternative hypothesis in the ARCH effect test is that there are ARCH effects, while the null hypothesis is that there are no ARCH effects. In the event that the Chi-squared p-value is less than 0.05, the null hypothesis is rejected, and it is concluded that there are ARCH effects in the observed series.

Table 5. ARCH-LM test during the COVID-19 period using daily returns.

	ARCH-LM	p-Value
Basic Materials	7.5819	0.0059 ***
Consumer Discretionary	2.2414	0.1344
Consumer Staples	8.0216	0.0046 ***
Financials	39.6608	0.0000 ***
Health Care	29.2613	0.0000 ***
Industrials	34.5164	0.0000 ***
Energy	64.6798	0.0000 ***
Technology	3.0564	0.0804 *
Telecommunications	3.6816	0.0550 *
JSE ALSI	13.5818	0.0002 ***

Note: The ARCH-LM test column presents the numerical values of the observed R-Squared. Note: *** 1% significance level, * 10% significance level.

presents the ARCH LM F-statistics derived from the mean equations of the respective sectors.

From the LM statistics, the majority of the sectors and the overall market index show evidence of ARCH effects, suggesting that the volatility in those sectors was not sufficiently captured by the mean equations. The evidence of ARCH effects means that the volatilities of the market and the sectors vary over time, and in the event of a market shock, volatility in the current period will spill over to the next period. The only sector without any indication of ARCH effects was consumer discretionary; technology and telecommunications were significant at a 10% confidence level. Therefore, the estimation of GARCH models for consumer discretionary is no longer appropriate because the volatility of the sector is sufficiently captured by the mean equation.

Since the remaining sectors and the JSE ALSI exhibit significant heteroskedasticity, this suggests that the prerequisites for employing the GARCH models were satisfied. Thus, the GARCH models were employed since they enable the modelling of time-varying volatility and can shed light on the persistence and clustering of volatility in the sectors.

4.5.2. Univariate GARCH Models

GARCH (1,1)

Table 6. GARCH (1,1) results.

	ω	α	β	α + β
Basic Materials	0.000025 **	0.090751 ***	0.854428 ***	0.94518
Consumer Staples	0.000008 ***	0.193578 ***	0.787735 ***	0.98131
Energy	0.000010 ***	0.130729 ***	0.882599 ***	1.01333
Financials	0.000016 ***	0.195712 ***	0.767979 ***	0.963691
Health Care	0.000018 **	0.098273 ***	0.846536 ***	0.94481
Industrials	0.000007 ***	0.113969 ***	0.866647 ***	0.98062
Technology	0.000048 ***	0.156479 ***	0.780299 ***	0.93678
Telecoms	0.000041 ***	0.112121 ***	0.811410 ***	0.92353
JSE ALSI	0.000015 ***	0.161199 ***	0.753829 ***	0.91503

Note: *** 1% significance level, ** 5% significance level.

reports the GARCH (1,1) estimates across the JSE ALSI and its main sectors. The constant term (ω) is positive and statistically significant for all indices, though relatively small, suggesting that in tranquil periods with no new information, baseline volatility levels remain low. Technology exhibits the highest ω, reflecting that even for absent shocks, this sector is inherently more volatile, likely due to its sensitivity to innovation cycles, interest rates, and speculative trading.

The news impact term (α) is significant across sectors and highlights how strongly volatility responds to recent shocks. Consumer Staples and Financials show the highest α, indicating strong short-term reactivity to news. Interestingly, Consumer Staples, often considered defensive, were highly sensitive during COVID-19, possibly due to supply chain disruptions and panic buying. Conversely, Health Care and Basic Materials exhibit relatively low α values, suggesting that shocks affect them less severely and their volatility stabilizes more quickly.

The persistence term (β) is generally high, reflecting that volatility is influenced by past shocks and remains clustered. However, Financials and the JSE ALSI show the lowest β values, suggesting that despite being reactive to recent shocks (high α), they revert more quickly to calmer volatility conditions. This faster mean reversion is consistent with broad market recovery mechanisms and financial market adaptability during crises.

