Dynamic co-movements of COVID-19 pandemic anxieties and stock market returns

Abstract

In this study, we constructed two pandemic anxiety indexes based on an assumption that people's emotions fluctuate with the COVID-19 reported cases and deaths, to examine the dynamic co-movements between these anxiety indexes and the stock markets in the BRICS and G7 countries. We found that the anxiety indexes are volatile over time but have an overall downtown trend. The correlations between stock market returns and the epidemic anxiety indexes are time varying. We found a common feature across the countries studied, namely that the correlation becomes weaker and has smaller fluctuations after the announcement of the mRNA-based COVID-19 vaccine.

1. Introduction

Since the COVID-19 pandemic emerged in January 2020, it spread rapidly and became a human tragedy as well as both a healthcare and economic crisis around the world. To date, the total reported cases of COVID-19 exceed one hundred million, meaning that nearly 1in 70 people in the world have been infected. As the original cause has not been identified, and it is not yet known how much longer the pandemic will last, general psychological disturbances, mental anxiety, and economic shocks remain elevated. There is no doubt that the COVID-19 pandemic has been an urgent topic of research for academicians and experts in different fields. For example, there is much research being done by governments and public health experts who are responsible for making decisions to control and suppress the COVID-19 crisis and prevent future epidemics (Atalan, 2020, Kanitkar, 2020, Noorbhai, 2020). Scholars in the field of medicine and epidemiology are studying the origin, health outcomes, treatments, and vaccine development for the COVID-19 disease (Hetkamp et al., 2020, Salameh et al., 2020, Wuyts et al., 2020) Researchers in economics are examining whether and how the pandemic has affected the economy in various ways (Allam, 2020, Ashraf, 2020a, Bernauer and Slowey, 2020, Hossain, 2021, Keogh-Brown et al., 2020).

In this study, we examine the dynamic correlation between the COVID-19 pandemic and stock market returns in 12 countries, namely Brazil, Russia, India, China, and South Africa (often referred to as the BRICS countries) and the G7 countries, i.e., the United States, England, Canada, Italy, Germany, France, and Japan. There are already a number of studies on the relationship between the COVID-19 pandemic and financial markets. Zhang et al. (2020) analyzed the systemic risks and potential consequences of policy interventions in the global financial markets during the pandemic. Ali et al. (2020) investigated the reactions of global financial markets as the pandemic spread from Asia to Europe and then to North America. Some studies examined the stock markets’ reaction in terms of volatility during COVID-19 (Baek et al., 2020, Corbet et al., 2021, Just and Echaust, 2020, Takyi and Bentum-Ennin, 2020, Topcu and Gulal, 2020, Yong and Laing, 2020, Ashraf, 2020b). Other studies focused on the pandemic's impact on connections across international markets (Guo et al., 2021, So et al., 2021).

Our study expands the literature on the relationship between the COVID-19 pandemic and financial markets. Our study's contributions can be summarized as follows: First, in contrast to studies that used the number of reported COVID-19 cases as a key variable (Just and Echaust, 2020, Xu, 2021, Yong and Laing, 2020), we constructed two kinds of pandemic anxiety  index based on the numbers of daily reported cases and deaths to examine the linkage between the COVID-19 pandemic and stock markets. Secondly, our approach differs from the methods used in previous studies, such as the event study, VAR models, the OLS regression approach, and Granger complex network (Baek et al., 2020, Lai and Hu, 2021, Sun et al., 2021, Topcu and Gulal, 2020, Xu, 2021). We first employed a Dynamic Conditional Correlation- Generalized Autoregressive Conditional Heteroskedasticity (DCC-GARCH) model to capture the dynamic co-movement between the pandemic's evolution and stock market volatility. Then, we selected BRICS and G7 countries as research objects, as they are representative of emerging and developed economies, respectively.

The remainder of the study is organized as follows: Section 2 introduces our methodologies, Section 3 describes and provides some basic statistics about the data used in the study, Section 4 details the empirical results, and Section 5 concludes the study.

2. Methodologies

2.1. Pandemic anxiety index

Following Salisu et al. (2020),we construct two types of pandemic anxiety indexes as proxies for measuring sentiments about the spread and severity of COVID-19, based on an assumption that people's anxiety emotions fluctuate with the changes of the COVID-19 reported cases and deaths. The first index, called COVID-19 Anxiety Index (AI), is similar to the approach in Salisu and Akanni (2020) and was constructed using daily reported cases and deaths (Salisu and Akanni, 2020). The second index, called COVID-19 Rolling Anxiety Index (RAI) was constructed based on the total number of reported cases and deaths in the 14-day incubation period. The two pandemic anxiety indexes can be specified as below:

$$A{I}_{i,t}=0.5\times \left[\frac{C_{i,t}}{C_{i,t}+C_{i,t-14}}+\frac{D_{i,t}}{D_{i,t}+D_{i,t-14}}\right]\times 100$$ (1)

$$RA{I}_{i,t}=0.5\times \left[\frac{\sum _{k=1}^{14}{C}_{i,t-14+k}}{\sum _{k=0}^{13}{C}_{i,t-14+k}}+\frac{\sum _{k=1}^{14}{D}_{i,t-14+k}}{\sum _{k=0}^{13}{D}_{i,t-14+k}}\right]$$ (2)

where C _i,t denotes the number of COVID-19 reported cases at day t for country i; D _i,t denotes the number of COVID-19 reported death at day t for country i; C _i,t − 14 and D _i,t − 14 are reported cases and deaths, respectively, in country i at the beginning of the incubation period. $\sum _{k=1}^{14}{C}_{i,t-14+k}$ and $\sum _{k=1}^{14}{D}_{i,t-14+k}$ denote the total number of reported cases and deaths, respectively, in the 14-day incubation period including day t; while $\sum _{k=0}^{13}{C}_{i,t-14+k}$ and $\sum _{k=0}^{13}{D}_{i,t-14+k}$ respectively denote the total number in the previous 14-day incubation period.

2.2. Granger causality test

After constructing the pandemic anxiety indexes, we next take Granger causality test to examine whether changes in the level of anxiety associated with the COVID-19 pandemic causes a change in stock market returns. The Granger causality test equations are specified as follows:

$$r_{i,t}={\phi }_i+{\alpha }_{i,1}{r}_{i,t-1}+\dots +{\alpha }_{i,p}{r}_{i,t-p}+{\beta }_{i,1}A{I}_{i,t-1}+\dots +{\beta }_{i,p}A{I}_{i,t-p}+{\epsilon }_i$$ (3)

$$r_{i,t}={\phi }_i+{\alpha }_{i,1}{r}_{i,t-1}+\dots +{\alpha }_{i,p}{r}_{i,t-p}+{\beta }_{i,1}RA{I}_{i,t-1}+\dots +{\beta }_{i,p}RA{I}_{i,t-p}+{\epsilon }_i$$ (4)

where r_{i , t} is the one-day log stock return at time t for country i; and AI_{i ,t} and  RAI_{i , t} are the two COVID-19 pandemic anxiety indexes constructed above. p is the lag order which is selected by BIC information criterion. We calculated the daily log return at time t for country i as r_{i, t} = 100*log(r_i,t/r_{i, t − 1}).The null hypothesis of the Granger causality test is ${\beta }_{i,t-1},\dots ,{\beta }_{i{,}_{t-\mathrm{p}}}=0$. Rejecting the null hypothesis means that the anxiety due to the pandemic causes a change in stock market.

