Next Article in Journal
The Effect of Twitter Messages and Tone on Stock Return: The Case of Saudi Stock Market “Tadawul”
Previous Article in Journal
Digital Financial Capability Scale
Previous Article in Special Issue
An Empirical Examination of Bitcoin’s Halving Effects: Assessing Cryptocurrency Sustainability within the Landscape of Financial Technologies
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
JRFM
Volume 17
Issue 9
10.3390/jrfm17090403
jrfm-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

COVID-19 and Uncertainty Effects on Tunisian Stock Market Volatility: Insights from GJR-GARCH, Wavelet Coherence, and ARDL

by
Emna Trabelsi
1,2
1
Social and Economic Policy Analysis Laboratory, Higher Institute of Management of Tunis, University of Tunis, Tunis 2000, Tunisia
2
Faculty of Economic and Management Sciences of Sousse, University of Sousse, Sousse 4023, Tunisia
J. Risk Financial Manag. 2024, 17(9), 403; https://doi.org/10.3390/jrfm17090403
Submission received: 24 January 2024 / Revised: 3 March 2024 / Accepted: 3 April 2024 / Published: 9 September 2024
(This article belongs to the Special Issue Stability of Financial Markets and Sustainability Post-COVID-19)
Browse Figures
Versions Notes

Abstract

:
This study rigorously investigates the impact of COVID-19 on Tunisian stock market volatility. The investigation spans from January 2020 to December 2022, employing a GJR-GARCH model, bias-corrected wavelet analysis, and an ARDL approach. Specific variables related to health measures and government interventions are incorporated. The findings highlight that confirmed and death cases contribute significantly to the escalation in TUNINDEX volatility when using both the conditional variance and the realized volatility. Interestingly, aggregate indices related to government interventions exhibit substantial impacts on the realized volatility, indicating a relative resilience of the Tunisian stock market amidst the challenges posed by COVID-19. However, the application of the bias-corrected wavelet analysis yields more subtle outcomes in terms of the correlations of both measures of volatility to the same metrics. Our econometric implications bear on the application of such a technique, as well as on the use of the realized volatility as an accurate measure of the “true” value of volatility. Nevertheless, the measures and actions undertaken by the authorities do not exclude fear and insecurity from investors due to another virus or any other crisis. The positive and long-term impact on the volatility of US equity market uncertainty, VIX, economic policy uncertainty (EPU), and the infectious disease EMV tracker (IDEMV) is obvious through the autoregressive distributed lag model (ARDL). A potential vulnerability of the Tunisian stock market to future shocks is not excluded. Government and stock market authorities should grapple with economic and financial fallout and always instill investor confidence. Importantly, our results put mechanisms such as overreaction to public news and (in)efficient use of information under test. Questioning the accuracy of announcements is then recommended.

1. Introduction

The outbreak of the COVID-19 pandemic has triggered unprecedented challenges across global financial markets, with profound implications for economic systems and investment landscapes (see Onali 2020; Liu et al. 2020; He et al. 2020; Ashraf 2020; Okorie and Lin 2021; Mazur et al. 2021; Endri et al. 2021; Chowdhury et al. 2022). Among the affected regions, Tunisia has encountered distinctive dynamics in its stock market, marked by heightened volatility and fluctuations. This research endeavors to comprehensively examine the impact of COVID-19 on Tunisian stock market volatility through the lens of a GJR-GARCH model, wavelet coherence analysis, and autoregressive distributed lag model (ARDL). The motivation behind this study stems from the critical need to understand and quantify the specificities of the financial repercussions experienced by Tunisia in the wake of the pandemic, shedding light on the relationships between the key variables influencing market dynamics. The significance of this research lies in its potential to provide valuable insights for investors, policymakers, and financial analysts navigating during the post-COVID economic recovery, facilitating informed decision-making in the Tunisian financial landscape.
Our research is distinguished by incorporating specific determinants related to the COVID-19 pandemic and aims to quantify their impact on the conditional volatility of the TUNINDEX stock market return. It is also important to investigate the predictive power of the uncertainty indicators for volatility in the Tunisian context while considering a very sensitive period of investigation. By doing so, we not only offer a thorough exploration of a burgeoning economy grappling with the pandemic but also contribute valuable insights, particularly in light of the scarce existing literature within the scope of our research goals.
The remainder of this paper is as follows. Section 2 reviews specific studies that relate the stock market volatility to the COVID-19 crisis. Section 3 presents data and materials to answer the research objective. The main findings are exposed in Section 4. Concluding remarks and policy implications of the results are displayed in Section 5 and Section 6, respectively.

2. Literature Review

The literature on the determinants of stock market volatility is abundant (see Binder and Merges 2001; Mazzucato and Semmler 2002; Nikmanesh and Nor 2016; Hussain et al. 2019; Hewamana et al. 2022; Biu and Kusuma 2023; Dhingra et al. 2023; etc.). Particular attention has been given to the linkages between incidences (events) and stock markets. Various catastrophic events, such as terrorist attacks (see Kollias et al. 2013; Aslam et al. 2014, 2018, 2020), earthquakes (Ferreira and Karali 2015; Hendricks et al. 2020), and natural disasters (Worthington 2008; Kowalewski and Śpiewanowski 2020), have historically impacted stock markets and other economic activities. Examining how sensitive a market’s behavior is when pended to an exogenous shock has been the topic of many studies. For example, Barrett et al. (1987) found that the immediate negative effects of serious accidents are typically limited to the first trading day after the event. Li et al. noted that meteorological disasters can trigger stock market volatility. Additionally, Hanabusa (2010) analyzed how terrorist attacks and natural disasters influenced the stock prices of Japanese oil firms. Robinson and Bangwayo-Skeete (2016) studied the economic and financial impacts of hurricanes and tropical storms on small islands, finding that the losses for stock market investors can be significant, exceeding reported losses from property and infrastructure damage. Bourdeau-Brien and Kryzanowski (2017) found that the second moments of local stock returns more than double when hurricanes, floods, winter storms, and extreme temperature episodes occur. Kowalewski and Śpiewanowski (2020) compared the stock market’s response to natural disasters versus man-made disasters, finding that the response to natural disasters is typically stronger. In this paper, we focus on research that relates stock market volatility to the COVID-19 crisis. Li et al. (2022) established a link between COVID-19 fear and stock market volatility, emphasizing its significance for portfolio diversification. Findings suggest that COVID-19 fear is a primary driver of public attention and stock market fluctuations, with associated impacts on both stock returns and GDP during the pandemic. Gao et al. (2022) used the quantile-on-quantile approach to compare the impact of new COVID-19 cases on U.S. and Chinese stock market volatility, revealing COVID-19 as the primary driver in the U.S. with implications for monetary policy and market stability amid the global spread of the virus. Kusumahadi and Permana (2021) examined the global impact of COVID-19 on stock return volatility in 15 countries and identified structural changes preceding the onset of COVID-19, suggesting the positive influence of the virus on return volatility, though the effect size is modest. Uddin et al. (2021) emphasized the importance of leveraging economic factors such as capitalism, monetary policy, and financial development in policymaking to address global stock market volatility and prevent potential financial crises. Bora and Basistha (2021) compared the stock return volatility performance of India before and during the COVID-19 crisis and documented a higher volatility due to the pandemic effect. According to Baek et al. (2020), COVID-19 news significantly influences U.S. stock market volatility, with a notable negativity bias and industry-wise changes in systematic risk. Yousef (2020) used regression analysis and GARCH models to show that the COVID-19 pandemic, daily new cases, and growth rate of daily new cases significantly increased volatility across all G7 stock indices. The same view is shared by Izzeldin et al. (2021). Through GARCH modeling, Chaudhary et al. (2020) found a negative mean return during the COVID period, with heightened volatility, suggesting a bearish trend in the top 10 countries according to GDP. The global COVID-19 fear index seems to have a pronounced effect on the volatility of 19 emerging stock markets (see Sadiq et al. 2021). Papadamou et al. (2020) highlighted that increased anxiety regarding COVID-19 contagion on Google correlates with heightened risk aversion in stock markets, especially in Europe. The volatility in stock prices caused by COVID-19 affects abnormal returns, prompting investors to implement risk management strategies amid uncertainty, while also creating potential speculative opportunities in an inefficient market (Endri et al. 2021). Lúcio and Caiado (2022) observed that amid the COVID-19 outbreak, all S&P 500 industries, except internet and direct marketing retail, witnessed a substantial rise in volatility, leading to a convergence in time-varying variance among them.
Worldwide investors respond differently to an exogenous shock. Those differences are mainly attributed to national culture and governance quality (Bakry et al. 2022). In Tunisia, akin to major global stock exchanges, the equity market exhibited rapid reactions. The detection of the initial positive COVID-19 cases since 2 March 2020, induced extensive market apprehension, culminating notably during the trading days of March 13, 16, and 17, with the TUNINDEX experiencing a notable decline of 10%. This urged the Central Bank of Tunisia (CBT) and the government to undertake emergency actions to curb the pandemic’s consequences on the overall economy, including stock markets.1
Studies on the effect of COVID-19 on stock return volatility in the Tunisian context are very recent. Using GARCH models, Akinlaso et al. (2022) detected high persistence in both Tunisia’s Islamic and conventional stock markets, with the conventional index negatively influencing Islamic stocks during the pandemic. Jeribi et al. (2015) and Fakhfekh et al. (2023) revealed increased volatility persistence in all Tunisian sectorial stock market indices post-COVID-19, with certain sectors showing significant asymmetric effects.
We are particularly interested in how the Tunisian stock market reacts to COVID-19 measures and to the government interventions to limit the spread of the virus. As technology advances and information is made available, metrics about the markets’ opinions on unexpected events are provided. This is relevant to understand investor sentiment and attention in response to crises (see Papadamou et al. 2020). COVID-19 is an event that puts market efficiency under test (Fama 1965). Supporters of behavioral finance (e.g., De Bondt and Thaler 1985; Barber and Odean 2001) documented irrational behavior, while Grossman and Stiglitz (1980) reject perfectly efficient markets. Particularly, stock markets are inclined to overreact to new (bad) events (Scherf et al. 2022). Such an argument was challenged in rational expectation models (e.g., De Bondt 1989; Veronesi 1999; Dissanaike 1997; Chari et al. 2017) and in game theoretic approaches (Morris and Shin 2002). It is further intriguing to ask whether more information necessarily means more or little use of information. Asset-market-based models highlighted the dilemma of (in)efficient use of information (see Angeletos and Pavan 2007). Our results show that Tunisian market participants pay attention to confirmed and mortality rates but show little interest to the measures undertaken by the government.

3. Data and Methodology

3.1. Data

We collect daily data on TUNINDEX returns amid the COVID-19 pandemic. The period spans from 2 January 2020 to 30 December 2022, making a total of 755 observations, omitting nontrading days. Since the objective is to analyze the impact of the sanitary crisis on the Tunisian stock return volatility, we gather data on the number of cases (total, new, and death) and on government interventions to limit the spread of the virus (see Table 1 for details).
Descriptive statistics of the above variables are displayed in Table 2. The TUNINDEX exhibits a mean return of 0.000150, indicating a generally modest performance. Figure 1 illustrates a compelling narrative of diminishing returns in TUNINDEX, particularly during the challenging times of the COVID-19 era. The total number of COVID-19 cases and deaths, with means of 551,839.1 and 16,432.85, respectively, reflects the significant impact of the pandemic on public health. The stringency index, with a mean of 49.03363, suggests varying degrees of government measures. Interestingly, the containment health index, economic support index, and government response index do not seem to strongly influence market volatility. The skewness values, particularly for TUNINDEX, indicate a negatively skewed distribution, and the kurtosis values suggest heavy tails in the distribution.

3.2. Methodology

The assessment of volatility has mandated the application of models within the autoregressive conditional heteroskedasticity (ARCH) framework in the realm of financial literature. Originating with the seminal ARCH model introduced by Engle (1982), subsequent advancements have been made, encompassing the development of models such as generalized ARCH (GARCH), GARCH-in-mean, quadratic GARCH, and threshold GARCH, among others.
The stock return is calculated following previous related literature using this formula:
R t = Ln ( P t P t − 1 )
where P t   a n d   P t − 1 are the stock market indices of the Tunisian Stock Exchange at times t and t − 1, respectively. Ln is the natural logarithm.
The logarithm transformation serves as a technique to induce stationarity in variables by mitigating trends and seasonality within time-series data. Consequently, augmented Dickey–Fuller (ADF) and Phillips–Perron (PP) tests are conducted to verify the attainment of stationarity in the level of returns. The associated results are available in Table 3.
The autoregressive (AR) (p) or autoregressive moving average (ARMA) (p, q) structures are commonly integrated as explanatory elements within the conditional mean equation of generalized autoregressive conditional heteroskedasticity (GARCH) models for the dependent variable. Following Insaidoo et al. (2021), an advantageous feature of the GARCH model lies in its adaptability to include explanatory variables and dummy variables in both the conditional mean and variance equations, thereby accommodating the specific objectives of the study. Hence, we incorporate each of the COVID-19 variables and the government intervention controls in both the mean equation and the variance equation separately to capture the effect of the pandemic on Tunisian stock return and volatility. Our econometric modeling is essentially based on Fakhfekh et al. (2023). However, we distinguish from their study by including specific controls related to the sanitary crisis instead of separating the pre-crisis and during-crisis periods. Thus, we estimate the following GARCH family models.

3.2.1. ARCH

The ARCH model characterizes the variance of a time series, particularly in situations involving dynamic and potentially volatile variance. It was introduced by Engle (1982). While ARCH models can potentially depict a slowly escalating variance trend over time, their primary application lies in scenarios featuring brief episodes of heightened variation. It is worth noting that situations involving a gradual increase in both variance and mean level may be more effectively addressed by transforming the variable.
The mean equation is written as follows:
R t = α + ∑ i = 1 p β i R t − i + ∑ i = 1 q θ i ε t − i + ε t
The choice between AR(1), MA(1), and ARMA(1,1) is made based on the conventional information criteria (Aikake and Schwarz) for which the value is minimized.
Our estimations show that AR(1) is the preferred one.
The conditional variance equation is defined as:
σ t 2 = a 0 + ∑ i = 1 p a i ε t − i 2
where a 0 > 0 ,   a i > 0 . This model is known as the linear ARCH(p). In financial data analysis, the model captures the phenomenon of volatility clustering, wherein significant (insignificant) price changes tend to be succeeded by other substantial (moderate) price changes, but with an unpredictable direction. To streamline the model and guarantee a consistently diminishing impact of more distant shocks, a makeshift linearly declining lag structure was frequently enforced in many early applications of the model (see Bollerslev et al. 1992; Bollerslev and Mikkelsen 1996).

3.2.2. GARCH

Bollerslev’s (1986) GARCH model is the most widely used model for estimating volatility. GARCH models have had great success in the literature due to their simple specification and ease of interpretation.
The conditional variance equation is defined as:
σ t 2 = a 0 + ∑ i = 1 p a i ε t − i 2 + ∑ j = 1 q b j σ t − j 2
where a 0 > 0 ,   a i > 0 ,   b j > 0 . For a GARCH(1,1), the constraint a 1 + b 1 < 1 implies that the unconditional variance of the ε yield series is finite and that the conditional variance σ t 2 evolves. It also provides what is necessary and sufficient for the stochastic process σ t ; t ∈ Z to be a unique process strictly stationary with E( σ t 2 ) < ∞.
Two key properties can be noted from the above equation. First, a high value of ε t − 1 2 or σ t − 1 2 gives rise to a high value of σ t 2 , and this generates the volatility clustering that is common in time series. Second, the tail distribution is thicker than that of a normal distribution (see Mestiri 2022).