Volatility persistence (α + β) is close to 1 for most indices, reflecting prolonged volatility clustering during COVID-19. Notably, the Energy sector has α + β > 1, implying explosive volatility persistence. This aligns with the collapse in global energy demand during lockdowns and the subsequent oil price crash.

Economic intuition: The GARCH (1,1) results show that volatility across the JSE sectors during the pandemic was highly persistent, though sectoral sensitivities differed. Technology and Energy were most inherently volatile, Consumer Staples unexpectedly mirrored shock-sensitive behavior, while Financials and the ALSI showed resilience by reverting more quickly to calmer volatility states.

E-GARCH

The E-GARCH results in

Table 7. E-GARCH results.

	ω	α	β	α + β	γ
Basic Materials	−0.481489 ***	0.141737 ***	0.951430 ***	1.093167	−0.107984 ***
Consumer Staples	−0.218682 ***	0.115144 ***	0.984967 ***	1.100111	−0.111728 ***
Energy	−0.608511 ***	0.240747 ***	0.930090 ***	1.170837	0.032438 **
Financials	−0.519095 ***	0.265631 ***	0.961146 ***	1.226777	−0.114963 ***
Technology	−0.729745 ***	0.266426 ***	0.928291 ***	1.194717	−0.080595 **
Telecoms	−0.208526 ***	0.027696	0.974926 ***	1.002622	−0.156032 ***
Health Care	−0.559257 ***	0.133839 ***	0.947862 ***	1.121701	−0.107327 ***
Industrials	−0.196258 ***	0.125266 ***	0.987586 ***	1.112852	−0.088025 ***
JSE ALSI	−0.621794 ***	0.156530 ***	0.942765 ***	1.099295	−0.194654 ***

Note: ** 5% significance level, *** 1% significance level.

provide additional insights by capturing asymmetries. Negative ω values, common in E-GARCH due to the log specification, reflect low average volatility in normal conditions. However, this model highlights how shocks of any size can rapidly amplify volatility.

The α terms indicate sensitivity to new information. Financials and Technology again show strong responsiveness, consistent with their macroeconomic and innovation-driven exposures. Telecoms, in contrast, display a statistically insignificant α, reinforcing their defensive and relatively shock-immune character.

The β values are uniformly high, showing strong persistence, particularly in Consumer Staples and Industrials, where supply chain bottlenecks and factory closures prolonged uncertainty.

The γ asymmetry parameter reveals a dominant leverage effect. All sectors except Energy show negative γ values, meaning that negative shocks (bad news) intensified volatility more than positive news. This is consistent with the pandemic context, where lockdowns, rising cases, and economic restrictions drove investor fear. Energy, however, exhibits a positive γ: here, recovery signals (stimulus, post-pandemic demand rebound, and the energy transition narrative) mattered more than the initial collapse.

Economic intuition: The E-GARCH results highlight the asymmetric nature of volatility in South Africa’s equity market during COVID-19, with most sectors more sensitive to bad news. Energy was an outlier, as positive recovery dynamics outweighed negative shocks, underscoring sector-specific heterogeneity in investor sentiment.

T-GARCH

Table 8. T-GARCH results.

	ω	α	β	α + β	λ
Basic Materials	0.000020 *	−0.011892	0.905130 *	0.893238	0.122603 *
Consumer Staples	0.000005 **	0.045294 ***	0.869276 *	0.914570	0.138191 *
Energy	0.000011 *	0.156624 *	0.878011 *	1.035781	−0.045726 **
Telecoms	0.000018 *	−0.048663 *	0.958787 *	0.910124	0.121325 *
Financials	0.000013 *	0.081308 **	0.808293 *	0.910124	0.154399 *
Health Care	0.000016 **	−0.005360	0.867431 *	0.862071	0.183278 *
Industrials	0.000004 **	0.039429 ***	0.911000 *	0.950429	0.080775 **
Technology	0.000042 **	0.038974	0.831004 *	0.869978	0.136015 **
JSE ALSI	0.000012 *	−0.034930	0.836943 *	0.802013	0.253775 *

Note: * 10% significance level, ** 5% significance level, *** 1% significance level.

presents the T-GARCH estimates, which also account for asymmetry but through leverage effects (λ).