2.3. DCC-GARCH model

Next step, we further constructed a bivariate DCC–GARCH (1,1) model to examine the dynamic correlation between stock volatility and COVID-19 pandemic anxiety indexes. Engle (2002) introduced the dynamic conditional correlation-Generalized Autoregressive Conditional Heteroskedasticity (DCC–GARCH) model to examine time-varying correlations between assets. The bivariate DCC–GARCH (1,1) model is specified below:

$$r_t=\mu +{\varnothing }_1{r}_{t-1}+\dots +{\varnothing }_p{r}_{t-p}+{\epsilon }_t$$ (5)

$${\epsilon }_t=\sqrt{h_t}·z_t{z}_t\sim i.i.d.\left(0,1\right){\epsilon }_t\sim N\left(0,H_t\right)$$ (6)

$$h_t=\omega +\alpha ·{{\epsilon }_{t-1}}^2+\beta ·h_{t-1}$$ (7)

$${\mathrm{H}}_t=D_t{R}_t{D}_t$$ (8)

$$Q_t=\left(1-a^{DCC}-b^{DCC}\right)\overline{Q}+a^{DCC}{z}_t{z}_{t-1}^T+b^{DCC}{Q}_{t-1}$$ (9)

$$R_t=diag{\left(Q_t\right)}^{-1}\left(Q_t\right)diag{\left(Q_t\right)}^{-1}$$ (10)

where  r_t = (r_stock,r_efi)′, r_stock is the stock market return and r_efi is the change of pandemic anxiety  index; z_t is the distributed error vector with z_t = (z _t,stock, z _t,efi)′; ε_t is the standardized residual vector of the mean equation with ε_t = (ε_t,stock, ε_t,efi)′; ${\mathrm{H}}_t$ is the dynamic conditional variance-covariance matrix with $H_t={\left(\sqrt{h_{t,stock}},\sqrt{h_{t,efi}}\right)}^{\prime }$;  ω, α  and β are the parameters of GARCH (1,1) model while   a^DCC and b^DCC are non-negative scalar parameters of the DCC procedure, where a^DCC + b^DCC < 1 . R_t is the dynamic correlation coefficient matrix and D_t is the conditional standard deviation matrix; Q_t is a symmetric positive definite matrix.

3. Data and descriptive statistics

We collected daily stock market data for the BRICS and G7 countries from the CEIC database. For each country, we selected one representative stock index, the details of which are shown in Table 1 . We extracted daily COVID-19 data from the Wind database, which captures data from the World Health Organization. The period used to construct the anxiety indexes spans from March 6, 2020 to January 29, 2021, with the starting date selected according to the date when the first case was reported in South Africa. The stock market data covers the period from March 19, 2020 to January 29, 2021

Table 1.

List of selected stock index series of BRICS and G7 countries.

Country	Stock index	Country	Stock index
Brazil	BOVESPA Index	England	FTSE 250 Index
Russian	RTS Index (US Dollar)	Canada	S&P/TSX Composite Index
India	Bombay Stock Exchange 100 Index	Italy	FTSE Italia STAR Index
China	Shanghai Composite Stock Index	Germany	DAX Index
South Africa	FTSE All Share Index	France	CAC 40 Index
United States	Dow Jones Composite Average Index	Japan	TSE TOPIX 1st Section Index

Notes: This table reports details on stock market index chosen for each country. Notably, although the S&P 500 index is used in most studies of the United States stock market for its more comprehension, we find the correlation coefficient between the Dow Jones Composite Average Index and the S&P 500 index is 0.9909 during our sample period (March 19, 2020 to January 29, 2021) while 0.9490 for the same period prior to the sample period (May 8,2019 to March 18, 2020). It indicates that the two stock indexes are highly consistent during the COVID-19 pandemic and it is suitable to select the Dow Jones Composite Average Index for this study here.

Fig. 1 depicts the daily values of the stock market indexes for the BRICS and G7 countries, and trends of daily stock log returns are reported in Fig. 6 in Appendix). Fig. 2 plots the trends of absolute number of daily COVID-19 reported cases and deaths for each country. The mean daily reported cases for Brazil, Russia, India, China, South Africa, the United States, England, Canada, Italy, Germany, France, and Japan are 27,263,11,555, 32,485,61,4357, 76,830,11,344,2307,7613, 6644,9331, and 1149, respectively. The corresponding mean daily reported deaths are 667, 219, 467, 5, 131, 1290, 313, 59, 264, 169, 225, and 17, respectively. As we know, the corresponding total population of these 12 countries are 210.000, 144.478, 1324.000, 1400.050, 56.52, 330.000, 65.580, 37.059, 60.800, 82.930, 66.987 and 124.776 million, respectively. If we standardize the daily numbers of reported cases by using the total population in each country, the corresponding infection destiny are 129.824, 79.978, 24.535, 0.044,77.088,232.818,172.980,62.252, 125.214,80.116,139.296 and 9.209 per million, respectively. This shows that among the countries included in this study, the U.S. has been most heavily affected by the pandemic (the infection destiny is 232.818 per million), while the impact on China has been the least severe (the infection destiny is 0.044 per million).

Fig. 1.

Trends in daily stock index value for the BRICS and G7countries.

Fig. 6.

Daily stock market returns for the BRICS and G7countries.

Fig. 2.

Trends in the daily COVID-19 reported cases and deaths for the BRICS and G7 countries. Note: Reported cases are plotted on the left axis (blue line), and reported deaths are plotted on the right axis (red line).

Fig. 3 plots trends of the COVID-19 pandemic anxiety indexes. We observed a downward trend in both the anxiety indexes as a whole, despite various phases of increases and decreases in reported cases and deaths across all the countries in our sample. In particular, in the second half of 2020, although the daily number of reported COVID- 19 cases and deaths hit an all-time high (except in China), the anxiety indexes were still lower than at the beginning of the COVID-19 pandemic.

Fig. 3.

Trends in the two daily COVID-19 anxiety indexes for the BRICS and G7 countries. Note: The pandemic AI is plotted on the left axis, and the pandemic RAI is plotted on the right axis.

Table 2 summarizes the descriptive statistics for the stock market returns and COVID−19 pandemic anxiety indexes of each country. All series, except market returns for China and Russia, are skewed and have excess kurtosis, indicating a leptokurtic distribution. As shown in Table 3 , the Augmented Dickey–Fuller (ADF) test results for the pandemic anxiety indexes, except for the AI in China, the United States, Germany, and France, show that we cannot reject the null hypothesis of a unit root, implying the time series are nonstationary. However, all the ADF test results for the differences in stock market returns and pandemic anxiety indexes (Δr, ΔAI, and ΔRAI) are significant at the 1% level.