3.2.3. EGARCH

This model is suggested by Nelson (1991) to account for leverage effects. It is an asymmetric exponential GARCH specification that distinguishes between “good” and “bad” news on volatility.
The conditional volatility equation is given by:
L n σ t 2 = a 0 + ∑ i = 1 p a i ε t − i + δ i ε t − i σ t − i + ∑ j = 1 q b j L n σ t − j 2
If there is “good” news, ε t − i > 0 , and the total effect is given by ( 1 + δ i ) ε t − i . If there is “bad” news, ε t − i < 0 , and the total effect is given by ( 1 − δ i ) ε t − i . “Bad” news is likely to have a more pronounced impact on volatility, and δ i is expected to be negative.

3.2.4. GJR-GARCH

Another version of an asymmetric GARCH model with leverage effects is the threshold GARCH specification, called also GJR-GARCH. The conditional volatility is given by:
σ t 2 = a 0 + ∑ i = 1 p a i ε t − i 2 + ∑ i = 1 p γ i S t − i ε t − i 2 + ∑ j = 1 q b j σ t − j 2
with
S t − i = 1   if   ε t − i < 0 0   if   ε t − i > 0
When ε t − i > 0 , the total effect is given by a i ε t − i 2 . If ε t − i < 0 , the total effect is given by a i + γ i ε t − i 2 . The coefficient γ i is expected to be positive so that “bad” news has a larger impact on volatility.
The best-fit model is chosen based on Akaike (AIC) and Schwarz (SC) information criteria for which the value is minimized.
According to Bera and Higgins (1993), the most applied financial work shows that GARCH (1,1) provides a flexible and parsimonious approximation of the conditional variance dynamics and is capable of representing the majority of financial series. Therefore, p = 1 and q = 1 are applied for all GARCH family models to be estimated.
Before estimating ARCH-type models, we need to test for the presence of ARCH(q) effects. We rely on the Lagrange multiplier (LM) test. We reject the null hypothesis of no ARCH effects (F-statistic = 342.0031, p-value = 0.000; Obs × R2 = 235.6154, p-value = 0.000).
To know what combination of variables is to be included in the regression, we provide the pairwise correlation matrix (see File S1 of the Supplemental File for details).

4. Results and Discussion

4.1. Impact of COVID-19 Announcements and Government Intervention on TUNINDEX Stock Return Volatility

Table 3 shows that GJR-GARCH is the preferred one, as the conventional information criteria exhibit the lowest values. To examine the impact of COVID-19 announcements and the government intervention variables, we regress the conditional stock return variance that results from the GJR-GARCH(1,1), AR(1) model on the variables of interest and a set of controls (see Bakry et al. 2022):
V O L t = α + β 1 C R t + β 2 D R t + β 2 G I I t + β 1 O i l _ V O L t + μ t
where CR: cases_rate, DR: death_rate, GRI: government intervention index (this includes the stringency_index, containment_health_index, economic_support_index, government_response_index, and sub-indicators of the stringency index (s1 to s9)), and Oil_Vol: conditional WTI oil volatility from GJR-GARCH(1,1) with a constant.
To handle the multicollinearity issue, we include the government intervention indicators separately. Furthermore, we control for autocorrelation and heteroscedasticity by estimating the equation with the Newey–West method. Results are available in Table 4 and Table 5. We re-estimate the variants of the equations while skipping for the cases_rate variable. All results mostly duplicate the ones in Table 4 and Table 5, except that death_rate becomes statistically significant (see Table 6 and Table 7).
According to Kamal and Wohar (2023), “Considering global market integration (Solnik 1974) and to remove misspecification error, [..]. We also lagged the independent variables by one period to examine the slow response of the market. Slow response supports the underreaction hypothesis, which suggests investors adjust slowly to new information.” Investors take some time to incorporate new information. We replicate the output while considering the variables of interest and controls lagged with one period (see Table 8, Table 9, Table 10 and Table 11).
The observed incidences of confirmed cases and mortality rates exhibit a positive association with heightened volatility in the TUNINDEX in addition to the WTI conditional volatility. Moreover, there exists a positive correlation between the stringency index and volatility in the Tunisian stock market. Conversely, the indices about containment health economic support and government response do not manifest a discernible impact on market volatility. Those observations can be attributed to increased uncertainty and economic impacts. Rising cases make investors cautious, leading to market fluctuations. Stringent government measures, reflected in the stringency index, also correlate positively with market volatility, indicating economic repercussions. However, indices related to health economic support and government response show no significant impact on volatility, suggesting that other factors may be more influential in shaping the Tunisian stock market.
File S2 of the Supplementary File shows the results when all variables are observed with a delay of two days. The results are qualitatively identical to our previous observations, but we depict a higher impact of COVID-19 health situations when there is an immediate release of confirmed and mortality rates. Market volatility through the GJR-GARCH model does not seem to respond to government intervention even with a delay, except the sub-indicator of the stringency index, which is related to public gathering information, which has an adverse effect on market volatility. Market participants seem either to be inattentive to the lockdown restrictions imposed by the government, or they do not appreciate some of them. Particularly, public information gathering (s9) deteriorates the stock market performance. The effect is more pronounced when the variable is lagged by two periods. We note that the flow of information on social distancing imposed by the government leads investors to be somewhat unresponsive. Much information and little use of it confirm the observations of Angeletos and Pavan (2007) on the inefficient use of information, while we present evidence on an overreaction to health news portrayed by the affected and dead people, aligning with Morris and Shin (2002).
According to Andersen and Teräsvirta (2009), it is important to differentiate between the conditional variance, standing for the (ex-ante) expected size of future squared return innovations over a certain period and the quadratic change in return over some horizons, reflecting the actual (ex-post) realization of return variation. Thus, we follow Mouselli and Mohialdeen (2021) and construct the realized volatility by following a simple weighted method encompassing S&P 500. Mouselli and Mohialdeen (2021) provided a nice hint in calculus while accounting for the absence of trading options in emerging markets. This includes the Tunisian stock market.
R V t = 252 n 1 N ∑ i = t − N + 1 t ln p i p i − 1 2 = 252 n 1 N ∑ i = t − N + 1 t R i 2
where n is the number of trading days, and N is set to 60 days. R i is the return as computed previously. We re-estimate the Newey–West time-series regressions while incorporating the same set of variables. First, we observe how market prices adjust to health situation updates, to the central bank announcements on interest rates, and to the government lockdown restrictions. It is noteworthy that interest rate differentials and government interventions exhibit an immediate effect on the realized volatility (see Table 12, Table 13, Table 14 and Table 15). These effects still show up significantly when the variables are delayed by one day (see Table 16, Table 17, Table 18 and Table 19) and by two days (see File S3). Both the Central Bank of Tunisia and the government’s information disclosure have a positive impact on stock market performance, contradicting the findings of Ben Ayed (2022). Our results support somewhat the resilience of the Tunisian stock market to the pandemic. As for the sub-indicators of the stringency index, we observe a similar pattern to when conditional variance is employed as a measure of volatility. Particularly, restrictions on public gatherings are incorporated into market prices, but investors respond to them negatively. Indeed, realized volatility increases when the stringency sub-indicator (s9) increases. It is, however, surprising to find that the same metric of volatility does not respond to confirmed and death rates. This observation is in line with Scherf et al. (2022).
Our findings emphasize that the decree issued by the president of the republic on March 22, which implemented a general lockdown, combined with the implementation of monetary policy measures, such as reducing the CBT’s key rate by 100 basis points to 6.75% and relaxing prudential regulations and refinancing policy, along with the government and CBT’s announcement of exceptional measures to support companies and individuals affected by the COVID-19 health crisis, collectively contributed to the restoration of investor confidence in listed shares. Apart from these measures, we have to premise that the financial market council issued rules such online meetings and the change in trading hours (closes at 12.10 p.m. instead of 2.30 p.m.). Particularly, the decision by market authorities to shorten trading sessions and limit the range of price variations to ±3% per session (down from ±6% previously) during the period from March 18 to 5 June 2020 has mitigated the negative impact on the market.

4.2. Wavelet Coherence Analysis

Wavelet coherence analysis is a method used to investigate the relationship between two time series in both the time and frequency domains. It is particularly useful for exploring the coherence between signals that vary in time and is well suited to handle series with both stationary and non-stationary properties. Then, we can explore the coherence or synchronization between two signals across different scales and time intervals, making it a valuable tool in the analysis of complex and dynamic relationships between time-varying processes (see Torrence and Webster 1998; Grinsted et al. 2004; Liu et al. 2007; Cazelles et al. 2008; Rouyer et al. 2008; Veleda et al. 2012).
Bias-corrected wavelet analysis refines traditional methods by mitigating edge effects and leakage issues associated with finite-length time series. This approach provides more accurate estimates of wavelet coefficients, enhances statistical inference, and improves the interpretability of time-frequency characteristics. It is particularly valuable for handling challenges related to discreteness and non-stationarity in signals, offering a robust tool for understanding the dynamics of time-varying processes. In this part, we follow this approach (e.g., Dhanya and Gupta 2014; Hosseini Baghanam et al. 2023).
The mathematical expressions provided will be based on Tissaoui et al. (2021), Tissaoui et al. (2022), and Li et al. (2022). A wavelet is a function Ψ(.) that can be either real-valued or complex-valued, and it meets the criterion of being square-integrable.
A wavelet is characterized as a compact or small-sized wave if
Ψ l , d t = 1 d Ψ ( t − l d )
where 1 d is the normalization factor, such that Ψ l , d t = 1 , l stands for the location parameter, providing the exact position of the wavelet, and d denotes the scale dilatation parameter of the wavelet.
The cross-wavelet technique can break down the x ( t ) function initially and subsequently reconstruct it, such that
x t = 1 C W ∫ 0 ∞ ∫ − ∞ ∞ W x l , s Ψ l , d t   d l d d d 2 ,   d > 0
The process involves projecting a particular wavelet to achieve the desired outcome. The primary goal of wavelet coherence is to calculate localized correlations within a time-frequency domain across a series. We have
R 2 l , d = S ( s − 1 ( W m n ( l , d ) 2 S s − 1 W m l , d 2 . s − 1 W n l , d 2
The given expression is assessed using the absolute smooth cross-wavelet value. R 2 l , d is very similar to the correlation coefficient between two signals, m and n .
Approaching zero, it indicates a weak correlation between m and n , whereas a value close to 1 signifies a strong correlation between the two variables.
The phase discrepancy serves as an indicator of the timing variation between oscillations in two variables about their frequency. The analysis of this phase difference involves examining the orientation of arrows depicted in wavelet coherence graphs. More precisely, the identification of the lead-lag relationship between two time series is made by observing the directional alignment of arrows. When arrows point in the right direction, it signifies the in-phase alignment of the two signals, whereas leftward-pointing arrows indicate an anti-phase alignment.
The horizontal axis of the graph displays time, while the vertical axis depicts frequency, with higher scales corresponding to lower frequencies. The wavelet coherence identifies regions in time-frequency space where two time series co-vary. Significantly related areas are represented by warmer colors (red), indicating a strong interrelation, whereas colder colors (blue) signify lower dependence between the series. Cold regions outside the significant areas indicate time and frequencies where there is no dependence in the series. In the wavelet coherence plots, arrows indicate lead/lag phase relations between the examined series. A zero-phase difference denotes that the two time series move together at a particular scale. Arrows pointing to the right (left) indicate in-phase (anti-phase) relationships between the time series.
Globally, the COVID-19 virus persisted in causing anxiety, uncertainty, and distress in Tunisia. This situation heightened volatility in financial markets and led to liquidity challenges. The Tunisian stock market was not exempt from the profound effects of the COVID-19 outbreak, as evidenced by the ratios of confirmed cases and death cases, respectively (see Figure 2 and Figure 3).
Among the government intervention metrics, the stringency index, the containment health index, and the government response indices seem to produce the most impact on stock market volatility during the COVID-19 times, while the economic support index had a neglected impact in the same period (see Figure 4, Figure 5, Figure 6 and Figure 7), confirming partially our previous results (see Table 4, Table 5, Table 6, Table 7, Table 8, Table 9, Table 10 and Table 11).
Now, we decompose the stringency index into sub-indices. Figure 8, Figure 9, Figure 10, Figure 11, Figure 12, Figure 13, Figure 14, Figure 15 and Figure 16 display the correlations between TUNINDEX stock market volatility and each of the nine stringency index sub-indicators.
We observe that the closure of schools immediately influences volatility during the sanitary crisis (see Figure 8). Similarly, workplace closures exhibit an immediate impact, persisting over a medium-term horizon (see Figure 9). The cancellation of public events appears to have a notable and enduring impact over a longer-term horizon, as evidenced by a relatively high correlation with stock market volatility that persists beyond the COVID-19 period (see Figure 10). This pattern is evident when considering the stringency subindex related to restrictions on public gatherings (see Figure 11). Restrictions on public gatherings seem to impact conditional volatility over a short-term horizon (see Figure 12). Stay-at-home requirements distinctly impact volatility over short- to medium-term horizons (see Figure 13). Restrictions on internal movements have an immediate impact on volatility triggered by the pandemic (see Figure 14). Control measures on international travel affect volatility over a medium-term horizon, with the effect extending post-COVID (see Figure 15). The gathering of public information influences volatility in the long run (see Figure 16).
The observed impact of various stringency measures on market volatility during the COVID-19 crisis holds significant implications for policymaking. Measures such as school closures and workplace shutdowns immediately contribute to volatility during the sanitary crisis, necessitating swift and targeted interventions to stabilize financial markets. Notably, workplace closures exhibit a sustained impact over a medium-term horizon, highlighting the importance of measures to repress prolonged market uncertainties. Public event cancellations demonstrate a lasting influence, emphasizing the need for strategies to manage and recover from enduring economic disruptions. The stringency subindex related to restrictions on public gatherings affects conditional volatility over a short-term horizon, requiring a delicate balance in policymaking to ensure public health measures without causing prolonged market disruptions. Stay-at-home requirements distinctly impact volatility over short- to medium-term horizons, necessitating adaptive financial measures. Internal movement restrictions trigger an immediate market response, highlighting the importance of swift financial interventions. International travel control affects volatility over a medium-term horizon, prompting policymakers to consider strategies supporting industries reliant on international travel and trade. Lastly, the influence of public information gathering on long-term volatility underscores the importance of transparent communication for maintaining market stability. Policymakers can leverage these insights to design targeted interventions addressing both immediate and prolonged economic implications during crises.
We substitute the conditional variance with the realized volatility and replicate the wavelet coherence analysis. Figure 17 and Figure 18 support the time-series regressions’ findings. Market prices seem to fairly incorporate news about the health situations.
As for the government intervention measures, investors rely heavily on the economic support of the government and to social distancing, especially during the COVID-19 period (see Figure 19, Figure 20, Figure 21 and Figure 22). As the restrictions become relaxed, stock market participants underreact to further news. Again, we observe that much information related to the stringency index engenders less attention and use by investors (Figure 23, Figure 24, Figure 25, Figure 26, Figure 27, Figure 28, Figure 29, Figure 30 and Figure 31), confirming again that more information does not necessarily mean total and efficient use of that information (Angeletos and Pavan 2007).
The COVID-19 crisis is once again reshuffling the deck when it comes to investing in the stock market. Whenever financial markets are hit by a (sudden) exogenous shock such as the health crisis, there is a surge of uncertainty. This is because the future is always surrounded by doubt and mystery. While we show that investors’ confidence is fairly restored through the CBT and government’ s actions, they still exhibit skeptical behavior. Unfortunately, measures of investors’ sentiments are not available for the Tunisian context. Therefore, we try to depict insights through uncertainty measures and disentangle their effect on market volatility in the long run.