The α coefficients are mixed: in sectors like Basic Materials and Health Care, current shocks had little or even dampened effects on volatility, potentially reflecting stabilizing policy measures or sector-specific resilience. Conversely, Energy, Financials, and Industrials show positive α, consistent with volatility spikes in response to pandemic shocks.

The β coefficients remain large and significant, demonstrating persistent volatility across all indices. Most sectors have α + β < 1, indicating eventual mean reversion. However, Energy again stands out with α + β > 1, confirming that shocks in this sector had unusually long-lasting effects, a result consistent with the global oil market collapse.

The leverage effect (λ) is positive and significant for nearly all sectors, except Energy, suggesting that negative shocks increased volatility more than positive shocks. The JSE ALSI has the highest λ, implying that the market overall was strongly driven by bad news (lockdowns, infection waves, policy uncertainty). Energy’s negative λ highlights its unique dynamic: positive shocks (such as recovery in oil prices and prospects for green energy) reduced volatility more than negative shocks amplified it.

Economic intuition: The T-GARCH findings reinforce the dominance of negative news in shaping South African equity volatility during COVID-19. However, sectoral heterogeneity emerges while most sectors reflected strong bad news-driven volatility, Energy uniquely benefited from positive developments, showing asymmetry in the opposite direction.

4.6. Diagnostics and Robustness Checks

To ensure the reliability of the GARCH family estimates, several diagnostic and robustness checks were performed.

• ARCH LM Tests

Post-estimation ARCH-LM tests were conducted on the residuals to verify whether conditional heteroskedasticity remained. The null hypothesis of no ARCH effects was consistently rejected at conventional significance levels before model estimation, confirming the presence of volatility clustering and justifying the use of GARCH-type models. After estimation, the ARCH-LM tests showed no remaining ARCH effects, indicating that the models adequately captured volatility dynamics.

• Residual Diagnostics

Standardized residuals and squared standardized residuals were examined through autocorrelation (ACF) and partial autocorrelation (PACF) plots. The absence of significant autocorrelation in residuals indicates that volatility shocks were well captured by the models. Jarque–Bera tests, however, confirmed departures from normality, which is common in financial return series. To account for this, robust standard errors were used in the estimation process.

• Robustness Across Specifications

To ensure robustness, three complementary models were estimated:

• GARCH (1,1) for baseline persistence and clustering;
• E-GARCH to capture asymmetries (leverage effects of good vs. bad news);
• T-GARCH to explicitly model the impact of positive vs. negative shocks.

Across specifications, consistent patterns emerged: volatility persistence was high in all sectors, the Energy sector showed the strongest and longest-lasting volatility, and asymmetry effects were significant, with most sectors exhibiting stronger reactions to negative shocks. This cross-model consistency strengthens the robustness of the findings.

• Comparison with Alternatives

While the study focused on univariate models for sector-level clarity, alternative models such as multivariate GARCH (BEKK, DCC-GARCH) could capture interlinkages across sectors. However, given the study’s focus on sector-specific volatility behavior during COVID-19, the univariate framework provides clearer sectoral insights. Future work could extend the analysis by applying multivariate models to test for volatility spillovers and co-movements across sectors.

• Stability and Persistence

The persistence estimates (α + β) were consistently close to or above unity across all models, supporting the interpretation of prolonged volatility during COVID-19. The consistency of persistence across different specifications confirms the robustness of this result, particularly for the Energy and Consumer Staples sectors, which exhibited the highest persistence.

• Model Suitability and Recommendation

Based on the diagnostic tests and information criteria, the E-GARCH model emerges as the most suitable specification for the South African equity market during COVID-19, as it effectively captures the leverage effect and asymmetric responses to shocks, which were a defining feature of pandemic-induced volatility. While the GARCH (1,1) adequately explains baseline volatility clustering, it underestimates the asymmetric impact of bad news. The T-GARCH also highlights asymmetry, but its performance is weaker in several sectors compared to E-GARCH.

Unlike previous South African studies that focused mainly on aggregate market indices using standard GARCH models, this study contributes by providing a sector-level analysis with asymmetric GARCH models. The results demonstrate that volatility drivers differ substantially across sectors, with Energy exhibiting persistence dominated by positive news, in contrast to the rest of the market, where negative news drives volatility. This sectoral heterogeneity provides a novel insight for portfolio managers and policymakers.