Table 2.

Descriptive statistics of stock return and pandemic anxiety index for BRICS and G7 countries.

	During Sample period								Pre-sample period
	Mean	Median	Min	Max	Std.	Skew.	Kurt.		Skew.	Kurt.
Stock return
Brazil	0.2306	0.1293	−5.6664	9.2485	1.8935	0.4720	6.0987		−2.5057	22.7264
Russian	0.1839	0.2161	−7.9258	8.8251	1.9762	−0.0588	5.9989		−2.4055	17.5197
India	0.2215	0.3204	−13.8810	5.8879	1.6903	−2.3655	24.3050		−1.9678	16.1304
China	0.1123	0.1050	−4.6026	5.5543	1.0922	0.0883	7.2201		−0.4307	5.1083
South Africa	0.2204	0.2212	−5.1063	7.2615	1.4794	0.5719	6.7628		−3.4867	21.3255
United States	0.1769	0.1672	−6.7783	10.8260	1.7333	1.0140	11.7140		−2.4479	22.5630
England	0.2015	0.1209	−4.6780	8.0388	1.6153	0.7004	6.6000		−3.2589	21.1077
Canada	0.1566	0.2079	−5.4026	11.2940	1.4527	1.4838	19.6610		−3.6370	33.0477
Italy	0.2161	0.2176	−3.8867	3.9050	1.0679	−0.1372	4.7785		−3.7974	32.7978
Germany	0.1968	0.0707	−4.5711	10.4140	1.7498	0.7763	8.5238		−3.8798	28.1049
France	0.1490	0.0921	−4.8205	8.0561	1.7008	0.6589	6.5039		−3.8884	27.4339
Japan	0.1519	0.0616	−3.7737	6.6398	1.2075	0.8203	7.1936		−1.6338	9.8920
Daily COVID-19 Anxiety Index (AI)
Brazil	0.5571	0.5216	0.2527	1.0000	0.1349	1.2141	5.0612
Russian	0.5607	0.5259	0.3967	1.0000	0.1266	2.0970	6.8727
India	0.5562	0.5380	0.0000	1.0000	0.1416	0.9748	5.0925
China	0.4742	0.4156	0.0000	0.9643	0.2555	0.2459	1.8762
South Africa	0.5674	0.5736	0.0000	1.0000	0.1603	−0.2035	3.7881
United States	0.5465	0.5213	0.0000	2.0600	0.1680	3.7563	32.5130
England	0.5405	0.5141	0.2055	0.9762	0.1562	0.6731	3.1817
Canada	0.5387	0.5295	0.0000	1.0000	0.1653	0.4055	4.4302
Italy	0.5054	0.4798	−0.3434	0.8933	0.1536	−0.4706	6.5596
Germany	0.5369	0.5328	0.2130	0.9918	0.1570	0.5192	3.2505
France	0.5265	0.5296	0.0000	1.3427	0.1971	0.1902	4.3378
Japan	0.5429	0.5531	0.1507	1.0000	0.1524	0.1030	3.1991
Daily COVID-19 Rolling Anxiety Index (RAI)
Brazil	1.0292	1.0061	0.9358	3.3321	0.1607	13.1400	188.1800
Russian	1.0158	1.0070	0.5661	1.5667	0.0852	0.3685	19.3940
India	1.0225	1.0109	0.9102	1.5477	0.0636	3.9264	26.1480
China	1.0306	0.9759	0.2053	26.8480	1.7399	14.5490	216.4300
South Africa	1.0175	1.0207	0.5000	2.0143	0.1033	2.8759	45.0370
United States	1.0224	1.0049	0.9483	2.0443	0.0963	7.3353	67.9960
England	1.0172	1.0043	0.9183	1.3107	0.0590	2.0571	9.6010
Canada	1.0180	1.0096	0.8853	1.7613	0.0763	5.0125	44.4820
Italy	1.0050	0.9967	0.9249	1.1374	0.0423	0.7462	3.3914
Germany	1.0214	1.0090	0.9065	2.5194	0.1148	10.0130	129.6000
France	1.0130	1.0089	0.7682	1.3583	0.0681	0.8155	7.6105
Japan	1.0162	1.0163	0.8661	1.4263	0.0548	1.8706	16.5610

Notes:This table shows descriptive statistics of stock market returns and pandemic anxiety indexes for the BRICS and G7 countries. In order to observe whether the skewness and kurtosis of the stock returns have changed from the pre-pandemic period, we also report the skewness and kurtosis for the same period prior to our sample in the right part of this table. Our sample period spans from March 20, 2020 to January 29, 2021 while the benchmarked pre-sample period covers May 9,2019 to March 19, 2020, whereby including 226 trading dates for each period. As shown in this table, the stock returns, except for India, are more skewed and have fatter tails during the pandemic than under “normal” times in all cases. The COVID-19 AI and RAI are shown in Eqs. (1) and (2), respectively. We calculated daily anxiety indexes from March 20, 2020 to January 29, 2021, using data from Monday to Friday each week to match stock trading dates.

Table 3.

Results of stationarity and ARCH tests.