4.3. Impact of Uncertainty on the TUNINDEX Stock Return Volatility Using ARDL

The earlier COVID-19 and government intervention metrics serve as indicators of the broader economic and health landscape, influencing investor confidence and risk perceptions. However, the uncertainty surrounding COVID-19 has a notable impact on stock market volatility. The dataset comprises daily values of economic policy uncertainty (EPU), the infectious disease EMV tracker (IDEMV), implied volatility (VIX), and U.S. equity market uncertainty (EMU). The development of EPU and IDEMV was initiated by Baker et al. (2016, 2019). Quoting (Tissaoui et al. 2022, p. 4), EPU quantifies the number of articles in major newspapers covering news related to the economy, uncertainty, monetary and trade policies, and financial regulation in the U.S. (https://www.policyuncertainty.com/us_monthly.html, accessed on 2 October 2023). Data on IDEMV are accessed from (https://www.policyuncertainty.com/infectious_EMV.html, accessed on 2 October 2023). Additionally, the VIX is the implied volatility index of the Chicago Board Options Exchange. We gather data on EMU from the Federal Reserve Bank of St. Louis https://fred.stlouisfed.org/series/WLEMUINDXD, accessed on 2 October 2023. Unit root and Johansen tests indicate that the variables are I(0) and I(1) and cointegrated (see File S4 in the Supplementary File). We adjust coefficients to the White estimator as only heteroscedasticity is detected when analyzing residuals that remain uncorrelated and stable according to the Ramsey test (not shown for tractable reasons but available upon request). All variables, except EPU, have both positive long- and short-term effects on Tunisian stock market volatility (see Table 20). This could be due to factors like the stronger influence of local economic conditions, political stability, and industry-specific dynamics on TUNINDEX volatility. When we substitute the dependent variable with the realized volatility, we observe the same pattern, except that the EPU becomes statistically significant (see Table 21). At the same time, the findings emphasize the importance of monitoring not only domestic factors but also global economic and health-related developments to anticipate and navigate volatility in the Tunisian stock market.

4.4. Discussion and Comparison with Other Studies

Our findings highlight investors’ overreaction to public announcements of infected and dead people. In Tunisia, the Ministry of Health released the most recorded rate of infections on 23 July 2021. Such public news triggered panic and fear among investors, leading to upward levels of volatility in consonance with the observations of Akhtaruzzaman et al. (2021) and Zhang et al. (2020). This effect is instantaneous and is significant for one day to two days ahead. Emergency healthcare programs are supposed to limit the spread of the virus (Alfano and Ercolano 2020; Greenstone and Nigam 2020). However, positive effects are observed only when the realized volatility is used as the dependent variable. The aggregate stringency index seems to decrease the realized volatility when taken as an aggregate measure. The government banned travel between cities and canceled all gatherings and events. Weekly markets are closed, and a curfew has been established between 7 p.m. and 5 a.m. In 2021, lockdown restrictions hold despite vaccination campaigns. Nevertheless, the sub-indicators (s4 and s9) have a conflicting effect. Cancelling or postponing events and gatherings of any size is meant to contain the pandemic to protect vulnerable people by maintaining hand hygiene and social distancing, but at the same time, some investors may consider some policies as reducing economic activities. Ashraf (2020) and Zaremba et al. (2020) observed negative outcomes due to social distancing. The economic support index has led to a decrease in the realized volatility of the Tunisian stock market. The collection of funds in the country and abroad, known as the 1818 program, seems to have a fairly positive impact on market prices. However, doubts surround how these funds are used and spent. Another form of socio-political anarchy can emerge due the political regime changes in Tunisia. Switching between the different elected governments might relatively affect investors’ behavior (Bonaccorsi et al. 2020; Tisdell 2020). CBT’s plans to contain the economic impact of the virus include, among others, delays in taxes and ease in credit payments. Importantly, the decision of the CBT to cut the interest rate by 100 basis points to 6.75 per cent to cope with the repercussions of the coronavirus has fruitful impacts on TUNINDEX market prices. Oil volatility plays a central role in shaping stock markets. We establish the variable to be a strong predictor of stock market variances for both measures of volatility. Particularly, we find that the oil price fluctuations have led to jumps in TUNINDEX stock market volatility, aligning with Christoffersen and Pan (2018).
We depict further similarities and disparities between our results and other studies. For instance, Albulescu (2021) observed a positive correlation between U.S. market volatility the occurrence of new COVID-19 cases and the fatality ratio. In contrast, Onali (2020) did not find a similar effect on returns in the United States. Evidence from other emerging markets, such as Asian stock markets, indicates a high persistence in volatility during the pre-COVID period, as noted by Yong et al. (2021). Some researchers posit that the impact of COVID-19 varies depending on the stock exchange index used, as suggested by Gherghina et al. (2021), or on the degree of freedom, as highlighted by Erdem (2020).
Due to the quarantine procedures, investors feel insecurity, shifting the focus to the predictive power of uncertainty or the impact of investors’ sentiments on the stock market (Szczygielski et al. 2024). It is acknowledged that the spillover of the U.S. macroeconomic shocks has its effect on emerging markets, albeit different depending on trade agreements, political coalitions, and portfolio investment opportunities. The VIX variable, which describes the uncertainty level of global equity along with EMU, EPU, and IDEMV, has a long-run effect on international investment diversification. Our perspective aligns with Li et al. (2020), Alqahtani et al. (2020), and Asgharian et al. (2023) in the context of other countries and regions. The positive contributions of the uncertainty indices are also aligned with the observations of Su et al. (2019) using GARCH-MIDAS for industrialized and emerging markets. However, our findings diverge from those of Megaritis et al. (2021), who argue that macroeconomic uncertainty surpasses standard uncertainty indices in forecasting volatility. It is worth noting that those observations depend on the methodologies used, contextual variations, market characteristics, and time horizons.
When relating stock markets to uncertainty, misinformation is likely to disturb investors’ sentiments (Arcuri et al. 2023). Such an intriguing concept reflects how fake or false announcements may drive market prices and explain abnormal trade activity, as some of the uncertainty variables are newspaper-gleaned information—that is, information about the future economic outlook. These measures provide a challenge to information precision. It is, then, of paramount importance to extract the source of accurate information. Rumors are another source of misleading behavior that should be accounted for in this context, particularly in a digitalized world (Wang et al. 2019).
Last but not least, our results emphasize that realized volatility yields unbiased and accurate variance, improving upon standard measures from the GARCH family (Ebens 1999). The differences in results between the conditional variance and the realized volatility show that the latter captures economic gain in portfolio allocation (Fleming et al. 2003).

5. Conclusions

This study provides valuable insights into the dynamics of financial markets in the face of unprecedented global challenges in the Tunisian context. The findings unveil the importance of considering not only traditional financial factors but also incorporating health-related variables to comprehensively assess and manage stock market risks.
The pronounced effect of both CBT and government intervention is particularly highlighted through the use of realized volatility and bias-corrected wavelet analysis, underlining the advantages of such a measure and empirical technique, respectively. Certainly, safeguarding population health is of paramount importance, but they should be supplemented by monetary policies. This is well highlighted through the key interest rate of CBT. Our results underline that the Tunisian government should ramp up efforts toward public emergency actions and economic-policy-sustained measures to ingrain confidence into investors. At some point, those policies should not be at the expense of investors’ sentiments. The proliferation of restrictions might be suffocating and trigger downward performance. Some public announcements are a premise of the misuse of information (more information but little use of it). This is emphasized through the nine sub-indicators of the stringency index on both measures of volatility despite its beneficial aggregate impact on the realized volatility.
Somewhat, the Tunisian stock market shows a relative resilience against the pandemic. Further factors explain such a pattern. Particularly, there was low free-float capitalization of foreign investors in Tunisia (between 2% and 4%), limiting potential outflows. Daily variation ranges for shares have been revised downward to ±3% per session (compared with ±6% previously), and the length of the trading session has been reduced.
Moreover, there is a low representation of the sectors hit the most by the crisis, such as tourism and air and sea transport, on the Tunisian Stock Exchange. Another reason—and not the least—for the relative resilience of the Tunisian equity market is that its listings do not reflect the reality of the economic fabric. Sectors likely to be impacted by the crisis are currently absent from the market. These include tourism related to hotels and restaurants, the craft industry, importing sectors such as franchises and international trade companies, and air and sea transport (with the exception of the national carrier Tunisair, which has been sanctioned on the stock market for several years due to structural problems of governance and over-indebtedness). Nevertheless, economic vulnerability to upcoming shock is not excluded.

6. Policy Implications

The results highlight how important the central bank and government actions are in reducing the effect of COVID-19 on stock market volatility in the Tunisian context. There is a need for policy adjustments, urging the government to reassess the efficacy of its crisis response measures, particularly in terms of health containment and economic support. Practically, investors and market participants should consider incorporating a broader set of indicators, beyond government indices, into their risk assessments to make more informed decisions amid market volatility. Our results emphasize an overreaction mechanism regarding health news. Investors are attentive to some government measures, while they do not respond to the stringency sub-indicators, underlining that the flow of announcements about lockdown restrictions is hardly incorporated into market prices. Such a phenomenon is well explained in theory (see Trabelsi 2024). While resilience to the COVID-19 crisis emerges, vigilance is appealing, as investors’ fear of such a grim circumstance could show up. The uncertainty indicators point to the deep impact of an exogenous shock at a longer horizon. From a methodological point of view, we detect some differences in results between the conditional variance and the realized volatility. These findings bear implications for capturing the true pattern of volatility. We argue that the GARCH models do not provide sufficiently good forecasts of volatility, and the realized volatility is more useful to understand the pattern of stock markets.
From a managerial perspective, businesses and financial institutions operating in Tunisia should adopt adaptive strategies that account for the inefficiencies in governmental responses, ensuring resilience and agility in navigating economic uncertainties linked to health crises. As for the stock market authorities, further measures should be executed by stock market authorities. This includes establishing a prudential metric that warns against possible shocks (Mouselli and Mohialdeen 2021). Other measures should be considered to improve the stock market. These include adjusting the timing of quotas and changing the formula for automatic trading suspension during exceptional events to avoid sharp drops in the index. Increasing the minimum amount for block trades can help boost trading and modifying stock exchange order provisions to make them clearer for investors. These actions can enhance liquidity, transparency, and transaction integrity in line with international standards. Stock market authorities should also protect investors’ savings by regulating markets to prevent manipulation, monitoring financial information dissemination, and penalizing violators. The Tunisian financial market needs reorganization and a new legal framework to encourage family businesses to open their capital and stimulate the primary and secondary markets. Favoring stock markets’ digitalization as a part of technology advancement should be a current preoccupation. This helps to resist potential shocks.
Advanced analytical tools such as GARCH-MIDAS could be employed to further explore the impact of other relevant factors such as governance quality, the overall economic resilience, etc. on stock market dynamics. Further research about investors’ sentiments and attention should be considered from a game theoretic approach (Trabelsi 2024) with a setup applied to the context of an emerging market. Then, data (when available) should be processed from an empirical point of view. One direction could be the construction of a “fake news” index in addition to the “fear” or a “sentiment” index and examining their impacts on stock market performance.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/jrfm17090403/s1.

Funding

This research received no external funding.

Data Availability Statement

Data are accessible upon request from the author.

Conflicts of Interest

The author has no competing interests.