5. Conclusions and Recommendations

This study analyzed the volatility dynamics of South African equity sectors during the COVID-19 pandemic, focusing on sector-level returns, correlation with the broader market (JSE ALSI), and the influence of economic policy uncertainty (EPU). Daily return data from 1 January 2020 to 31 March 2022 were used, and univariate GARCH-family models (GARCH (1,1), E-GARCH, and T-GARCH) were employed to estimate conditional volatility.

Key Findings

• Volatility Clustering:

Graphical analysis of daily returns revealed clear volatility clustering across all sectors, with periods of low volatility followed by sharp spikes. Notably, March 2020 saw a substantial increase in volatility coinciding with the COVID-19 outbreak. Energy and Technology sectors displayed relative stability, reaching their lowest returns later in the sample period, likely due to factors such as oil price cuts (Energy) and increased demand from digital adoption (Technology).

• Sector-Market Correlation:

All sectors exhibited positive correlations with the JSE ALSI, indicating that sector performance generally influenced the broader market. Basic Materials, Consumer Discretionary, and Financials were most strongly correlated, while the Energy sector showed the weakest correlation, highlighting its potential role in portfolio diversification as a partially independent hedge.

• Economic Policy Uncertainty:

Correlation analysis between the SA EPU index and sector returns showed a weak negative relationship for most sectors, suggesting that policy uncertainty marginally impacted market returns. The Energy sector was an exception, showing a weak positive correlation, potentially reflecting government interventions such as subsidies or stimulus measures that mitigated policy-related risks.

• Volatility Dynamics (GARCH Models):

  ○

  GARCH (1,1): All sectors exhibited volatility clustering and persistence. Consumer Discretionary lacked ARCH effects and was excluded from model estimation. The Energy sector showed α + β > 1, indicating explosive volatility, while other sectors displayed mean-reverting behavior. Recent news (α) had significant impacts, especially in Financials and Consumer Staples, whereas Health Care and Basic Materials were less sensitive. Past shocks (β) contributed to volatility persistence across most sectors, though less so for the JSE ALSI.

  ○

  E-GARCH: The asymmetry parameter (γ) indicated that market sentiment was largely driven by negative news, except in the Energy sector, where positive developments dominated. This highlights sector-specific heterogeneity in volatility response.

  ○

  T-GARCH: Leverage effects (λ) were positive and significant for most sectors, confirming that negative shocks generally increased volatility more than positive shocks. The Energy sector had a negative λ, reinforcing that positive news shaped its volatility dynamics during the pandemic.

• Economic and Policy Interpretation:

The study suggests that SA equity market behavior during COVID-19 was influenced by both unique policy interventions (e.g., R500 billion stimulus package, interest rate cuts) and market inefficiencies (e.g., electricity supply challenges, exchange rate volatility). Sectoral heterogeneity is pronounced: Energy behaves independently, offering potential for portfolio diversification, while Consumer Staples and Financials are highly reactive to shocks, emphasizing the need for risk-adjusted portfolio weighting.

• Recommendations:
  • For investors: During extreme market events, portfolios should be weighted toward more resilient sectors, such as Energy and Technology, to mitigate downside risk from highly sensitive sectors. Diversification strategies should consider sector independence and asymmetry in shock responses.
  • For policymakers: Targeted interventions can support sectors negatively impacted by extreme events and reduce investor anxiety. Minimizing the circulation of negative news and implementing protective measures can enhance market stability.
  • For future research:

    ○

    Extend the analysis to other extreme events (e.g., Russia–Ukraine war) and compare pre- and post-pandemic periods.

    ○

    Incorporate additional exogenous volatility regressors to improve model explanatory power.

    ○

    Explore structural breaks and long-term persistence in sectoral volatility.

    ○

    Consider broader GARCH-family models (e.g., I-GARCH, APARCH) or multivariate GARCH specifications to study spillovers and co-movements across sectors.

    ○

    Increase the sample period beyond two years for greater robustness.

This study highlights the importance of sector-specific analysis for understanding volatility during extreme market events and provides practical guidance for both investors and policymakers in managing risk and supporting market stability.