	Brazil	Russian	India	China	South Africa	United States	England	Canada	Italy	Germany	France	Japan
ADF test
r	−11.174⁎⁎⁎	−10.824⁎⁎⁎	−13.004⁎⁎⁎	−9.8768⁎⁎⁎	−9.7788⁎⁎⁎	−10.299⁎⁎⁎	−9.5583⁎⁎⁎	−13.215⁎⁎⁎	−9.0987⁎⁎⁎	−9.716⁎⁎⁎	−9.7338⁎⁎⁎	−9.1193⁎⁎⁎
	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
AI	−0.8311	−0.5901	−1.0958	−1.8092	−1.4892	−1.9458⁎⁎	−1.6024	−1.5509	−1.3854	−1.6225*	−1.7251*	−1.2233
	(0.3466)	(0.4349)	(0.2497)	(0.0671)	(0.1277)	(0.0496)	(0.1029)	(0.114)	(0.1542)	(0.0987)	(0.0801)	(0.2031)
RAI	−0.8781	0.0015	−1.422	−7.314⁎⁎⁎	−0.4518	−2.1701⁎⁎	−0.8082	−0.8131	−0.6521	−1.0579	−0.6577	−0.3339
	(0.3294)	(0.6515)	(0.1445)	(0.001)	(0.4855)	(0.0292)	(0.355)	(0.3533)	(0.4122)	(0.2636)	(0.4101)	(0.5287)
Δr	−18.164⁎⁎⁎	−18.642⁎⁎⁎	−18.92⁎⁎⁎	−17.905⁎⁎⁎	−17.999⁎⁎⁎	−16.665⁎⁎⁎	−16.233⁎⁎⁎	−18.728⁎⁎⁎	−17.379⁎⁎⁎	−16.288⁎⁎⁎	−16.618⁎⁎⁎	−16.266⁎⁎⁎
	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
ΔAI	−16.818⁎⁎⁎	−13.381⁎⁎⁎	−22.11⁎⁎⁎	−13.234⁎⁎⁎	−20.111⁎⁎⁎	−17.046⁎⁎⁎	−17.69⁎⁎⁎	−21.313⁎⁎⁎	−17.407⁎⁎⁎	−15.522⁎⁎⁎	−18.066⁎⁎⁎	−14.801⁎⁎⁎
	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
ΔRAI	−32.284⁎⁎⁎	−11.181⁎⁎⁎	−20.663⁎⁎⁎	−18.203⁎⁎⁎	−18.253⁎⁎⁎	−18.066⁎⁎⁎	−17.51⁎⁎⁎	−24.974⁎⁎⁎	−18.24⁎⁎⁎	−19.736⁎⁎⁎	−17.495⁎⁎⁎	−15.648⁎⁎⁎
	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
ARCH test
r	62.9430⁎⁎⁎	2.7316*	6.8791⁎⁎⁎	0.4201	97.5740⁎⁎⁎	3.1322*	32.4630⁎⁎⁎	24.1930⁎⁎⁎	0.2188	0.0092	5.1058⁎⁎	10.9080⁎⁎⁎
	(0.0000)	(0.0984)	(0.0087)	(0.5169)	(0.0000)	(0.0768)	(0.0000)	(0.0000)	(0.6400)	(0.9236)	(0.0238)	(0.0010)
AI	76.1870⁎⁎⁎	164.430⁎⁎⁎	167.480⁎⁎⁎	115.830⁎⁎⁎	29.066⁎⁎⁎	12.294⁎⁎⁎	178.380⁎⁎⁎	33.859⁎⁎⁎	177.980⁎⁎⁎	153.580⁎⁎⁎	30.763⁎⁎⁎	42.162⁎⁎⁎
	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0005)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)
RAI	0.0004	80.839⁎⁎⁎	52.788⁎⁎⁎	0.0045	24.600⁎⁎⁎	79.410⁎⁎⁎	160.820⁎⁎⁎	8.0837⁎⁎⁎	171.680⁎⁎⁎	69.751⁎⁎⁎	34.527⁎⁎⁎	27.450⁎⁎⁎
	(0.9848)	(0.0000)	(0.0000)	(0.9464)	(0.0000)	(0.0000)	(0.0000)	(0.0045)	(0.0000)	(0.0000)	(0.0000)	(0.0000)
Δr	10.7430⁎⁎⁎	31.3160⁎⁎⁎	12.2070⁎⁎⁎	24.9260⁎⁎⁎	10.5880⁎⁎⁎	62.4120⁎⁎⁎	5.2259⁎⁎	52.8850⁎⁎⁎	22.8200⁎⁎⁎	57.5380⁎⁎⁎	16.6700⁎⁎⁎	49.2290⁎⁎⁎
	(0.0010)	(0.0000)	(0.0005)	(0.0000)	(0.0011)	(0.0000)	(0.0223)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)
ΔAI	89.5560⁎⁎⁎	26.0540⁎⁎⁎	0.4179	12.7860⁎⁎⁎	17.9920⁎⁎⁎	56.9230⁎⁎⁎	78.1490⁎⁎⁎	65.9700⁎⁎⁎	43.9610⁎⁎⁎	18.8340⁎⁎⁎	43.4150⁎⁎⁎	43.1280⁎⁎⁎
	(0.0000)	(0.0000)	(0.518)	(0.0003)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)
ΔRAI	56.6610⁎⁎⁎	5.3016⁎⁎	144.430⁎⁎⁎	54.9800⁎⁎⁎	91.6700⁎⁎⁎	92.6440⁎⁎⁎	22.1300⁎⁎⁎	77.0560⁎⁎⁎	3.8357*	0.0004	43.7590⁎⁎⁎	50.8230⁎⁎⁎
	(0.0000)	(0.0213)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0502)	(0.9845)	(0.0000)	(0.0000)

Notes: This table reports the results of stationarity and ARCH tests for stock market returns and pandemic anxiety indexes. r, AI, and RAI represent   stock market returns, the pandemic anxiety index, and the rolling anxiety index, respectively. Δr, ΔAI, and ΔRAI denote the corresponding differential series, ^Δrt ^=rt^{- r}t−1^{, ΔAI}t ^= 100*(AI t^{- AI}t − 1^{) and ΔRAI}t ^= 100*(RAI t^{- RAI}t − 1^{). The p-values are presented in the parentheses. ***}, ^{and *** indicate significance at the 10%, 5%, and 1% levels, respectively.}

Moreover, the corresponding ARCH test results for the differential series ΔAI, except for India, and for the series ΔRAI, except for Germany, are statistically significant. This highlights that the data is suitable to estimate a GARCH model using the changes in stock market returns and the pandemic anxiety indexes, namely Δr, ΔAI, and ΔRAI.

4. Empirical results

Table 5 summarizes the unconditional correlation matrix and the results of the Granger causality tests as shown in Eqs. (3) and (4) for each country, where the optimal lag orders are selected based on the Bayesian Information Criterion (BIC) shown in Table 4 . As shown on the left side of Table 5 , the Chinese stock market has the weakest correlation with its domestic pandemic anxiety index (C_{Δr- ΔAI} = 0.0127; C_Δr-ΔRAI = 0.0040), while India has the strongest unconditional correlation with its domestic pandemic anxiety index (C_Δr-ΔAI = −0.5729; C_Δr-ΔRAI = −0.6318).

Table 5.

Summary of unconditional correlation matrix and Granger Causality test results.

	Correlation Matrix				Granger Causality Test
	CΔr-ΔAI	CΔr-ΔRAI	CΔF-ΔRAI		ΔAI causing Δr	ΔRAI causing Δr
Brazil	0.3255	0.3292	0.5281		2.7054⁎⁎(−0.0314)	6.5700⁎⁎⁎(0.0001)
Russian	−0.1391	−0.2245	0.2862		0.5111(0.6751)	2.1687*(0.0736)
India	−0.5729	−0.6318	0.6725		13.6857⁎⁎⁎(0.0000)	6.6752⁎⁎⁎(0.0000)
China	0.0127	0.004	0.3959		1.9761(0.1411)	0.7695(0.5462)
South Africa	0.0906	0.2781	0.4568		0.1676(0.9547)	3.6041⁎⁎⁎(0.0073)
United States	0.0183	0.2109	0.2632		0.4794(0.7509)	4.0446⁎⁎⁎(0.0035)
England	−0.0807	0.0991	0.6973		0.4443(0.6418)	2.1412(0.0769)
Canada	0.1061	0.0223	0.8006		3.3055⁎⁎(0.0118)	7.5664⁎⁎⁎(0.0000)
Italy	0.1016	0.08	0.3479		0.5874(0.6721)	0.0436(0.9573)
Germany	−0.0420	0.1349	0.2941		0.3170(0.8131)	2.6677⁎⁎(0.0486)
France	0.0032	−0.0244	0.5676		1.0063(0.4052)	0.3361(0.8534)
Japan	−0.1360	−0.1491	0.6299		8.4998⁎⁎⁎(0.0039)	7.3871⁎⁎⁎(0.0071)