Note

1
Lassoued et al. (2020) provided a very comprehensive timeline of the pandemic’s spread in Tunisia.

References

  1. Akhtaruzzaman, Med, Sabri Boubaker, and Ahmet Sensoy. 2021. Financial contagion during COVID–19 crisis. Finance Research Letters 38: 101604. [Google Scholar] [CrossRef] [PubMed]
  2. Akinlaso, Mustapha Ishaq, Aroua Robbana, and Nura Mohamed. 2022. Volatility transmission within financial markets during the COVID-19 pandemic: Are faith-based investors well off in Tunisia? Journal of Islamic Accounting and Business Research 13: 98–113. [Google Scholar] [CrossRef]
  3. Albulescu, Claudiu Tiberiu. 2021. COVID-19 and the United States financial markets’ volatility. Finance Research Letters 38: 101699. [Google Scholar] [CrossRef]
  4. Alfano, Vincenzo, and Salvatore Ercolano. 2020. The efficacy of lockdown against COVID-19: A cross-country panel analysis. Applied Health Economics and Health Policy 18: 509–17. [Google Scholar] [CrossRef] [PubMed]
  5. Alqahtani, Abdullah, Michael J. Wither, Zhankui Dong, and K. R. Goodwin. 2020. Impact of news-based equity market volatility on international stock markets. Journal of Applied Economics 23: 224–34. [Google Scholar] [CrossRef]
  6. Andersen, Torben G., and Timo Teräsvirta. 2009. Realized volatility. In Handbook of Financial Time Series. Berlin and Heidelberg: Springer, pp. 555–75. Available online: https://link.springer.com/chapter/10.1007/978-3-540-71297-8_24 (accessed on 2 October 2023).
  7. Angeletos, George Marios, and Alessandro Pavan. 2007. Efficient use of information and social value of information. Econometrica 75: 1103–42. [Google Scholar] [CrossRef]
  8. Arcuri, Maria Cristina, Gino Gandolfi, and Ivan Russo. 2023. Does fake news impact stock returns? Evidence from US and EU stock markets. Journal of Economics and Business 125: 106130. [Google Scholar] [CrossRef]
  9. Asgharian, Hossein, Charlotte Christiansen, and Ai Jun Hou. 2023. The effect of uncertainty on stock market volatility and correlation. Journal of Banking & Finance 154: 106929. [Google Scholar] [CrossRef]
  10. Ashraf, Badar Nadeem. 2020. Economic impact of government interventions during the COVID-19 pandemic: International evidence from financial markets. Journal of Behavioral and Experimental Finance 27: 100371. [Google Scholar] [CrossRef]
  11. Aslam, Faheem, Amir Rafique, Aneel Salman, Hyong Goo Kang, and Wahbeeah Mohti. 2018. The impact of terrorism on financial markets: Evidence from Asia. The Singapore Economic Review 63: 1183–204. [Google Scholar] [CrossRef]
  12. Aslam, Faheem, Chanyoung Eom, and Hyong Goo Kang. 2014. How Terrorist Attacks Affect the Volatility of the Pakistani Stock Market? 남아시아연구 19: 151–81. [Google Scholar]
  13. Aslam, Faheen, Amir Rasheed, and Masood Hassan. 2020. The Effect of Terrorism on Capital Market Returns: An Empirical Analysis of Emerging Market. Paradigms S1: 39–46. [Google Scholar]
  14. Baek, Seungho, Sunil K. Mohanty, and Mina Glambosky. 2020. COVID-19 and stock market volatility: An industry level analysis. Finance Research Letters 37: 101748. [Google Scholar] [CrossRef] [PubMed]
  15. Baker, Scott R., Nicholas Bloom, and Steven J. Davis. 2016. Measuring economic policy uncertainty. The Quarterly Journal of Economics 131: 1593–636. [Google Scholar] [CrossRef]
  16. Baker, Scott R., Nicholas Bloom, Steven J. Davis, and Kyle J. Kost. 2019. Policy News and Stock Market Volatility (No. w25720). National Bureau of Economic Research. Available online: https://www.nber.org/papers/w25720 (accessed on 2 October 2023).
  17. Bakry, Walid, Peter John Kavalmthara, Vivienne Saverimuttu, Yiyang Liu, and Sojan Cyril. 2022. Response of stock market volatility to COVID-19 announcements and stringency measures: A comparison of developed and emerging markets. Finance Research Letters 46: 102350. [Google Scholar] [CrossRef] [PubMed]
  18. Barber, Brad M., and Terrance Odean. 2001. The internet and the investor. Journal of Economic Perspectives 15: 41–54. Available online: https://pubs.aeaweb.org/doi/pdf/10.1257/jep.15.1.41 (accessed on 2 October 2023).
  19. Barrett, W. Brian, Andrea J. Heuson, Robert W. Kolb, and Gabriele H. Schropp. 1987. The adjustment of stock prices to completely unanticipated events. Financial Review 22: 345–54. [Google Scholar] [CrossRef]
  20. Ben Ayed, Wassim. 2022. The Tunisian stock market before invoking Article 80 of the Constitution: The (in) direct impact of government interventions during the sanitary crisis. Journal of Business and Socio-economic Development 4: 20–36. [Google Scholar] [CrossRef]
  21. Bera, Anil K., and Matthew L. Higgins. 1993. ARCH models: Properties, estimation and testing. Journal of Economic Surveys 7: 305–66. [Google Scholar] [CrossRef]
  22. Binder, John J., and Matthias J. Merges. 2001. Stock market volatility and economic factors. Review of Quantitative Finance and Accounting 17: 5–26. Available online: https://link.springer.com/article/10.1023/A:1011207919894 (accessed on 2 October 2023).
  23. Biu, Gilberth S., and Prambudi K. Kusuma. 2023. Stock Market Volatility Analysis During the Global Financial Crisis: Literature Review. Riwayat: Educational Journal of History and Humanities 6: 2510–20. Available online: https://jurnal.usk.ac.id/riwayat/article/view/34539 (accessed on 2 October 2023).
  24. Bollerslev, Tim. 1986. Generalized autoregressive conditional heteroskedasticity. Journal of Econometrics 31: 307–27. [Google Scholar] [CrossRef]
  25. Bollerslev, Tim, and Hans Ole Mikkelsen. 1996. Modeling and pricing long memory in stock market volatility. Journal of Econometrics 73: 151–84. [Google Scholar] [CrossRef]
  26. Bollerslev, Tim, Ray Y. Chou, and Kenneth F. Kroner. 1992. ARCH modeling in finance: A review of the theory and empirical evidence. Journal of Econometrics 52: 5–59. [Google Scholar] [CrossRef]
  27. Bonaccorsi, Giovanni, Francesco Pierri, Matteo Cinelli, Andrea Flori, Alessandro Galeazzi, Fabio Porcelli, Ana Lucia Schmidt, Carlo Michele Valensise, Antonio Scala, Walter Quattrociocchi, and et al. 2020. Economic and social consequences of human mobility restrictions under COVID-19. Proceedings of the National Academy of Sciences of the United States of America 117: 15530–35. Available online: https://www.pnas.org/doi/abs/10.1073/pnas.2007658117 (accessed on 2 October 2023).
  28. Bora, Debakchi, and Daisy Basistha. 2021. The outbreak of COVID-19 pandemic and its impact on stock market volatility: Evidence from a worst-affected economy. Journal of Public Affairs 21: e2623. [Google Scholar] [CrossRef] [PubMed]
  29. Bourdeau-Brien, Michael, and Lawrence Kryzanowski. 2017. The impact of natural disasters on the stock returns and volatilities of local firms. The Quarterly Review of Economics and Finance 63: 259–70. [Google Scholar] [CrossRef]
  30. Cazelles, Bernard, Mario Chavez, Dominique Berteaux, Frédéric Ménard, Jon Olav Vik, Stéphanie Jenouvrier, and Nil C. Stenseth. 2008. Wavelet analysis of ecological time series. Oecologia 156: 287–304. [Google Scholar] [CrossRef] [PubMed]
  31. Chari, S., Purva Hegde-Desai, and Nilesh Borde. 2017. A Review of Literature on Short Term Overreaction Generated by News Sentiment in Stock Market. Available online: http://irgu.unigoa.ac.in/drs/handle/unigoa/4828 (accessed on 2 October 2023).
  32. Chaudhary, Rashmi, Priti Bakhshi, and Hemendra Gupta. 2020. Volatility in international stock markets: An empirical study during COVID-19. Journal of Risk and Financial Management 13: 208. [Google Scholar] [CrossRef]
  33. Chowdhury, Emon K., Bablu K. Dhar, and Alessandro Stasi. 2022. Volatility of the US stock market and business strategy during COVID-19. Business Strategy Development 5: 350–60. [Google Scholar] [CrossRef]
  34. Christoffersen, Peter, and Xihui N. Pan. 2018. Oil volatility risk and expected stock returns. Journal of Banking Finance 95: 5–26. [Google Scholar] [CrossRef]
  35. De Bondt, Werner F. 1989. Stock Price Reversals and Overreaction to News Events: A Survey of Theory and Evidence. Berlin and Heidelberg: Springer, pp. 57–84. [Google Scholar]
  36. De Bondt, Werner F., and R. Thaler. 1985. Does the stock market overreact? The Journal of Finance 40: 793–805. [Google Scholar] [CrossRef]
  37. Dhanya, C. T., and Abhishek Gupta. 2014. Improved bias correction method using wavelet decomposition. The International Journal of Civil Engineering Research 5: 1–8. [Google Scholar]
  38. Dhingra, Barkha, Shallu Batra, Vaibhav Aggarwal, Mahender Yadav, and Pankaj Kumar. 2023. Stock market volatility: A systematic review. Journal of Modelling in Management 19: 925–52. [Google Scholar] [CrossRef]
  39. Dissanaike, G. 1997. Do stock market investors overreact? Journal of Business Finance Accounting 24: 27–50. [Google Scholar] [CrossRef]
  40. Ebens, Heiko. 1999. Realized Stock Volatility. Department of Economics, Johns Hopkins University. Available online: https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=f429a022f2a40d2878b05cc4570bcb696bde2f85 (accessed on 2 October 2023).
  41. Endri, Endri, Widya Aipama, and Renil Septiano. 2021. Stock price volatility during the COVID-19 pandemic: The GARCH model. Investment Management Financial Innovations 18: 12. [Google Scholar] [CrossRef]
  42. Engle, Robert F. 1982. Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom inflation. Econometrica: Journal of the Econometric Society 50: 987–1007. [Google Scholar] [CrossRef]
  43. Erdem, Orhan. 2020. Freedom and stock market performance during Covid-19 outbreak. Finance Research Letters 36: 101671. [Google Scholar] [CrossRef] [PubMed]
  44. Fakhfekh, Mohamed, Ahmed Jeribi, and Marwa Ben Salem. 2023. Volatility dynamics of the Tunisian stock market before and during the COVID-19 outbreak: Evidence from the GARCH family models. International Journal of Finance Economics 28: 1653–66. [Google Scholar] [CrossRef]
  45. Fama, Eugene F. 1965. The behavior of stock-market prices. The Journal of Business 38: 34–105. [Google Scholar] [CrossRef]
  46. Ferreira, Susana, and B. Karali. 2015. Do earthquakes shake stock markets? PLoS ONE 10: e0133319. [Google Scholar] [CrossRef]
  47. Fleming, Jeff, Chris Kirby, and Barbara Ostdiek. 2003. The economic value of volatility timing using “realized” volatility. Journal of Financial Economics 67: 473–509. [Google Scholar] [CrossRef]
  48. Gao, Xue, Yixin Ren, and Muhammad Umar. 2022. To what extent does COVID-19 drive stock market volatility? A comparison between the US and China. Economic Research-Ekonomska Istraživanja 35: 1686–706. [Google Scholar] [CrossRef]
  49. Gherghina, Ștefan C., Daniel Ș. Armeanu, and Camelia C. Joldeș. 2021. COVID-19 pandemic and Romanian stock market volatility: A GARCH approach. Journal of Risk and Financial Management 14: 341. [Google Scholar] [CrossRef]
  50. Greenstone, Michael, and Vishan Nigam. 2020. Does Social Distancing Matter? University of Chicago, Becker Friedman Institute for Economics Working Paper, (2020–2026). Available online: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3561244 (accessed on 2 October 2023).
  51. Grinsted, Aslak, John C. Moore, and Svetlana Jevrejeva. 2004. Application of the cross wavelet transform and wavelet coherence to geophysical time series. Nonlinear Processes in Geophysics 11: 561–66. [Google Scholar] [CrossRef]
  52. Grossman, Sanford J., and Joseph E. Stiglitz. 1980. On the impossibility of informationally efficient markets. The American Economic Review 70: 393–408. Available online: https://www.jstor.org/stable/1805228 (accessed on 2 October 2023).
  53. Hanabusa, Kunihiru. 2010. Effects of foreign disasters on the petroleum industry in Japan: A financial market perspective. Energy 35: 5455–63. [Google Scholar] [CrossRef]
  54. He, Xi, Eric H. Lau, P. Wu, Xilong Deng, Jian Wang, X. Hao, Yiu Chung Lau, Jessica Y. Wong, Yujuan Guan, Xinghua Tan, and et al. 2020. Temporal dynamics in viral shedding and transmissibility of COVID-19. Nature Medicine 26: 672–675. Available online: https://www.nature.com/articles/s41591-020-0869-5 (accessed on 2 October 2023).
  55. Hendricks, Kevin B., Brian W. Jacobs, and Vinod R. Singhal. 2020. Stock market reaction to supply chain disruptions from the 2011 Great East Japan Earthquake. Manufacturing Service Operations Management 22: 683–99. [Google Scholar] [CrossRef]
  56. Hewamana, Ranil, Dinesha Siriwardhane, and Rmak Rathnayake. 2022. Determinants of Stock Price Volatility: A Literature Review. Available online: http://dr.lib.sjp.ac.lk/handle/123456789/12689 (accessed on 2 October 2023).
  57. Hosseini Baghanam, Aida, Vahid Nourani, Ehsan Norouzi, Amirezza T. Vakili, and Hüseyin Gökçekuş. 2023. Application of Wavelet Transform for Bias Correction and Predictor Screening of Climate Data. Sustainability 15: 15209. [Google Scholar] [CrossRef]
  58. Hussain, Sartaj, K. V. Bhanu Murthy, and Amit Singh. 2019. Stock market volatility: A review of the empirical literature. IUJ Journal of Management 7: 1–14. Available online: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3418148 (accessed on 2 October 2023).
  59. Insaidoo, Michael, Lilian Arthur, Samuel Amoako, and Francis K. Andoh. 2021. Stock market performance and COVID-19 pandemic: Evidence from a developing economy. Journal of Chinese Economic and Foreign Trade Studies 14: 60–73. [Google Scholar] [CrossRef]
  60. Izzeldin, Marwan, Yaz G. Muradoğlu, Vasileis Pappas, and Sheeja Sivaprasad. 2021. The impact of COVID-19 on G7 stock markets volatility: Evidence from a ST-HAR model. International Review of Financial Analysis 74: 101671. [Google Scholar] [CrossRef]
  61. Jeribi, Ahmed, Mohmaed Fakhfekh, and Anis Jarboui. 2015. Tunisian revolution and stock market volatility: Evidence from FIEGARCH model. Managerial Finance 41: 1112–35. [Google Scholar] [CrossRef]
  62. Kamal, Joved B., and Mark Wohar. 2023. Heterogenous responses of stock markets to covid related news and sentiments: Evidence from the 1st year of pandemic. International Economics 173: 68–85. [Google Scholar] [CrossRef]
  63. Kollias, Christos, Stephanos Papadamou, and Vangelis Arvanitis. 2013. Does terrorism affect the stock-bond covariance? Evidence from European countries. Southern Economic Journal 79: 832–48. [Google Scholar] [CrossRef]