Note:The left side of this table presents unconditional correlations between stock market returns and COVID-19 anxiety indexes for each country; the right side shows the results of the Granger causality tests. C_Δr-ΔAI denotes the correlation coefficient between daily changes in stock market returns (Δr) and pandemic anxiety index (ΔAI), C_Δr-ΔRAI is the correlation coefficient between daily changes in stock market returns (Δr) and pandemic rolling anxiety index (ΔRAI), while C_ΔF-ΔRAI denotes the correlation between AI and RAI. The first Granger causality test, for ΔAI causing Δr, uses Eq. (3), while the second Granger causality test, for ΔRAI causing Δr, uses Eq. (4). F statistics from the Granger causality tests are reported and corresponding p-values are shown in parentheses. Rejection of the null hypothesis means ΔAI Granger-causes Δr or ΔRAI Granger-causes Δr. Parameter estimates results for the two Granger causality tests are reported in Table 8, Table 9, respectively. ^⁎⁎⁎, and *** indicate significance at the 10%,5%, and 1% level, respectively.

Table 4.

Results of optimal lags order selection.

	Lags selection for stock-AI						Lags selection for stock-RAI
Lags	0	1	2	3	4		0	1	2	3	4
Brazil	12.258	11.622	11.533	11.387	11.279*		10.145	9.608	9.409	9.385	9.382*
Russian	11.194	10.744	10.694	10.632*	10.681		11.579	11.198	11.015	10.948	10.933*
India	10.071	9.500	9.404*	9.422	9.423		9.515	8.871	8.735	8.688	8.603*
China	12.349	12.035	12.029*	12.057	12.088		17.470	16.999	16.886	16.887	16.858*
South Africa	12.624	12.285	12.076	11.994	11.986*		11.557	11.377	11.117	10.998	10.990*
United States	13.202	12.604	12.547	12.574	12.529*		9.971	8.632	8.647	8.508	8.455*
England	11.293	10.866	10.815*	10.825	10.883		9.217	9.030	8.954	8.868	8.866*
Canada	12.407	11.732	11.619	11.475	11.399*		10.787	9.675	9.440	9.280	9.182*
Italy	11.257	10.738	10.657	10.659	10.620*		8.013	7.631	7.491*	7.508	7.500
Germany	12.024	11.530	11.548	11.486*	11.504		9.939	9.474	9.477	9.407*	9.423
France	13.438	12.776	12.738	12.725	12.639*		11.244	10.726	10.690	10.573	10.451*
Japan	11.929	11.247*	11.314	11.276	11.257		10.211	9.522*	9.559	9.548	9.529

Notes: This table reports the results of lag selections for bivariate VAR model based on the BIC criterion. The selected lags will be used to take Granger tests for whether the change in pandemic anxieties causing change in stock market returns and also will be used in the DCC-GARCH model hereinafter. The left part shows the results for pandemic anxiety index AI and stock market returns while the right part shows the results for pandemic rolling anxiety index RAI and stock market returns in each country. * presents the optimal lag order.

The results of the Granger causality tests shown in the right side of Table 5 indicate that the null hypothesis is rejected for 8 out of 12 selected countries, namely, Brazil, Russia, India, South Africa, the United States, Canada, Germany and Japan, indicating that the COVID-19 Anxiety Index changes, causing stock market returns in these eight countries. Interestingly, among these eight countries, the unconditional correlations between the changes in stock market returns and changes in the COVID-19 Anxiety Index are positive for Brazil, South Africa, the United States, Canada and Germany, while the correlations are negative for Russia, India and Japan. It highlights that the impact of the COVID-19 pandemic on stock market are not consistent for different countries.

Tables 6 and 7 show the estimated parameters of the bivariate DCC–GARCH model based on Eqs. (5) to (10). Table 6 reports the results from the model for stock market returns and the AI, while results for the model involving stock returns and the RAI, are presented in Table 7. We concentrate on the estimated parameters of the dynamic conditional correlation based on Eq. (9), shown in the columns labeled a^DCCand b^DCC.

Table 6.

Estimate results of DCC–GARCH (1,1) model for stock volatility and Anxiety Index (AI).

	ωr	ωAI	αr	αAI	βr	βAI	aDCC	bDCC
Brazil	1.6463⁎⁎⁎	6.4087⁎⁎	0.6352⁎⁎⁎	0.1860⁎⁎⁎	0.2439⁎⁎⁎	0.7702⁎⁎⁎	0.0248	0.9621⁎⁎⁎
	(0.4607)	(2.1396)	(0.0899)	(0.0548)	(0.0634)	(0.0510)	(0.0305)	(0.0153)
Russian	0.4971	0.1097	0.2394	0.0682	0.7144⁎⁎	0.9258⁎⁎⁎	0.0194	0.7363⁎⁎
	(0.7036)	(0.1162)	(0.2780)	(0.0561)	(0.3090)	(0.0732)	(0.0319)	(0.2851)
India	0.2640	0.1951	0.1757	0.0766⁎⁎⁎	0.6964	0.9101⁎⁎⁎	0.0429	0.9320⁎⁎⁎
	(0.6665)	(0.1616)	(0.3545)	(0.0263)	(0.6183)	(0.0218)	(0.041)	(0.0364)
China	0.9256⁎⁎⁎	17.6722⁎⁎⁎	0.0866	0.0578	0.4225⁎⁎	0.4374⁎⁎	0.0709	0.0146
	(0.2184)	(38.5883)	(0.2184)	(0.2184)	(0.2184)	(0.2184)	(0.2184)	(0.9599)
South Africa	0.1445⁎⁎⁎	3.3042	0.2378⁎⁎	0.1450⁎⁎⁎	0.7060⁎⁎⁎	0.8490⁎⁎⁎	0.0742	0.5271⁎⁎⁎
	(0.0523)	(2.2516)	(0.0936)	(0.0292)	(0.0774)	(0.0189)	(0.1050)	(0.1165)
United States	0.4314⁎⁎⁎	15.1446⁎⁎	0.6796⁎⁎⁎	0.9253⁎⁎	0.3110⁎⁎⁎	0.0742	0.0258⁎⁎	0.9470⁎⁎⁎
	(0.1627)	(6.39)	(0.1148)	(0.3934)	(0.0809)	(0.0556)	(0.0121)	(0.0422)
England	0.2193	10.2107⁎⁎	0.1334	0.5301*	0.8254⁎⁎⁎	0.3502⁎⁎	0.1783	0.4835
	(0.1850)	(5.1142)	(0.0828)	(0.2107)	(0.1159)	(0.1514)	(0.1039)	(0.3451)
Canada	0.3821	69.1409	0.7648	0.8137	0.2515	0.1028	0.2347	0.1644
	(0.1147)	(38.7416)	(0.1507)	(0.3302)	(0.0936)	(0.1108)	(0.3751)	(0.179)
Italy	1.0155⁎⁎⁎	3.0054	0.4703⁎⁎⁎	0.5603*	0.1108	0.4460⁎⁎	0.0683	0.8970⁎⁎⁎
	(0.2568)	(3.4139)	(0.1248)	(0.3066)	(0.1032)	(0.2004)	(0.0509)	(0.0579)
Germany	0.0797⁎⁎⁎	1.1032⁎⁎⁎	0.0658⁎⁎⁎	0.0443⁎⁎⁎	0.8999⁎⁎⁎	0.9351⁎⁎⁎	0.1141⁎⁎⁎	0.4982
	(0.0051)	(0.0034)	(0.0197)	(0.0139)	(0.0238)	(0.0753)	(0.0012)	(0.3005)
France	1.1899	3.2664	0.4854⁎⁎⁎	0.2962*	0.3930*	0.7015⁎⁎⁎	0.2911⁎⁎⁎	0.1649
	(0.8003)	(2.8436)	(0.1652)	(0.1521)	(0.2150)	(0.0791)	(0.0874)	(0.2358)
Japan	1.1643⁎⁎⁎	0.7901	0.6003⁎⁎⁎	0.2898⁎⁎	0.0483	0.7020⁎⁎⁎	0.1394*	0.6182⁎⁎
	(0.2387)	(1.0380)	(0.1629)	(0.1354)	(0.0703)	(0.1102)	(0.0713)	(0.3066)