  64. Kowalewski, Oskar, and Piotr Śpiewanowski. 2020. Stock market response to potash mine disasters. Journal of Commodity Markets 20: 100124. [Google Scholar] [CrossRef]
  65. Kusumahadi, Teresia A., and Fikri C. Permana. 2021. Impact of COVID-19 on global stock market volatility. Journal of Economic Integration 36: 20–45. Available online: https://www.jstor.org/stable/26985574 (accessed on 2 October 2023).
  66. Lassoued, Abira, Afef B. Saad, Hela Lassoued, Raouf Ketata, and Olfa Boubaker. 2020. Dataset on the COVID-19 Pandemic Situation in Tunisia with Application to SIR Model. medRxiv. Available online: https://www.medrxiv.org/content/10.1101/2020.04.23.20076802.abstract (accessed on 2 October 2023).
  67. Li, Tao, Feng Ma, Xuehua Zhang, and Yaojie Zhang. 2020. Economic policy uncertainty and the Chinese stock market volatility: Novel evidence. Economic Modelling 87: 24–33. [Google Scholar] [CrossRef]
  68. Li, Weiqing, Fengsheng Chien, Hafiz Waqas Kamran, Talla M. Aldeehani, Muhammad Sadiq, Van C. Nguyen, and Farhad Taghizadeh-Hesary. 2022. The nexus between COVID-19 fear and stock market volatility. Economic Research-Ekonomska Istraživanja 35: 1765–85. Available online: https://hrcak.srce.hr/file/435909 (accessed on 2 October 2023).
  69. Liu, Yonggang, Geng Peng, Lanyi Hu, Jichang Dong, and Qinqing Zhang. 2020. Using Google trends and Baidu index to analyze the impacts of disaster events on company stock prices. Industrial Management Data Systems 120: 350–65. [Google Scholar] [CrossRef]
  70. Liu, Yonggang, X. San Liang, and Robert H. Weisberg. 2007. Rectification of the bias in the wavelet power spectrum. Journal of Atmospheric and Oceanic Technology 24: 2093–102. [Google Scholar] [CrossRef]
  71. Lúcio, Francisco, and Jorge Caiado. 2022. COVID-19 and stock market volatility: A clustering approach for S&P 500 industry indices. Finance Research Letters 49: 103141. [Google Scholar] [CrossRef]
  72. Mazur, Mieszko, Man Dang, and Miguel Vega. 2021. COVID-19 and the march 2020 stock market crash. Evidence from S&P1500. Finance Research Letters 38: 101690. [Google Scholar] [CrossRef]
  73. Mazzucato, Mariana, and Willi Semmler. 2002. The determinants of stock price volatility: An industry study. Nonlinear Dynamics, Psychology, and Life Sciences 6: 197–216. [Google Scholar] [CrossRef]
  74. Megaritis, Anastasios, Nikolaos Vlastakis, and Athanasios Triantafyllou. 2021. Stock market volatility and jumps in times of uncertainty. Journal of International Money and Finance 113: 102355. [Google Scholar] [CrossRef]
  75. Mestiri, Sami. 2022. Modeling the volatility of Bitcoin returns using Nonparametric GARCH models. Journal of Academic Finance 13: 2–16. [Google Scholar] [CrossRef]
  76. Morris, Stephen, and Hyun Song Shin. 2002. Social value of public information. American Economic Review 92: 1521–34. [Google Scholar] [CrossRef]
  77. Mouselli, Sulaiman, and Razi Mohialdeen. 2021. The Impact of Corona Pandemic on the Volatility of Damascus Securities Exchange Index. Journal of Hama University 4: 1–16. Available online: https://hama-univ.edu.sy/ojs/index.php/huj/article/download/788/657 (accessed on 2 October 2023).
  78. Nelson, Daniel B. 1991. Conditional heteroskedasticity in asset returns: A new approach. Econometrica: Journal of the Econometric Society 59: 347–70. [Google Scholar] [CrossRef]
  79. Nikmanesh, Lida, and Abu Hassan S. M. Nor. 2016. Macroeconomic determinants of stock market volatility: An empirical study of Malaysia and Indonesia. Asian Academy of Management Journal 21: 161. [Google Scholar]
  80. Okorie, David I., and Boqiang Lin. 2021. Stock markets and the COVID-19 fractal contagion effects. Finance Research Letters 38: 101640. [Google Scholar] [CrossRef] [PubMed]
  81. Onali, Enrico. 2020. COVID-19 and Stock Market Volatility. Available online: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3573622 (accessed on 2 October 2023).
  82. Papadamou, Stephanos, Athanasios Fassas, Dimitris Kenourgios, and Dimitrios Dimitriou. 2020. Direct and Indirect Effects of COVID-19 Pandemic on Implied Stock Market Volatility: Evidence from Panel Data Analysis. Available online: https://mpra.ub.uni-muenchen.de/100020/ (accessed on 2 October 2023).
  83. Robinson, C. Justin, and Prosper Bangwayo-Skeete. 2016. The Financial Impact of Natural Disasters: Assessing the Effect of Hurricanes & Tropical Storms On Stock Markets in the Caribbean. Available online: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2845429 (accessed on 2 October 2023).
  84. Rouyer, Tristan, Jean Marc Fromentin, Nils C. Stenseth, and Bernard Cazelles. 2008. Analysing multiple time series and extending significance testing in wavelet analysis. Marine Ecology Progress Series 359: 11–23. [Google Scholar] [CrossRef]
  85. Sadiq, Muhammad, C. C. Hsu, YunQian Zhang, and Fengsheng Chien. 2021. COVID-19 fear and volatility index movements: Empirical insights from ASEAN stock markets. Environmental Science and Pollution Research 28: 67167–84. [Google Scholar] [CrossRef] [PubMed]
  86. Scherf, Mattias, Xena Matschke, and Marc O. Rieger. 2022. Stock market reactions to COVID-19 lockdown: A global analysis. Finance Research Letters 45: 102245. [Google Scholar] [CrossRef] [PubMed]
  87. Solnik, Bruno H. 1974. An equilibrium model of the international capital market. Journal of Economic Theory 8: 500–24. [Google Scholar] [CrossRef]
  88. Su, Zhi, Tong Fang, and Libo Yin. 2019. Understanding stock market volatility: What is the role of US uncertainty? The North American Journal of Economics and Finance 48: 582–90. [Google Scholar] [CrossRef]
  89. Szczygielski, Jan J., Ailie Charteris, Princess R. Bwanya, and Janusz Brzeszczyński. 2024. Google search trends and stock markets: Sentiment, attention or uncertainty? International Review of Financial Analysis 91: 102549. [Google Scholar] [CrossRef]
  90. Tisdell, Clement A. 2020. Economic, social and political issues raised by the COVID-19 pandemic. Economic Analysis and Policy 68: 17–28. [Google Scholar] [CrossRef] [PubMed]
  91. Tissaoui, Kais, Besma Hkiri, Mariem Talbi, Waleed Alghassab, and Khaled I. Alfreahat. 2021. Market volatility and illiquidity during the COVID-19 outbreak: Evidence from the Saudi stock exchange through the wavelet coherence approaches. The North American Journal of Economics and Finance 58: 101521. [Google Scholar] [CrossRef]
  92. Tissaoui, Kais, Besma Hkiri, Taha Zaghdoudi, and Jamel Azibi. 2022. Illiquidity, Uncertainty Indices, and COVID-19 Outbreak Conditions: Empirical Evidence from the US Financial Market. Complexity 2022: 1–23. Available online: https://downloads.hindawi.com/journals/complexity/2022/2818633.pdf (accessed on 2 October 2023).
  93. Torrence, Christopher, and Peter J. Webster. 1998. The annual cycle of persistence in the El Nño/Southern Oscillation. Quarterly Journal of the Royal Meteorological Society 124: 1985–2004. [Google Scholar] [CrossRef]
  94. Trabelsi, Emna. 2024. Welfare fragmented information effects: The cost-benefit analysis and Trade-offs. Journal of Information Economics 2: 22. [Google Scholar] [CrossRef]
  95. Uddin, Moshfique, Anup Chowdhury, Keith Anderson, and Kausik Chaudhuri. 2021. The effect of COVID–19 pandemic on global stock market volatility: Can economic strength help to manage the uncertainty? Journal of Business Research 128: 31–44. [Google Scholar] [CrossRef] [PubMed]
  96. Veleda, Doris, Raul Montagne, and Moacyr Araujo. 2012. Cross-wavelet bias corrected by normalizing scales. Journal of Atmospheric and Oceanic Technology 29: 1401–8. [Google Scholar] [CrossRef]
  97. Veronesi, Pietro. 1999. Stock market overreactions to bad news in good times: A rational expectations equilibrium model. The Review of Financial Studies 12: 975–1007. [Google Scholar] [CrossRef]
  98. Wang, Jun, Zhilong Xie, Qing Li, Jinghua Tan, Rong Xing, Yuanzhu Chen, and Fengyun Wu. 2019. Effect of digitalized rumor clarification on stock markets. Emerging Markets Finance and Trade 55: 450–74. [Google Scholar] [CrossRef]
  99. Worthington, Andrew C. 2008. The impact of natural events and disasters on the Australian stock market: A GARCH-M analysis of storms, floods, cyclones, earthquakes and bushfires. Global Business and Economics Review 10: 1–10. [Google Scholar] [CrossRef]
  100. Yong, Jordan N. C., Sayyed M. Ziaei, and Kenneth R. Szulczyk. 2021. The impact of covid-19 pandemic on stock market return volatility: Evidence from Malaysia and Singapore. Asian Economic and Financial Review 11: 191. Available online: https://www.researchgate.net/profile/Sayyed-Ziaei/publication/349763199_The_Impact_of_COVID-19_Pandemic_on_Stock_Market_Return_Volatility_Evidence_from_Malaysia_and_Singapore/links/6044cd4aa6fdcc9c781b7080/The-Impact-of-COVID-19-Pandemic-on-Stock-Market-Return-Volatility-Evidence-from-Malaysia-and-Singapore.pdf (accessed on 2 October 2023).
  101. Yousef, Ibrahim. 2020. Spillover of COVID-19: Impact on stock market volatility. International Journal of Psychosocial Rehabilitation 24: 18069–81. [Google Scholar]
  102. Zaremba, Adam, Renatas Kizys, David Y. Aharon, and Ender Demir. 2020. Infected markets: Novel coronavirus, government interventions, and stock return volatility around the globe. Finance Research Letters 35: 101597. [Google Scholar] [CrossRef] [PubMed]
  103. Zhang, Dayong, Min Hu, and Qiang Ji. 2020. Financial markets under the global pandemic of COVID-19. Finance Research Letters 36: 101528. [Google Scholar] [CrossRef] [PubMed]
Jrfm 17 00403 g001
Figure 1. Evolution of TUNINDEX stock return (2 January 2020–30 December 2022). Source: The Author.
Figure 1. Evolution of TUNINDEX stock return (2 January 2020–30 December 2022). Source: The Author.
Jrfm 17 00403 g001
Jrfm 17 00403 g002
Figure 2. Wavelet transform coherence: Tunisian stock return volatility versus COVID-19 cases rate. Notes: The black contour shows where the spectrum is significantly different from red noise at the 5% level. The lighter shade represents the cone of influence, marking high-power areas and indicating the autocorrelation of wavelet power at each scale. The horizontal axis represents time from 3 March 2020 to 30 December 2022, and the vertical axis denotes scale bands with daily frequency. Arrows to the right (left) indicate in-phase (out-of-phase) relationships, meaning a positive (negative) connection. If arrows move right and up (down), the first variable “m” (cases rate) drives (follows), while if arrows move left and up (down), variable “n” (TUNINDEX volatility) leads (lags). This visualization helps understand the dynamic relationships between variables. Source: The Author.
Figure 2. Wavelet transform coherence: Tunisian stock return volatility versus COVID-19 cases rate. Notes: The black contour shows where the spectrum is significantly different from red noise at the 5% level. The lighter shade represents the cone of influence, marking high-power areas and indicating the autocorrelation of wavelet power at each scale. The horizontal axis represents time from 3 March 2020 to 30 December 2022, and the vertical axis denotes scale bands with daily frequency. Arrows to the right (left) indicate in-phase (out-of-phase) relationships, meaning a positive (negative) connection. If arrows move right and up (down), the first variable “m” (cases rate) drives (follows), while if arrows move left and up (down), variable “n” (TUNINDEX volatility) leads (lags). This visualization helps understand the dynamic relationships between variables. Source: The Author.
Jrfm 17 00403 g002
Jrfm 17 00403 g003
Figure 3. Wavelet transform coherence: Tunisian stock return volatility versus COVID-19 death rate. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 23 March 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 3. Wavelet transform coherence: Tunisian stock return volatility versus COVID-19 death rate. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 23 March 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g003
Jrfm 17 00403 g004
Figure 4. Wavelet transform coherence: Tunisian stock return volatility versus stringency index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 4. Wavelet transform coherence: Tunisian stock return volatility versus stringency index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g004
Jrfm 17 00403 g005
Figure 5. Wavelet transform coherence: Tunisian stock return volatility versus containment health index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 5. Wavelet transform coherence: Tunisian stock return volatility versus containment health index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g005
Jrfm 17 00403 g006
Figure 6. Wavelet transform coherence: Tunisian stock return volatility versus economic support index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 6. Wavelet transform coherence: Tunisian stock return volatility versus economic support index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g006
Jrfm 17 00403 g007
Figure 7. Wavelet transform coherence: Tunisian stock return volatility versus government response index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 7. Wavelet transform coherence: Tunisian stock return volatility versus government response index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g007
Jrfm 17 00403 g008
Figure 8. Wavelet transform coherence: Tunisian stock return volatility versus school closing index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 8. Wavelet transform coherence: Tunisian stock return volatility versus school closing index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g008
Jrfm 17 00403 g009
Figure 9. Wavelet transform coherence: Tunisian stock return volatility versus workplace closing. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 9. Wavelet transform coherence: Tunisian stock return volatility versus workplace closing. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g009
Jrfm 17 00403 g010
Figure 10. Wavelet transform coherence: Tunisian stock return volatility versus public events canceling. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 10. Wavelet transform coherence: Tunisian stock return volatility versus public events canceling. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g010