Table 7.

Estimate results of DCC-GARCH (1,1) model for stock volatility and Rolling Anxiety Index (RAI).

	ωr	ωRAI	αr	αRAI	βr	βRAI	aDCC	bDCC
Brazil	1.2649⁎⁎⁎	0.5175	0.3622⁎⁎	0.2780	0.2685	0.6539⁎⁎⁎	0.0739	0.8238⁎⁎⁎
	(0.4675)	(0.4731)	(0.1661)	(0.2022)	(0.2207)	(0.2265)	(0.0665)	(0.1356)
Russian	0.197	0.0191	0.0907	0.5384*	0.8605⁎⁎⁎	0.4451⁎⁎⁎	0.0099	0.9716⁎⁎⁎
	(0.1991)	(0.0182)	(0.0709)	(0.3085)	(0.1047)	(0.063)	(0.0673)	(0.0175)
India	0.4005⁎⁎	0.0277	0.7711*	0.7957*	0.2015⁎⁎	0.1028*	0.1453⁎⁎⁎	0.7985⁎⁎⁎
	(0.1678)	(0.0240)	(0.4325)	(0.2568)	(0.1003)	(0.0492)	(0.0368)	(0.0372)
China	0.6885⁎⁎⁎	26.3169⁎⁎⁎	0.3957⁎⁎⁎	0.2301⁎⁎⁎	0.3034⁎⁎⁎	0.1227⁎⁎⁎	0.0476	0.1034
	(0.0022)	(3.3467)	(0.0017)	(0.0026)	(0.0031)	(0.0023)	(0.0314)	(0.0012)
South Africa	0.1301	10.9143⁎⁎⁎	0.1774*	0.8362⁎⁎⁎	0.7923⁎⁎⁎	0.1187	0.1196	0.7289⁎⁎⁎
	(0.0844)	(3.8207)	(0.0900)	(0.2351)	(0.1092)	(0.1316)	(0.1340)	(0.1425)
United States	0.2971	0.6368	0.8180⁎⁎⁎	0.8119*	0.1982⁎⁎⁎	0.1388	0.2701	0.7139⁎⁎⁎
	(0.6726)	(1.005)	(0.1309)	(0.3481)	(0.0647)	(0.1772)	(1.1600)	(0.1334)
England	0.1103	0.3312	0.1014	0.1995⁎⁎⁎	0.8470⁎⁎⁎	0.7654⁎⁎⁎	0.0601	0.4917⁎⁎⁎
	(0.0947)	(0.2337)	(0.0691)	(0.0671)	(0.0888)	(0.0628)	(0.0483)	(0.1489)
Canada	0.1828⁎⁎⁎	4.0735⁎⁎⁎	0.5727⁎⁎⁎	0.4548⁎⁎⁎	0.4155⁎⁎⁎	0.4073⁎⁎⁎	0.0429⁎⁎⁎	0.7203⁎⁎⁎
	(0.0072)	(0.1082)	(0.0002)	(0.0006)	(0.003)	(0.0037)	(0.0034)	(0.0290)
Italy	1.0447⁎⁎⁎	0.1130	0.5204⁎⁎⁎	0.1653⁎⁎⁎	0.0754	0.8186⁎⁎⁎	0.2025⁎⁎	0.0871
	(0.2809)	(0.0697)	(0.1190)	(0.0567)	(0.0802)	(0.0572)	(0.0913)	(0.2300)
Germany	0.1999	2.8437	0.2097⁎⁎	0.2082⁎⁎	0.7446⁎⁎⁎	0.5701⁎⁎⁎	0.0103	0.7635⁎⁎⁎
	(0.1223)	(1.9064)	(0.0898)	(0.099)	(0.0937)	(0.2153)	(0.0576)	(0.2485)
France	0.1810⁎⁎⁎	0.2317⁎⁎	0.1771⁎⁎⁎	0.1316⁎⁎⁎	0.7763⁎⁎⁎	0.8455⁎⁎⁎	0.1039	0.5188⁎⁎⁎
	(0.0415)	(0.1131)	(0.0186)	(0.0126)	(0.0149)	(0.0086)	(0.0643)	(0.199)
Japan	1.2143⁎⁎⁎	−0.0159	0.5887⁎⁎⁎	0.2250⁎⁎⁎	0.0251	0.7518⁎⁎⁎	0.0271	0.5863*
	(0.2130)	(0.1011)	(0.1629)	(0.0830)	(0.0506)	(0.0547)	(0.2460)	(0.2508)

Notes: this table reports the estimate results of the bivariate DCC-GARCH (1,1) model involving stock return and COVID-19 rolling anxiety index (RAI). a_r, and a_r are the parameters of univariate GARCH (1,1) model for stock return while ω_RAI, α_RAI  and  β_RAI  are the parameters of univariate GARCH (1,1) for RAI. a^DCC and b^DCC are parameters of the DCC procedure. Standard errors are presented in parentheses. ^⁎⁎⁎and ^⁎⁎⁎ indicate the significance level at 10%,5% and 1%, respectively.

For the DCC between stock market returns and the AI, the estimated results of a^DCC are significant for the United States, Germany, France and Japan, while the estimates for b^DCC are significant in all cases except for China, England, Canada, Germany and France, as shown in Table 6. The estimated parameters a^DCC and b^DCCfor the dynamic conditional correlation between stock market returns and the RAI, are shown in Table 7. The results for a^DCC are significant for India, Canada and Italy, while the estimates for b^DCCare statistically significant for Brazil, Russia, India, South Africa, the United States, England, Canada, Germany, France and Japan.