Jrfm 17 00403 g011
Figure 11. Wavelet transform coherence: Tunisian stock return volatility versus public gathering restrictions. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 11. Wavelet transform coherence: Tunisian stock return volatility versus public gathering restrictions. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g011
Jrfm 17 00403 g012
Figure 12. Wavelet transform coherence: Tunisian stock return volatility versus public transport closure. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 12. Wavelet transform coherence: Tunisian stock return volatility versus public transport closure. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g012
Jrfm 17 00403 g013
Figure 13. Wavelet transform coherence: Tunisian stock return volatility versus stay-at-home requirements. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 13. Wavelet transform coherence: Tunisian stock return volatility versus stay-at-home requirements. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g013
Jrfm 17 00403 g014
Figure 14. Wavelet transform coherence: Tunisian stock return volatility versus internal movement restrictions. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 14. Wavelet transform coherence: Tunisian stock return volatility versus internal movement restrictions. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g014
Jrfm 17 00403 g015
Figure 15. Wavelet transform coherence: Tunisian stock return volatility versus international travel control. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 15. Wavelet transform coherence: Tunisian stock return volatility versus international travel control. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g015
Jrfm 17 00403 g016
Figure 16. Wavelet transform coherence: Tunisian stock return volatility versus public information gathering. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 16. Wavelet transform coherence: Tunisian stock return volatility versus public information gathering. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g016
Jrfm 17 00403 g017
Figure 17. Wavelet transform coherence: Tunisian realized volatility versus the COVID-19 cases rate. Notes: The black contour shows where the spectrum is significantly different from red noise at the 5% level. The lighter shade represents the cone of influence, marking high-power areas, indicating autocorrelation of wavelet power at each scale. The horizontal axis represents time from 3 March 2020 to 30 December 2022, and the vertical axis denotes scale bands with daily frequency. Arrows to the right (left) indicate in-phase (out-of-phase) relationships, meaning a positive (negative) connection. If arrows move right and up (down), the first variable “m” (cases rate) drives (follows), while if arrows move left and up (down), variable “n” (TUNINDEX volatility) leads (lags). This visualization helps understand the dynamic relationships between variables. Source: The Author.
Figure 17. Wavelet transform coherence: Tunisian realized volatility versus the COVID-19 cases rate. Notes: The black contour shows where the spectrum is significantly different from red noise at the 5% level. The lighter shade represents the cone of influence, marking high-power areas, indicating autocorrelation of wavelet power at each scale. The horizontal axis represents time from 3 March 2020 to 30 December 2022, and the vertical axis denotes scale bands with daily frequency. Arrows to the right (left) indicate in-phase (out-of-phase) relationships, meaning a positive (negative) connection. If arrows move right and up (down), the first variable “m” (cases rate) drives (follows), while if arrows move left and up (down), variable “n” (TUNINDEX volatility) leads (lags). This visualization helps understand the dynamic relationships between variables. Source: The Author.
Jrfm 17 00403 g017
Jrfm 17 00403 g018
Figure 18. Wavelet transform coherence: Tunisian realized volatility versus COVID-19 death rate. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 23 March 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 18. Wavelet transform coherence: Tunisian realized volatility versus COVID-19 death rate. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 23 March 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g018
Jrfm 17 00403 g019
Figure 19. Wavelet transform coherence: Tunisian realized volatility versus stringency index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 19. Wavelet transform coherence: Tunisian realized volatility versus stringency index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g019
Jrfm 17 00403 g020
Figure 20. Wavelet transform coherence: Tunisian realized volatility versus containment health index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 20. Wavelet transform coherence: Tunisian realized volatility versus containment health index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g020
Jrfm 17 00403 g021
Figure 21. Wavelet transform coherence: Tunisian realized volatility versus economic support index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 21. Wavelet transform coherence: Tunisian realized volatility versus economic support index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g021
Jrfm 17 00403 g022
Figure 22. Tunisian realized volatility versus government response index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 22. Tunisian realized volatility versus government response index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g022
Jrfm 17 00403 g023
Figure 23. Wavelet transform coherence: Tunisian realized volatility versus school closing index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 23. Wavelet transform coherence: Tunisian realized volatility versus school closing index. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g023
Jrfm 17 00403 g024
Figure 24. Wavelet transform coherence: Tunisian realized volatility versus workplace closing. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 24. Wavelet transform coherence: Tunisian realized volatility versus workplace closing. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g024
Jrfm 17 00403 g025
Figure 25. Wavelet transform coherence: Tunisian stock return volatility versus public events canceling. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 25. Wavelet transform coherence: Tunisian stock return volatility versus public events canceling. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g025
Jrfm 17 00403 g026
Figure 26. Wavelet transform coherence: Tunisian realized volatility versus public gathering restrictions. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 26. Wavelet transform coherence: Tunisian realized volatility versus public gathering restrictions. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g026
Jrfm 17 00403 g027
Figure 27. Wavelet transform coherence: Tunisian realized volatility versus public transport closure. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 27. Wavelet transform coherence: Tunisian realized volatility versus public transport closure. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g027
Jrfm 17 00403 g028
Figure 28. Wavelet transform coherence: Tunisian realized volatility versus stay-at-home requirements. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 28. Wavelet transform coherence: Tunisian realized volatility versus stay-at-home requirements. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g028
Jrfm 17 00403 g029
Figure 29. Wavelet transform coherence: Tunisian realized volatility versus internal movement restrictions. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 29. Wavelet transform coherence: Tunisian realized volatility versus internal movement restrictions. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g029
Jrfm 17 00403 g030
Figure 30. Wavelet transform coherence: Tunisian realized volatility versus international travel control. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 30. Wavelet transform coherence: Tunisian realized volatility versus international travel control. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g030
Jrfm 17 00403 g031
Figure 31. Wavelet transform coherence: Tunisian realized volatility versus public information gathering. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Figure 31. Wavelet transform coherence: Tunisian realized volatility versus public information gathering. Notes: The black contour identifies areas where the spectrum is statistically significant at the 5% level compared to red noise. The lighter shade designates the cone of influence, outlining high-power regions. Time, ranging from 3 January 2020 to 30 December 2022, is represented on the horizontal axis, while the vertical axis denotes scale bands with daily frequency. Source: The Author.
Jrfm 17 00403 g031
Table 1. Data definitions.
Table 1. Data definitions.
VariableSymbolDefinitionSource
TUNINDEXtunThe daily closing price of the Tunisian market indexhttps://www.investing.com/indices/tunindex-historical-data, accessed on 1 December 2023
Total casestotal_casesCumulative number of the confirmed cases due to COVID-19Our World in Data https://ourworldindata.org/coronavirus/country/tunisia, accessed on 1 December 2023
Total deaths total_deathCumulative number of the confirmed death cases due to COVID-19
New casesnew_casesTotal number of the new confirmed cases due to COVID-19
New deaths new_deathTotal number of new confirmed death cases due to COVID-19
Rate of confirmed casescases_rateNews cases/Total casesCalculus
Rate of deathsdeath_rateDeath cases/Total deathsCalculus
Stringency indexstringency_indexRanges from 0 to 100, where higher values indicate stricter government policies and restrictions in response to the COVID-19 pandemicOxford COVID-19 Government Response Tracker (OxCGRT) https://github.com/OxCGRT/covid-policy-dataset, accessed on 1 December 2023
Containment health indexcontainement_health_indexTypically ranges from 0 to 100, where higher values indicate better virus containment
Economic policy indexeconomic_policy_indexRanges from 0 to 100, where higher values indicate a more accommodative economic policy stance and lower values indicate a more restrictive stance
Government response indexgovernment_response_indexRanges between 0 and 100. A higher score means a better country’s response to the pandemic
WTI crude oil pricewti_priceCrude oil price (United States)Energy Information Administration database https://www.eia.gov/dnav/pet/pet_pri_spt_s1_d.htm, accessed on 1 December 2023
Interest rateintnKey rate if interest Central bank of Tunisia https://www.bct.gov.tn/bct/siteprod/tableau_statistique_a.jsp?params=PL203260&la=AN, accessed on 1 December 2023
School closings1Sub-indicator 1 of the stringency index Oxford COVID-19 Government Response Tracker (OxCGRT) https://github.com/OxCGRT/covid-policy-dataset, accessed on 1 December 2023
Workplace closings2Sub-indicator 2 of the stringency index
Cancel public eventss3Sub-indicator 3 of the stringency index
Restrictions on public gatheringss4Sub indicator 4 of the stringency index
Closures of public transports5Sub-indicator 5 of the stringency index
Stay-at-home requirementss6Sub-indicator 6 of the stringency index
Restrictions on internal movementss7Sub indicator 7 of the stringency index
International travel controls8Sub-indicator 8 of the stringency index
Public information gatheringss9Sub indicator 9 of the stringency index
Source: The Author.
Table 2. Descriptive statistics of the variables.
Table 2. Descriptive statistics of the variables.
RETURN_TUNTOTAL_CASESTOTAL_DEATHSSTRINGENCY_INDEXS1S2S3S4S5S6S7S8S9CONTAINMENT_HEALTH_INDEXECONOMIC_SUPPORT_INDEXGOVERNMENT_RESPONSE_INDEXWTI_OIL_RETURN
Mean0.000150551,839.116,432.8549.033631.0340431.1446811.5971633.2624110.3078010.8085110.6397161.8765961.84680950.2839146.8439749.854030.001816
Median0.000339603,981.021,941.0045.310001.0000001.0000002.0000004.0000000.0000000.0000000.0000002.0000002.00000049.9600050.0000048.800000.002294
Maximum0.0189741,147,57129,284.0090.740003.0000003.0000002.0000004.0000002.0000002.0000002.0000004.0000002.00000077.7400075.0000077.400000.425832
Minimum−0.0418591.0000003.0000000.0000000.0000000.0000000.0000000.0000000.0000000.0000000.0000000.0000000.0000000.0000000.0000000.000000−0.281382
Std.Dev.0.004802444,625.611,959.8221.889171.0177280.8041460.7571571.5468320.6032670.9676210.9189361.1220060.49061618.0448825.6963517.329350.045112
Skewness−2.2034200.060338−0.263656−0.1193520.5468060.438110−1.480724−1.6322031.7990000.3900660.7721500.298314−3.148689−1.074264−0.287851−1.0554071.353766
Kurtosis20.121681.3932141.3310612.6711552.0972172.8430623.3840403.6729804.9933791.1836481.6329382.23149411.525944.1098551.7850514.54190129.12992
Jarque-Bera9181.80772.0480083.478094.85036859.0732623.27648261.9562326.3342497.0006114.7899124.953027.805393300.243171.783553.09627200.719320,271.78
Probability0.0000000.0000000.0000000.0884620.0000000.0000090.0000000.0000000.0000000.0000000.0000000.0000010.0000000.0000000.0000000.0000000.000000
Sum0.1057093.68 × 10810,747,08334,568.71729.0000807.00001126.0002300.000217.0000570.0000451.00001323.0001302.00035,450.1633,025.0035,147.091.280245
SumSq.Dev.0.0162351.31 × 10149.34 × 1010337,311.6729.1830455.2426403.59431684.454256.2071659.1489594.4879886.2638169.4553229,234.8464,852.8211,415.71.432724
Observations705666654705705705705705705705705705705705705705705
Source: The Author.
Table 3. Results of ARCH/GARCH family models for the TUNINDEX volatility and oil volatility.
Table 3. Results of ARCH/GARCH family models for the TUNINDEX volatility and oil volatility.
Panel A: Stock return volatility
ModelARCH(1), constantARCH(1), AR(1)ARCH(1), MA(1)ARCH(1), ARMA(1,1)GARCH(1,1), constantGARCH(1,1), AR(1)GARCH(1,1), MA(1)GARCH(1,1), ARMA(1,1)
AIC−8.248105−8.283928−8.274759−8.281654−8.332585−8.376225−8.365392−8.378094
SC−8.229721−8.259390−8.250246−8.250982−8.308073−8.345553−8.334751−8.341287
ModelEGARCH(1,1), constantEGARCH(1,1), AR(1)EGARCH(1,1), MA(1)EGARCH(1,1), ARMA(1,1)TGARCH(1,1), constantTGARCH(1,1), AR(1)TGARCH(1,1), MA(1)TGARCH(1,1), ARMA(1,1)
AIC−8.323690−8.369747−8.358517−8.369772−8.330636−8.374358−8.363444−8.376183
SC−8.293050−8.332940−8.321748−8.326830−8.299996−8.337551−8.326675−8.333242