The time-varying conditional correlations between the stock market and AI and RAI for each country are depicted in Fig. 4, Fig. 5 , respectively. We found that the conditional correlations between the stock market and the two pandemic anxiety indexes were time varying in all cases. With respect to the range of dynamic conditional correlations, the links between the stock market and both the pandemic anxiety indexes are weakest in China (−0.2~ 0.2 for AI; −0.05~0.05 for RAI), reflecting the fact that the number of reported COVID-19 cases in China is the lowest among these countries, in line with the lowest unconditional correlations and the insignificant Granger causality test result shown in Table 5. The range of correlations between the stock market and AI is largest in India (−1.0~0.5), while the range of correlations for the stock market and RAI is strongest in the United States (−1.0~1.0). It seems that the lower the pandemic AI, the less correlated it is with the stock market returns. A decrease in the anxiety index could be attributed to greater knowledge of the COVID-19 virus, more preventive measures, and especially, the availability of a vaccine. On November 18,2020, Pfizer Inc. and BioNTech SE announced the results of the Phase 3 clinical trial of their mRNA-based COVID-19 vaccine, which showed a efficacy rate of 95% (p < 0.0001).

Fig. 4.

Dynamic conditional correlations between stock market returns and the COVID-19 AI for each country. Note: Graphs are portioned into two phases by a dotted line, depicting the co-movements of stock markets and the AI before and after the announcement of the mRNA-based COVID-19 vaccine.

Fig. 5.

Dynamic conditional correlations between stock market returns and the COVID-19 RAI for each country. Note: Graphs are portioned into two phases by a dotted line, depicting the co-movements of stock markets-RAI before and after the announcement of the mRNA-based COVID-19 vaccine.

As shown in Figs. 4 and 5, after the COVID-19 vaccine was announced, the co-movements between the stock markets and pandemic anxiety indexes weakened and fluctuations in the correlations declined in the twelve countries, despite the fact that the number of reported cases and deaths (except in India and China) hit their highest levels during this period (see Fig. 2). In addition, except in India, the highest correlations occurred in the first half of 2020, corresponding to the beginning or the first small surge in the pandemic for each country.

5. Conclusion

The COVID-19 pandemic spread around the world in 2020, and the link between the pandemic and financial markets is worth studying empirically. In this study, we constructed two anxiety indexes (AI and RAI) related to the COVID-19 pandemic based on daily reported cases and deaths and used them to examine the co-movements between the pandemic and stock market returns in the BRICS and G7 countries, employing a bivariate DCC–GARCH model over the period from March 23, 2020 to January 29, 2021.

We showed that the COVID-19 pandemic anxiety indexes were volatile over the aforementioned period but displayed a downtown trend over time for each country. Even when the number of daily reported cases and deaths hit the highest level in the second half of 2020, the pandemic anxiety indexes were still below the previous level reached in the early stages of the pandemic. We also showed that the conditional correlations between stock market returns and the two pandemic anxiety indexes fluctuate over time for each country, experiencing phases of increased and decreased correlation. A common feature of the co-movements between stock market returns and the pandemic anxiety indexes, found for all of the BRICS and G7 countries, is that the correlations weakened and had smaller fluctuations after the COVID-19 vaccine results were announced. Moreover, compared with other countries, the results show that China's stock market has the weakest correlation with its domestic COVID-19 anxiety indexes.

One weakness of our study is the fact that the empirical results show that the correlations between changes in stock returns and changes in the COVID-19 pandemic anxiety are not consistent and stable for different countries, sometimes positive for some countries and sometimes negative for the others, but we cannot figure out the specific reasons behind these phenomena, in that there are many other sources of uncertainty or outcomes (such as economic policy, pandemic prevention measures and the United States Presidential election, etc.) that affect stock returns. Secondly, we construct the anxiety index based on the assumption that people's emotions fluctuate with the changes of COVID-19 reported cases and death. However, the actual sentiment changes are more complicated than that in real life. In addition, we just select 12 countries but not all countries in the world for this study. Some other countries successfully suppressing the spread of COVID-19 (e.g., New Zealand, Vietnam) could have been useful for comparing China's results. As part of future analysis, it would be interesting to extend our study to resolve these limitations.

6. Authorship contribution statement

Xiaoling Yu: Conceptualization; Methodology; Formal analysis; Writing - original draft; Revision; Writing - review & editing.

Kaitian Xiao: Conceptualization; Data curation; Software; Writing - original draft; Writing - review & editing; Revision.

Junping Liu:Conceptualization; Writing - original draft; Revision; Writing - review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix

Table 8.

Estimated results of the Granger causality test for the AI causing stock market returns.

	ϕi	αi,1	αi,2	αi,3	αi,4	βi,1	βi,2	βi,3	βi,4
Brazil	−0.0655	−0.8618⁎⁎⁎	−0.5675⁎⁎⁎	−0.4195⁎⁎⁎	−0.258⁎⁎⁎	−0.0271*	−0.0123	−0.0389⁎⁎	−0.0336⁎⁎
	(0.1270)	(0.0653)	(0.0798)	(0.0743)	(0.0622)	(0.0147)	(0.0168)	(0.0165)	(0.0136)
Russian	−0.0447	−0.8185⁎⁎⁎	−0.5919⁎⁎⁎	−0.3428⁎⁎⁎		−0.0109	−0.032	−0.0103
	(0.1396)	(0.0645)	(0.0737)	(0.0619)		(0.0239)	(0.0259)	(0.0238)
India	0.0167	−0.4933⁎⁎⁎	−0.1756⁎⁎⁎			0.0925⁎⁎⁎	0.0738⁎⁎⁎
	(0.1044)	(0.0617)	(0.0538)			(0.0226)	(0.0172)
China	−0.0112	−0.6061⁎⁎⁎	−0.3035⁎⁎⁎			−0.001	−0.0088*
	(0.0809)	(0.0629)	(0.0620)			(0.0046)	(0.0045)
South Africa	−0.058	−0.7296⁎⁎⁎	−0.5552⁎⁎⁎	−0.397⁎⁎⁎	−0.1728⁎⁎⁎	−0.0035	−0.0066	−0.0038	−0.0009
	(0.0981)	(0.0675)	(0.0791)	(0.0756)	(0.0610)	(0.0072)	(0.0087)	(0.0086)	(0.0071)
United States	−0.0565	−0.8107⁎⁎⁎	−0.3611⁎⁎⁎	−0.3188⁎⁎⁎	−0.2764⁎⁎⁎	−0.0059	−0.0039	0.0015	−0.0055
	(0.1129)	(0.0651)	(0.0818)	(0.0736)	(0.0573)	(0.0074)	(0.0083)	(0.0083)	(0.0073)
England	−0.0297	−0.4881⁎⁎⁎	−0.2167⁎⁎⁎			0.0119	0.0149
	(0.1163)	(0.0633)	(0.0607)			(0.0166)	(0.0165)
Canada	−0.0593	−0.9008⁎⁎⁎	−0.658⁎⁎⁎	−0.535⁎⁎⁎	−0.2658⁎⁎⁎	−0.011	−0.0119	−0.0208⁎⁎	−0.0213⁎⁎⁎
	(0.0835)	(0.0637)	(0.0768)	(0.0675)	(0.0542)	(0.0070)	(0.0091)	(0.0088)	(0.0063)
Italy	−0.0304	−0.8295⁎⁎⁎	−0.5674⁎⁎⁎	−0.4662⁎⁎⁎	−0.2858⁎⁎⁎	−0.0039	−0.0124	−0.0019	−0.0044
	(0.0756)	(0.0659)	(0.0811)	(0.0808)	(0.0647)	(0.0085)	(0.0101)	(0.0101)	(0.0085)
Germany	−0.0503	−0.6954⁎⁎⁎	−0.4088⁎⁎⁎	−0.2943⁎⁎⁎		−0.0104	−0.0012	0.0071
	(0.1246)	(0.0647)	(0.0732)	(0.0612)		(0.0142)	(0.0161)	(0.0143)
France	−0.0449	−0.7338⁎⁎⁎	−0.4766⁎⁎⁎	−0.4337⁎⁎⁎	−0.1934⁎⁎⁎	−0.0028	0.0122	0.01	0.0046
	(0.1187)	(0.0671)	(0.0781)	(0.0757)	(0.0632)	(0.0073)	(0.0094)	(0.0094)	(0.0073)
Japan	−0.0083	−0.5195⁎⁎⁎				0.0179⁎⁎⁎
	(0.0937)	(0.0560)				(0.0061)