Panel B: WTI oil volatility
ModelARCH(1), constantARCH(1), AR(1)ARCH(1), MA(1)ARCH(1), ARMA(1,1)GARCH(1,1), constantGARCH(1,1), AR(1)GARCH(1,1), MA(1)GARCH(1,1), ARMA(1,1)
AIC−3.856832−3.990224−4.003052−4.000366−4.272393−4.269562−4.269562−4.257070
SC−3.837436−3.964362−3.977190−3.968039−4.246531−4.237234−4.237235−4.231253
EGARCH(1,1), constantEGARCH(1,1), AR(1)EGARCH(1,1), MA(1)EGARCH(1,1), ARMA(1,1)TGARCH(1,1), constantTGARCH(1,1), AR(1)TGARCH(1,1), MA(1)TGARCH(1,1), ARMA(1,1)
AIC−4.278102−4.275790−4.275864−3.344925−4.292975−4.290166−4.290171−4.287428
SC−4.245774−4.236997−4.237071−3.299666−4.260648−4.251373−4.251377−4.242170
Note: Colored and bold columns pertain to the optimal model. Source: The Author.
Table 4. Effect of COVID-19 announcements and government intervention on TUNINDEX stock return volatility.
Table 4. Effect of COVID-19 announcements and government intervention on TUNINDEX stock return volatility.
Eq NameEQ01EQ02EQ03EQ04
cases_rate0.000357 ***0.000356 ***0.000364 ***0.000387 ***
[3.006][2.976][3.035][3.190]
death_rate0.0000490.0000530.0000360.000009
[0.554][0.599][0.404][0.100]
condvar_oil0.001428 ***0.001423 ***0.001423 ***0.001386 ***
[3.233][3.213][3.239][3.202]
dtintn−0.000022−0.000022−0.000023−0.000023
[−1.322][−1.319][−1.320][−1.314]
dstringencyindex0.000001
[1.147]
dcontainmenthealthindex 0.000001
[0.961]
dgovernmentresponseindex 0.000001
[1.156]
deconomicsupportindex 0.000001
[1.573]
_cons0.000011 ***0.000011 ***0.000011 ***0.000011 ***
[13.036][13.057][13.480][13.810]
N°observations654654654654
Note: t statistics in brackets. * p < 0.1, ** p < 0.05, *** p < 0.01. Source: The Author.
Table 5. Effect of stringency index sub-indicators on TUNINDEX stock return volatility.
Table 5. Effect of stringency index sub-indicators on TUNINDEX stock return volatility.
Eq NameEQ01EQ02EQ03EQ04EQ05EQ06EQ07EQ08EQ09
cases_rate0.000345 ***0.000352 ***0.000343 ***0.000365 ***0.000355 ***0.000366 ***0.000343 ***0.000346 ***0.000346 ***
[2.891][2.990][2.884][3.169][2.969][3.119][2.814][2.902][2.897]
death_rate0.0000780.0000590.0000760.0000240.0000610.0000410.0000820.0000770.000077
[0.889][0.691][0.864][0.277][0.716][0.510][0.904][0.880][0.881]
condvar_oil0.001410 ***0.001437 ***0.001421 ***0.001441 ***0.001408 ***0.001407 ***0.001410 ***0.001411 ***0.001410 ***
[3.160][3.240][3.202][3.301][3.182][3.192][3.153][3.163][3.162]
dtintn−0.000021−0.000022−0.000022−0.000023−0.000022−0.000022−0.000021−0.000021−0.000022
[−1.312][−1.319][−1.315][−1.319][−1.312][−1.321][−1.312][−1.311][−1.313]
ds1−0.000002
[−0.719]
ds2 0.000008
[1.313]
ds3 0.000005
[0.620]
ds4 0.000011 *
[1.656]
ds5 0.000014
[0.953]
ds6 0.000017 *
[1.726]
ds7 −0.000002
[−0.408]
ds8 −0.000002
[−0.936]
ds9 0.000004 ***
[5.676]
_cons0.000011 ***0.000011 ***0.000011 ***0.000011 ***0.000011 ***0.000011 ***0.000011 ***0.000011 ***0.000011 ***
[12.382][12.880][12.258][13.564][12.662][13.170][12.560][12.351][12.386]
N°observations654654654654654654654654654
Note: t statistics in brackets. * p < 0.1, ** p < 0.05, *** p < 0.01. Source: The Author.
Table 6. Effect of COVID-19 announcements and government intervention on TUNINDEX stock return TUNINDEX volatility, excluding cases_rate.
Table 6. Effect of COVID-19 announcements and government intervention on TUNINDEX stock return TUNINDEX volatility, excluding cases_rate.
Eq NameEQ01EQ02EQ03EQ04
death_rate0.000215 ***0.000220 ***0.000215 ***0.000209 ***
[5.759][5.682][5.316][4.828]
condvar_oil0.001973 ***0.001965 ***0.001970 ***0.001970 ***
[2.799][2.785][2.788][2.795]
dtintn−0.000028−0.000027−0.000028−0.000028
[−1.392][−1.391][−1.391][−1.390]
dstringencyindex0.000000
[0.701]
dcontainmenthealthindex 0.000000
[0.439]
dgovernmentresponseindex 0.000000
[0.619]
deconomicsupportindex 0.000000
[1.215]
_cons0.000012 ***0.000012 ***0.000012 ***0.000012 ***
[14.944][15.159][15.323][15.933]
N°observations654654654654
Note: t statistics in brackets. * p < 0.1, ** p < 0.05, *** p < 0.01. Source: The Author.
Table 7. Effect of stringency index sub-indicators on TUNINDEX stock return volatility, excluding cases_rate.
Table 7. Effect of stringency index sub-indicators on TUNINDEX stock return volatility, excluding cases_rate.
Eq NameEQ01EQ02EQ03EQ04EQ05EQ06EQ07EQ08EQ09
death_rate0.000226 ***0.000215 ***0.000222 ***0.000193 ***0.000224 ***0.000206 ***0.000238 ***0.000226 ***0.000226 ***
[6.233][5.829][6.233][4.742][5.767][5.692][5.896][6.174][6.171]
condvar_oil0.001955 ***0.001981 ***0.001967 ***0.002001 ***0.001958 ***0.001977 ***0.001941 ***0.001957 ***0.001957 ***
[2.783][2.812][2.805][2.843][2.782][2.813][2.756][2.788][2.786]
dtintn−0.000027−0.000028−0.000027−0.000029−0.000027−0.000028−0.000027−0.000027−0.000027
[−1.389][−1.391][−1.391][−1.389][−1.390][−1.392][−1.388][−1.389][−1.391]
ds1−0.000002
[−0.823]
ds2 0.000005
[1.143]
ds3 0.000008
[0.837]
ds4 0.000008
[1.495]
ds5 0.000002
[0.251]
ds6 0.000012
[1.440]
ds7 −0.000006
[−1.641]
ds8 −0.000001
[−0.514]
ds9 0.000006 ***
[9.493]
_cons0.000012 ***0.000012 ***0.000012 ***0.000012 ***0.000012 ***0.000012 ***0.000012 ***0.000012 ***0.000012 ***
[15.148][15.368][15.045][15.484][15.329][15.293][15.283][15.150][15.153]
N°observations654654654654654654654654654
Note: t statistics in brackets. * p < 0.1, ** p < 0.05, *** p < 0.01. Source: The Author.
Table 8. Effect of lagged COVID-19 announcements and government intervention on stock return volatility.
Table 8. Effect of lagged COVID-19 announcements and government intervention on stock return volatility.
Eq NameEQ01EQ02EQ03EQ04
L.cases_rate0.000293 ***0.000291 ***0.000296 ***0.000296 ***
[2.766][2.734][2.738][2.722]
L.death_rate0.0000620.0000680.0000590.000066
[0.945][1.022][0.804][0.860]
L.dtintn−0.000067−0.000067−0.000067−0.000067
[−0.995][−0.994][−0.995][−0.992]
L.condvar_oil0.001199 ***0.001193 ***0.001191 ***0.001171 ***
[3.051][3.020][3.014][2.954]
L.dstringencyindex0.000001
[1.027]
L.dcontainmenthealthindex 0.000001
[0.910]
L.dgovernmentresponseindex 0.000001
[1.000]
L.deconomicsupportindex 0.000000
[0.860]
_cons0.000011 ***0.000011 ***0.000011 ***0.000011 ***
[14.284][14.329][14.519][14.734]
N°observations653653653653
Note: t statistics in brackets. * p < 0.1, ** p < 0.05, *** p < 0.01. Source: The Author.
Table 9. Effect of lagged stringency index sub-indicators on TUNINDEX stock return volatility.
Table 9. Effect of lagged stringency index sub-indicators on TUNINDEX stock return volatility.
Eq NameEQ01EQ02EQ03EQ04EQ05EQ06EQ07EQ08EQ09
L.cases_rate0.000281 ***0.000287 ***0.000272 ***0.000303 ***0.000283 ***0.000292 ***0.000272 **0.000281 ***0.000281 ***
[2.680][2.753][2.681][2.979][2.673][2.788][2.533][2.685][2.684]
L.death_rate0.000092 *0.0000740.0000890.0000290.0000880.0000720.000109 *0.000092 *0.000092 *
[1.680][1.262][1.607][0.389][1.429][1.211][1.749][1.673][1.673]
L.dtintn−0.000066−0.000067−0.000067−0.000069−0.000067−0.000067−0.000066−0.000066−0.000066
[−0.990][−0.993][−0.993][−0.998][−0.990][−0.992][−0.988][−0.990][−0.990]
L.condvar_oil0.001180 ***0.001206 ***0.001217 ***0.001217 ***0.001180 ***0.001179 ***0.001178 ***0.001180 ***0.001180 ***
[2.954][3.067][3.137][3.146][2.958][2.950][2.946][2.955][2.955]
L.ds1−0.000001
[−0.472]
L.ds2 0.000008
[1.319]
L.ds3 0.000017
[1.152]
L.ds4 0.000013
[1.527]
L.ds5 0.000004
[0.389]
L.ds6 0.000009
[1.530]
L.ds7 −0.000006
[−1.598]
L.ds8 −0.000000
[−0.088]
L.ds9 0.000004 ***
[6.110]
_cons0.000011 ***0.000011 ***0.000011 ***0.000011 ***0.000011 ***0.000011 ***0.000011 ***0.000011 ***0.000011 ***
[14.062][14.387][13.882][14.969][14.166][14.538][14.037][14.025][14.060]
N°observations653653653653653653653653653
Note: t statistics in brackets. * p < 0.1, ** p < 0.05, *** p < 0.01. Source: The Author.
Table 10. Effect of lagged COVID-19 announcements and government intervention on TUNINDEX volatility, excluding cases_rate.
Table 10. Effect of lagged COVID-19 announcements and government intervention on TUNINDEX volatility, excluding cases_rate.
Eq Name:EQ01EQ02EQ03EQ04
L.death_rate0.000198 ***0.000204 ***0.000204 ***0.000220 ***
[4.836][4.907][4.680][4.147]
L.dtintn−0.000072−0.000072−0.000071−0.000071
[−1.021][−1.020][−1.020][−1.018]
L.condvar_oil0.001646 ***0.001636 ***0.001635 ***0.001618 ***
[2.786][2.767][2.762][2.728]
L.dstringencyindex0.000000
[0.638]
L.dcontainmenthealthindex 0.000000
[0.465]
L.dgovernmentresponseindex 0.000000
[0.411]
L.deconomicsupportindex −0.000000
[−0.495]
_cons0.000012 ***0.000012 ***0.000012 ***0.000012 ***
[16.227][16.593][16.627][18.033]
N°observations653653653653
Note: t statistics in brackets. * p < 0.1, ** p < 0.05, *** p < 0.01. Source: The Author.
Table 11. Effect of lagged stringency index sub-indicators on TUNINDEX stock return volatility, excluding cases_rate.
Table 11. Effect of lagged stringency index sub-indicators on TUNINDEX stock return volatility, excluding cases_rate.
Eq NameEQ01EQ02EQ03EQ04EQ05EQ06EQ07EQ08EQ09
L.death_rate0.000213 ***0.000202 ***0.000205 ***0.000169 ***0.000218 ***0.000203 ***0.000232 ***0.000213 ***0.000213 ***
[5.109][4.849][5.831][3.352][4.843][4.478][5.239][5.101][5.101]
L.dtintn−0.000071−0.000072−0.000071−0.000073−0.000071−0.000072−0.000070−0.000071−0.000071
[−1.017][−1.020][−1.020][−1.025][−1.017][−1.020][−1.014][−1.018][−1.018]
L.condvar_oil0.001622 ***0.001649 ***0.001649 ***0.001682 ***0.001618 ***0.001634 ***0.001599 ***0.001624 ***0.001624 ***
[2.742][2.794][2.810][2.847][2.732][2.753][2.703][2.744][2.744]
L.ds1−0.000002
[−0.583]
L.ds2 0.000006
[1.054]
L.ds3 0.000020
[1.202]
L.ds4 0.000011
[1.208]
L.ds5 −0.000007
[−1.063]
L.ds6 0.000006
[1.010]
L.ds7 −0.000010 ***
[−2.704]
L.ds8 0.000001
[0.258]
L.ds9 0.000005 ***
[9.502]
_cons0.000012 ***0.000012 ***0.000012 ***0.000012 ***0.000012 ***0.000012 ***0.000012 ***0.000012 ***0.000012 ***
[17.371][17.171][17.217][16.422][17.548][17.448][17.726][17.368][17.378]
N°observations653653653653653653653653653
Note: t statistics in brackets. * p < 0.1, ** p < 0.05, *** p < 0.01. Source: The Author.
Table 12. Effect of COVID-19 announcements and government intervention on TUNINDEX realized volatility.
Table 12. Effect of COVID-19 announcements and government intervention on TUNINDEX realized volatility.
Eq. NameEQ01EQ02EQ03EQ04
cases_rate0.0686860.0675680.0539390.029122
[0.926][0.924][0.758][0.422]
death_rate−0.012786−0.0113350.0143730.037779
[−0.168][−0.154][0.213][0.632]
condvar_oil5.350280 ***5.354720 ***5.355325 ***5.413046 ***
[11.381][11.378][11.382][11.569]
dtintn−0.053879 ***−0.053868 ***−0.053194 ***−0.053067 ***
[−9.722][−9.753][−10.209][−10.451]
dstringencyindex−0.000839 *
[−1.769]
dcontainmenthealthindex −0.001099 *
[−1.864]
dgovernmentresponseindex −0.001528 **
[−2.306]
deconomicsupportindex −0.001159 ***
[−2.942]
_cons0.053226 ***0.053245 ***0.053160 ***0.053198 ***
[40.496][40.365][40.524][40.101]
N°observations654654654654
Note: t statistics in brackets. * p < 0.1, ** p < 0.05, *** p < 0.01. Source: The Author.
Table 13. Effect of stringency index sub-indicators on TUNINDEX realized volatility.
Table 13. Effect of stringency index sub-indicators on TUNINDEX realized volatility.
Eq. NameEQ01EQ02EQ03EQ04EQ05EQ06EQ07EQ08
cases_rate0.0872190.0836370.0649250.0680310.0792530.0762950.0871340.086742
[1.128][1.094][0.891][0.939][1.044][0.983][1.126][1.121]
death_rate−0.058369−0.0490180.004556−0.026478−0.043969−0.038608−0.057719−0.057569
[−0.671][−0.595][0.066][−0.358][−0.552][−0.450][−0.661][−0.659]
condvar_oil5.379272 ***5.366026 ***5.342108 ***5.382491 ***5.379450 ***5.375517 ***5.378758 ***5.378363 ***
[11.439][11.389][11.388][11.506][11.487][11.464][11.435][11.435]
dtintn−0.055314 ***−0.055010 ***−0.053278 ***−0.054589 ***−0.054937 ***−0.054799 ***−0.055269 ***−0.055289 ***
[−8.754][−9.013][−10.098][−9.393][−9.095][−9.103][−8.741][−8.751]
ds10.001770
[0.315]
ds2 −0.003696
[−0.508]
ds4 −0.013077 ***
[−2.689]
ds5 −0.028130
[−1.534]
ds6 −0.006273
[−0.690]
ds7 −0.007234
[−0.638]
ds8 −0.003069
[−0.521]
ds9 −0.001666
[−1.204]
_cons0.053456 ***0.053411 ***0.053218 ***0.053419 ***0.053391 ***0.053376 ***0.053420 ***0.053444 ***
[39.344][39.590][39.950][39.779][39.606][40.035][39.181][39.188]
N°observations654654654654654654654654
Note: t statistics in brackets. * p < 0.1, ** p < 0.05, *** p < 0.01. Source: The Author.
Table 14. Effect of COVID-19 announcements and government intervention on realized TUNINDEX volatility, excluding cases_rate.
Table 14. Effect of COVID-19 announcements and government intervention on realized TUNINDEX volatility, excluding cases_rate.
Eq. NameEQ01EQ02EQ03EQ04
death_rate0.0191560.0202950.0408830.052863
[0.329][0.359][0.835][1.300]
condvar_oil5.455116 ***5.457813 ***5.436255 ***5.457014 ***
[12.156][12.133][12.064][12.163]
dtintn−0.054896 ***−0.054860 ***−0.053928 ***−0.053430 ***
[−9.040][−9.122][−9.814][−10.357]
dstringencyindex−0.000894 *
[−1.855]
dcontainmenthealthindex −0.001174 *
[−1.936]
dgovernmentresponseindex −0.001616 **
[−2.387]
deconomicsupportindex −0.001201 ***
[−3.080]
_cons0.053428 ***0.053444 ***0.053311 ***0.053275 ***
[41.703][41.674][41.988][41.598]
N°observations654654654654
Note: t statistics in brackets. * p < 0.1, ** p < 0.05, *** p < 0.01. Source: The Author.
Table 15. Effect of stringency index sub-indicators on TUNINDEX realized volatility, excluding cases_rate.
Table 15. Effect of stringency index sub-indicators on TUNINDEX realized volatility, excluding cases_rate.
Eq. NameEQ01EQ02EQ03EQ04EQ05EQ06EQ07EQ08EQ09
death_rate−0.020821−0.011808−0.0195090.0346420.004763−0.008266−0.004016−0.020268−0.020288
[−0.282][−0.176][−0.259][0.650][0.082][−0.130][−0.059][−0.273][−0.274]
condvar_oil5.516901 ***5.495188 ***5.512694 ***5.441774 ***5.487786 ***5.502996 ***5.493677 ***5.516354 ***5.515391 ***
[12.320][12.315][12.282][12.177][12.281][12.364][12.245][12.314][12.311]
dtintn−0.056744 ***−0.056326 ***−0.056678 ***−0.054246 ***−0.055619 ***−0.056163 ***−0.055970 ***−0.056702 ***−0.056714 ***
[−7.852][−8.183][−7.843][−9.531][−8.769][−8.326][−8.314][−7.840][−7.852]
ds10.001593
[0.289]
ds2 −0.004341
[−0.578]
ds3 −0.002024
[−0.264]
ds4 −0.013618 **
[−2.546]
ds5 −0.030568
[−1.572]
ds6 −0.007261
[−0.774]
ds7 −0.008131
[−0.730]
ds8 −0.002860
[−0.486]
ds9 −0.001291
[−0.952]
_cons0.053735 ***0.053671 ***0.053719 ***0.053414 ***0.053629 ***0.053633 ***0.053608 ***0.053701 ***0.053723 ***
[40.029][40.443][39.983][41.026][40.949][40.565][41.092][39.875][39.869]
N°observations654654654654654654654654654
Note: t statistics in brackets. * p < 0.1, ** p < 0.05, *** p < 0.01. Source: The Author.
Table 16. Effect of lagged COVID-19 announcements and government intervention on realized volatility.
Table 16. Effect of lagged COVID-19 announcements and government intervention on realized volatility.
Eq NameEQ01EQ02EQ03EQ04
L.cases_rate−0.034246−0.035573−0.049672−0.042196
[−0.502][−0.529][−0.722][−0.656]
L.death_rate0.129279 **0.130501 **0.154870 ***0.118714 **
[2.189][2.253][2.663][2.381]
L.condvar_oil5.316555 ***5.324149 ***5.330161 ***5.383070 ***
[11.095][11.087][11.078][11.263]
L.dtintn−0.018730 **−0.018746 **−0.018154 **−0.019561 **
[−2.062][−2.072][−2.108][−2.070]
L.dstringencyindex−0.001331 **
[−2.113]
L.dcontainmenthealthindex −0.001719 **
[−2.157]
L.dgovernmentresponseindex −0.002053 **
[−2.477]
L.deconomicsupportindex −0.000735 **
[−2.454]
_cons0.052887 ***0.052922 ***0.052852 ***0.053080 ***
[40.851][40.727][40.647][39.950]
N°observations653653653653
Note: t statistics in brackets. * p < 0.1, ** p < 0.05, *** p < 0.01. Source: The Author.
Table 17. Effect of lagged stringency index sub-indicators on TUNINDEX realized volatility.
Table 17. Effect of lagged stringency index sub-indicators on TUNINDEX realized volatility.
Eq NameEq01Eq02Eq03Eq04Eq05Eq06Eq07Eq08Eq09
L.cases_rate−0.006974−0.0156870.002751−0.028292−0.033772−0.012208−0.024125−0.005724−0.005711
[−0.106][−0.242][0.041][−0.402][−0.513][−0.188][−0.362][−0.088][−0.088]
L.death_rate0.0602580.0861740.0607870.122789 *0.105019 *0.0701660.0918520.0582290.058208
[0.917][1.443][0.908][1.954][1.801][1.156][1.454][0.903][0.903]
L.condvar_oil5.358549 ***5.320837 ***5.327282 ***5.323396 ***5.367313 ***5.362021 ***5.356067 ***5.360940 ***5.360934 ***
[11.221][10.980][11.122][11.105][11.253][11.244][11.240][11.214][11.213]
L.dtintn−0.020915 **−0.020051 **−0.020541 **−0.018879 **−0.019915 **−0.020661 **−0.020098 **−0.020977 **−0.020977 **
[−1.971][−2.012][−1.977][−2.082][−2.061][−1.992][−2.014][−1.975][−1.975]
L.ds1−0.004532
[−0.918]
L.ds2 −0.012094
[−1.170]
L.ds3 −0.015882
[−0.871]
L.ds4 −0.013596
[−1.397]
L.ds5 −0.042358
[−1.404]
L.ds6 −0.005518
[−0.612]
L.ds7 −0.012838
[−1.331]
L.ds8 0.000494
[0.085]
L.ds9 0.002428 *
[1.781]
_cons0.053215 ***0.053121 ***0.053162 ***0.053000 ***0.053196 ***0.053189 ***0.053113 ***0.053243 ***0.053243 ***
[39.881][40.360][40.182][40.296][40.156][40.121][40.187][39.679][39.727]
N°observations653653653653653653653653653
Note: t statistics in brackets. * p < 0.1, ** p < 0.05, *** p < 0.01. Source: The Author.
Table 18. Effect of lagged COVID-19 announcements and government intervention on TUNINDEX volatility, excluding cases_rate.
Table 18. Effect of lagged COVID-19 announcements and government intervention on TUNINDEX volatility, excluding cases_rate.
Eq NameEQ01EQ02EQ03EQ04
L.death_rate0.113355 ***0.113849 ***0.130459 ***0.096859 ***
[2.728][2.736][3.015][2.907]
L.condvar_oil5.264298 ***5.269886 ***5.255650 ***5.319379 ***
[11.110][11.085][11.038][11.269]
L.dtintn−0.018223 **−0.018224 **−0.017477 **−0.019035 **
[−2.075][−2.087][−2.124][−2.082]
L.dstringencyindex−0.001304 **
[−2.114]
L.dcontainmenthealthindex −0.001680 **
[−2.138]
L.dgovernmentresponseindex −0.001972 **
[−2.429]
L.deconomicsupportindex −0.000674 **
[−2.561]
_cons0.052786 ***0.052817 ***0.052714 ***0.052967 ***
[42.808][42.762][42.614][41.712]
N°observations653653653653
Note: t statistics in brackets. * p < 0.1, ** p < 0.05, *** p < 0.01. Source: The Author.
Table 19. Effect of lagged stringency index sub-indicators on TUNINDEX realized volatility, excluding cases_rate.
Table 19. Effect of lagged stringency index sub-indicators on TUNINDEX realized volatility, excluding cases_rate.
Eq NameEQ01EQ02EQ03EQ04EQ05EQ06EQ07EQ08EQ09
L.death_rate0.0572560.079196 *0.0619590.109680 **0.089511 **0.0646660.080914 *0.0557690.055753
[1.201][1.852][1.233][2.486][2.261][1.522][1.865][1.187][1.187]
L.condvar_oil5.347547 ***5.296617 ***5.331660 ***5.279976 ***5.315056 ***5.342994 ***5.318713 ***5.351903 ***5.351914 ***
[11.370][11.077][11.322][11.178][11.279][11.358][11.277][11.373][11.373]
L.dtintn−0.020801 **−0.019804 **−0.020586 **−0.018457 **−0.019404 **−0.020472 **−0.019728 **−0.020882 **−0.020883 **
[−1.970][−2.018][−1.973][−2.091][−2.071][−1.995][−2.019][−1.974][−1.974]
L.ds1−0.004518
[−0.919]
L.ds2 −0.011973
[−1.167]
L.ds3 −0.015857
[−0.869]
L.ds4 −0.013361
[−1.416]
L.ds5 −0.041148
[−1.403]
L.ds6 −0.005365
[−0.605]
L.ds7 −0.012554
[−1.328]
L.ds8 0.000480
[0.083]
L.ds9 0.002403 *
[1.869]
_cons0.053193 ***0.053072 ***0.053171 ***0.052915 ***0.053091 ***0.053152 ***0.053039 ***0.053224 ***0.053225 ***
[41.537][42.288][41.711][41.965][42.153][41.960][42.029][41.356][41.397]
N°observations653653653653653653653653653
Note: t statistics in brackets. * p < 0.1, ** p < 0.05, *** p < 0.01. Source: The Author.
Table 20. Long-run and short-run results of uncertainty impact on TUNINDEX conditional volatility.
Table 20. Long-run and short-run results of uncertainty impact on TUNINDEX conditional volatility.
Using EPU and VIXLevels Equation
Case 2: Restricted Constant and No Trend
VariableCoefficientStd. Errort-StatisticProb.
EPU−1.10 × 10−82.72 × 10−8−0.4060810.6848
VIX2.44 × 10−69.45 × 10−72.5768580.0102
C−4.02 × 10−51.99 × 10−5−2.0223070.0435
EC = CONDVAR_TUN − (−0.0000 × EPU + 0.0000 × VIX − 0.0000)
F-Bounds TestNull Hypothesis: No levels of relationship
Test StatisticValueSignif.I(0)I(1)
Asymptotic: n = 1000
F-statistic20.2250110%2.633.35
k25%3.13.87
2.5%3.554.38
1%4.135
ECM Regression
Case 2: Restricted Constant and No Trend
VariableCoefficientStd. Errort-StatisticProb.
D(CONDVAR_TUN(−1))0.3052000.0360728.4609050.0000
D(CONDVAR_TUN(−2))0.1635970.0378894.3178480.0000
D(CONDVAR_TUN(−3))−0.1354070.038054−3.5582620.0004
D(CONDVAR_TUN(−4))0.0429660.0365671.1749830.2404
D(CONDVAR_TUN(−5))0.1567550.0365354.2905960.0000
D(CONDVAR_TUN(−6))−0.0642290.036672−1.7514650.0803
D(VIX)4.99 × 10−71.75 × 10−72.8460850.0046
D(VIX(−1))5.59 × 10−71.83 × 10−73.0607800.0023
CointEq(−1) *−0.2140800.023749−9.0141190.0000
Using EMU + VIXLevels Equation
Case 2: Restricted Constant and No Trend
VariableCoefficientStd. Errort-StatisticProb.
EMU1.02 × 10−75.16 × 10−81.9828380.0478
VIX1.33 × 10−65.94 × 10−72.2339820.0258
C−2.63 × 10−51.49 × 10−5−1.7693590.0773
EC = CONDVAR_TUN − (0.0000 × EMU + 0.0000 × VIX − 0.0000)
F-Bounds TestNull Hypothesis: No levels relationship
Test StatisticValueSignif.I(0)I(1)
Asymptotic: n = 1000
F-statistic16.6801610%2.633.35
k25%3.13.87
2.5%3.554.38
1%4.135
ECM Regression
Case 2: Restricted Constant and No Trend
VariableCoefficientStd. Errort-StatisticProb.
D(CONDVAR_TUN(−1))0.2803600.0377207.4326600.0000
D(CONDVAR_TUN(−2))0.1695880.0394024.3039950.0000
D(CONDVAR_TUN(−3))−0.1475480.039403−3.7446380.0002
D(CONDVAR_TUN(−4))0.0337690.0374660.9013240.3677
D(CONDVAR_TUN(−5))0.1537000.0373124.1193770.0000
D(CONDVAR_TUN(−6))−0.0806140.037319−2.1601500.0311
D(EMU)−5.65 × 10−96.22 × 10−9−0.9083230.3640
D(EMU(−1))−2.82 × 10−87.75 × 10−9−3.6384820.0003
D(EMU(−2))−2.49 × 10−88.39 × 10−9−2.9680320.0031
D(EMU(−3))−3.79 × 10−88.57 × 10−9−4.4183110.0000
D(EMU(−4))−2.99 × 10−88.34 × 10−9−3.5876800.0004
D(EMU(−5))−3.58 × 10−97.60 × 10−9−0.4714560.6375
D(EMU(−6))1.44 × 10−86.23 × 10−92.3152350.0209
D(VIX)4.39 × 10−71.75 × 10−72.5131630.0122
D(VIX(−1))7.64 × 10−71.87 × 10−74.0913210.0000
D(VIX(−2))3.64 × 10−71.88 × 10−71.9331890.0536
D(VIX(−3))4.45 × 10−71.89 × 10−72.3559370.0188
D(VIX(−4))2.90 × 10−71.90 × 10−71.5235890.1281
D(VIX(−5))2.81 × 10−71.89 × 10−71.4899220.1367
D(VIX(−6))4.49 × 10−71.80 × 10−72.4883580.0131
CointEq(−1) *−0.2083230.025447−8.1864560.0000
Using IDEMV + VIXLevels Equation
Case 2: Restricted Constant and No Trend
VariableCoefficientStd. Errort-StatisticProb.
IMEDV7.74 × 10−73.47 × 10−72.2305790.0260
VIX1.56 × 10−66.78 × 10−72.2937250.0221
C−3.15 × 10−51.69 × 10−5−1.8601520.0633
EC = CONDVAR_TUN − (0.0000 × IMEDV + 0.0000 × VIX − 0.0000)
F-Bounds TestNull Hypothesis: No levels relationship
Test StatisticValueSignif.I(0)I(1)
Asymptotic: n = 1000
F-statistic16.1659610%2.633.35
k25%3.13.87
2.5%3.554.38
1%4.135
ECM Regression
Case 2: Restricted Constant and No Trend
VariableCoefficientStd. Errort-StatisticProb.
D(CONDVAR_TUN(−1))0.2946400.0385417.6448520.0000
D(CONDVAR_TUN(−2))0.1567280.0401283.9057260.0001
D(CONDVAR_TUN(−3))−0.1481190.039993−3.7036290.0002
D(CONDVAR_TUN(−4))0.0320520.0379880.8437370.3991
D(CONDVAR_TUN(−5))0.1494790.0379473.9391360.0001
D(CONDVAR_TUN(−6))−0.0663080.037763−1.7559020.0796
D(IMEDV)4.17 × 10−86.49 × 10−80.6429590.5205
D(IDEMV(−1))−1.54 × 10−78.11 × 10−8−1.8933840.0587
D(IDEMV(−2))−1.26 × 10−78.97 × 10−8−1.4094330.1592
D(IDEMV(−3))−1.98 × 10−79.24 × 10−8−2.1393180.0328
D(IDEMV(−4))−1.67 × 10−78.99 × 10−8−1.8543060.0641
D(IDEMV(−5))−5.06 × 10−88.15 × 10−8−0.6205600.5351
D(IDEMV(−6))1.28 × 10−76.64 × 10−81.9259140.0545
D(VIX)4.08 × 10−71.81 × 10−72.2594210.0242
D(VIX(−1))7.04 × 10−71.95 × 10−73.6076290.0003
D(VIX(−2))3.73 × 10−71.98 × 10−71.8896940.0592
D(VIX(−3))3.89 × 10−71.98 × 10−71.9660480.0497
D(VIX(−4))1.44 × 10−71.99 × 10−70.7248400.4688
D(VIX(−5))2.50 × 10−71.94 × 10−71.2900370.1975
D(VIX(−6))4.65 × 10−71.84 × 10−72.5220340.0119
CointEq(−1) *−0.2127750.026401−8.0595000.0000
Note: * Cointegrating equation term.
Table 21. Long-run and short-run results of uncertainty impact on TUNINDEX realized volatility.
Table 21. Long-run and short-run results of uncertainty impact on TUNINDEX realized volatility.
Using EPU and VIXLevels Equation
Case 2: Restricted Constant and No Trend
VariableCoefficientStd. Errort-StatisticProb.
EPU0.0001526.74 × 10−52.2554910.0244
VIX0.0026210.0011882.2058600.0277
C−0.0316370.025702−1.2309150.2188
EC = RV − (0.0002 × EPU + 0.0026 × VIX − 0.0316)
F-Bounds TestNull Hypothesis: No levels relationship
Test StatisticValueSignif.I(0)I(1)
Asymptotic: n = 1000
F-statistic 9.06231210% 2.633.35
k25% 3.13.87
2.5% 3.554.38
1% 4.135
ECM Regression
Case 2: Restricted Constant and No Trend
VariableCoefficientStd. Errort-StatisticProb.
D(RV(−1))−0.8322470.038215−21.778030.0000
D(RV(−2))−0.8507050.047015−18.094180.0000
D(RV(−3))−0.6219160.051961−11.969000.0000
D(RV(−4))−0.4711660.046596−10.111760.0000
D(RV(−5))0.0982570.0365862.6856280.0074
D(EPU)−3.38 × 10−58.59 × 10−6−0.3933250.6942
D(EPU(−1))−3.14 × 10−59.86 × 10−6−3.1796220.0015
D(EPU(−2))−1.77 × 10−58.65 × 10−6−2.0414720.0416
D(VIX)−6.80 × 10−50.000204−0.3330570.7392
D(VIX(−1))0.0006290.0002122.9664450.0031
D(VIX(−2))7.76 × 10−50.0002130.3636780.7162
D(VIX(−3))8.51 × 10−50.0002140.3971240.6914
D(VIX(−4))−2.47 × 10−50.000214−0.1155180.9081
D(VIX(−5))0.0009360.0002074.5164070.0000
CointEq(−1) *−0.1313080.021761−6.0339820.0000
Using EMU + VIXLevels Equation
Case 4: Unrestricted Constant and Restricted Trend
VariableCoefficientStd. Errort-StatisticProb.
EMU0.0002500.0001481.6836780.0927
VIX0.0018620.0010291.8091860.0709
@TREND−3.65 × 10−51.91 × 10−5−1.9103760.0565
EC = RV − (0.0002 × EMU + 0.0019 × VIX − 0.0000 × @TREND)
F-Bounds TestNull Hypothesis: No levels relationship
Test StatisticValueSignif.I(0)I(1)
Asymptotic: n = 1000
F-statistic 9.75488610% 3.384.02
k25% 3.884.61
2.5% 4.375.16
1% 4.995.85
ECM Regression
Case 4: Unrestricted Constant and Restricted Trend
VariableCoefficientStd. Errort-StatisticProb.
C−2.36 × 10−50.000514−0.0459750.9633
D(RV(−1))−0.8411760.037294−22.555280.0000
D(RV(−2))−0.8658850.046192−18.745170.0000
D(RV(−3))−0.6555730.051174−12.810590.0000
D(RV(−4))−0.4933930.046515−10.607250.0000
D(RV(−5))0.0823680.0364462.2600350.0241
D(EMU)−1.53 × 10−57.15 × 10−6−2.1342000.0332
D(EMU(−1))−3.15 × 10−59.11 × 10−6−3.4566140.0006
D(EMU(−2))−2.91 × 10−59.58 × 10−6−3.0330140.0025
D(EMU(−3))−3.51 × 10−59.37 × 10−6−3.7390970.0002
D(EMU(−4))−2.67 × 10−58.57 × 10−6−3.1150080.0019
D(EMU(−5))−1.93 × 10−57.14 × 10−6−2.7095240.0069
D(VIX)−0.0001370.000202−0.6761100.4992
D(VIX(−1))0.0006970.0002153.2438370.0012
D(VIX(−2))1.96 × 10−50.0002140.0915990.9270
D(VIX(−3))0.0001080.0002170.4981180.6186
D(VIX(−4))2.34 × 10−50.0002170.1078710.9141
D(VIX(−5))0.0009430.0002104.4940510.0000
CointEq(−1) *−0.1215610.019417−6.2603880.0000
Using IMEDV + VIXLevels Equation
Case 4: Unrestricted Constant and Restricted Trend
VariableCoefficientStd. Errort-StatisticProb.
IMEDV−0.0009200.000998−0.9214650.3571
VIX0.0046720.0017392.6871710.0074
@TREND−6.39 × 10−52.62 × 10−5−2.4440310.0148
EC = RV − (−0.0009 × IMEDV + 0.0047 × VIX − 0.0001 × @TREND)
F-Bounds TestNull Hypothesis: No levels relationship
Test StatisticValueSignif.I(0)I(1)
Asymptotic: n = 1000
F-statistic 9.03548510% 3.384.02
k25% 3.884.61
2.5% 4.375.16
1% 4.995.85
ECM Regression
Case 4: Unrestricted Constant and Restricted Trend
VariableCoefficientStd. Errort-StatisticProb.
C−0.0020700.000614−3.3710470.0008
D(RV(−1))−0.8229800.038000−21.657320.0000
D(RV(−2))−0.8411940.046688−18.017480.0000
D(RV(−3))−0.6113660.051893−11.781180.0000
D(RV(−4))−0.4596130.046307−9.9254020.0000
D(RV(−5))0.0971880.0364232.6682810.0078
D(IMEDV)−0.0003196.09 × 10−5−5.2341770.0000
D(VIX)−6.01 × 10−50.000202−0.2980590.7657
D(VIX(−1))0.0005270.0002172.4297440.0154
D(VIX(−2))−0.0001550.000216−0.7182910.4728
D(VIX(−3))−9.15 × 10−50.000214−0.4268810.6696
D(VIX(−4))−7.96 × 10−50.000215−0.3707240.7110
D(VIX(−5))0.0007390.0002143.4469900.0006
D(VIX(−6))−0.0003450.000210−1.6427740.1009
CointEq(−1) *−0.1109250.018411−6.0251020.0000
Note: * Cointegrating equation term.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Trabelsi, E. COVID-19 and Uncertainty Effects on Tunisian Stock Market Volatility: Insights from GJR-GARCH, Wavelet Coherence, and ARDL. J. Risk Financial Manag. 2024, 17, 403. https://doi.org/10.3390/jrfm17090403

AMA Style

Trabelsi E. COVID-19 and Uncertainty Effects on Tunisian Stock Market Volatility: Insights from GJR-GARCH, Wavelet Coherence, and ARDL. Journal of Risk and Financial Management. 2024; 17(9):403. https://doi.org/10.3390/jrfm17090403

Chicago/Turabian Style

Trabelsi, Emna. 2024. "COVID-19 and Uncertainty Effects on Tunisian Stock Market Volatility: Insights from GJR-GARCH, Wavelet Coherence, and ARDL" Journal of Risk and Financial Management 17, no. 9: 403. https://doi.org/10.3390/jrfm17090403

Article Metrics

Back to TopTop