Notes: This table shows the parameter estimate results of the Granger causality tests for changes in the COVID-19 AI causing changes in stock market returns for the BRICS and G7 countries. φi,α_i,1, α_i,2, α_i,3, α_i,4, β_i,1, β_i,2, β_i,3 and β_i,4 are the parameters based on Eq. (3), with the standard errors presented in the parentheses. ***, and *** indicate significance at 10%, 5%, and 1% levels, respectively.

Table 9.

Estimated results of the Granger causality test for the RAI causing stock market returns.

	ϕi	αi,1	αi,2	αi,3	αi,4	βi,1	βi,2	βi,3	βi,4
Brazil	−0.0041	−0.8932⁎⁎⁎	−0.5875⁎⁎⁎	−0.3561⁎⁎⁎	−0.2163⁎⁎⁎	−0.0155	0.0582⁎⁎⁎	0.0228*	0.013
	(0.1237)	(0.0668)	(0.0872)	(0.0836)	(0.0637)	(0.0410)	(0.0152)	(0.0130)	(0.0092)
Russian	−0.0581	−0.8563⁎⁎⁎	−0.6647⁎⁎⁎	−0.4787⁎⁎⁎	−0.0989	−0.0297	0.028	−0.0044	0.0413*
	(0.1365)	(0.0694)	(0.0855)	(0.0837)	(0.0689)	(0.0230)	(0.0240)	(0.0228)	(0.0217)
India	0.0149	−0.7646⁎⁎⁎	−0.5042⁎⁎⁎	−0.2699⁎⁎⁎	−0.0971⁎⁎⁎	0.074⁎⁎	0.1029⁎⁎	0.1878⁎⁎⁎	0.1227⁎⁎⁎
	(0.0923)	(0.0641)	(0.0718)	(0.0669)	(0.0576)	(0.0347)	(0.0440)	(0.0404)	(0.0283)
China	−0.017	−0.7152⁎⁎⁎	−0.4928⁎⁎⁎	−0.3441⁎⁎⁎	−0.2582⁎⁎⁎	0.0003	−0.0001	0.0004	0.0003
	(0.0780)	(0.0659)	(0.0788)	(0.0769)	(0.0639)	(0.0004)	(0.0005)	(0.0005)	(0.0004)
South Africa	−0.0499	−0.6905⁎⁎⁎	−0.5883⁎⁎⁎	−0.4475⁎⁎⁎	−0.2364⁎⁎⁎	−0.0389⁎⁎⁎	−0.0075	−0.0215⁎⁎	0.0072
	(0.0951)	(0.0671)	(0.0772)	(0.0779)	(0.0630)	(0.0110)	(0.0111)	(0.0106)	(0.0104)
United States	−0.0257	−0.7892⁎⁎⁎	−0.3464⁎⁎⁎	−0.2192⁎⁎⁎	−0.1875⁎⁎⁎	−0.0758	0.0171	0.0531⁎⁎	−0.0022
	(0.1104)	(0.0671)	(0.0838)	(0.0803)	(0.0635)	(0.0509)	(0.0366)	(0.0230)	(0.0182)
England	−0.0285	−0.5937⁎⁎⁎	−0.424⁎⁎⁎	−0.3505⁎⁎⁎	−0.1575⁎⁎	0.0228	0.0249	−0.0241	0.0899⁎⁎
	(0.1108)	(0.0669)	(0.0742)	(0.0732)	(0.0617)	(0.0432)	(0.0462)	(0.0446)	(0.0402)
Canada	−0.0508	−0.8365⁎⁎⁎	−0.5888⁎⁎⁎	−0.4819⁎⁎⁎	−0.2482⁎⁎⁎	−0.0195	0.0047	−0.0003	−0.0441⁎⁎⁎
	(0.0809)	(0.0634)	(0.0778)	(0.0719)	(0.0561)	(0.0175)	(0.0245)	(0.0224)	(0.0145)
Italy	−0.0218	−0.7043⁎⁎⁎	−0.2709⁎⁎⁎			0.0118	0.0028
	(0.0805)	(0.0644)	(0.0641)			(0.0400)	(0.0398)
Germany	−0.0282	−0.7001⁎⁎⁎	−0.3726⁎⁎⁎	−0.2800⁎⁎⁎	−0.0449	0.0469	0.0299⁎⁎
	(0.1229)	(0.0645)	(0.0738)	(0.0604)	(0.0369)	(0.0367)	(0.0147)
France	−0.0455	−0.7319⁎⁎⁎	−0.4824⁎⁎⁎	−0.4288⁎⁎⁎	−0.1979⁎⁎⁎	0.0122	0.0052	−0.0011	0.0157
	(0.1196)	(0.0670)	(0.0788)	(0.0766)	(0.0638)	(0.0216)	(0.0262)	(0.0263)	(0.0209)
Japan	−0.0109	−0.5189⁎⁎⁎				0.0427⁎⁎⁎
	(0.0939)	(0.0563)				(0.0157)

Notes: This table shows the parameter estimate results of the Granger causality tests for changes in the COVID-19 RAI causing changes in stock market returns for the BRICS and G7 countries. φi,α_i,1, α_i,2, α_i,3, α_i,4, β_i,1, β_i,2, β_i,3 and β_i,4 are the parameters based on Eq. (4) with the standard errors presented in parentheses. ***, and ^⁎⁎⁎ indicate significance at the 10%, 5%, and 1% levels, respectively.