Volatility Persistence and Spillover Effects of Indian Market in the Global Economy: A Pre- and Post-Pandemic Analysis Using VAR-BEKK-GARCH Model

Abstract

This study examines how the COVID-19 pandemic impacted stock market volatility and interconnectedness between India and other selected global economies. The analysis, using data from 2016 to 2024, reveals a substantial rise in volatility within both the Indian market and those of several other countries after the pandemic. Interestingly, the volatility transmission patterns also changed. While the Indian market’s volatility significantly influenced Brazil, China, and Mexico throughout the entire period, the influence of the US market became negligible post-pandemic. In contrast, Russia exhibited a weak but statistically significant impact on India’s volatility only after the pandemic. These findings highlight the lasting impact of the pandemic on global financial markets and emphasize the need for investors and policymakers to adapt. By understanding these new dynamics, investors can make more informed decisions, and policymakers can develop stronger risk management strategies and international coordination during periods of increased volatility. This study offers valuable insights for navigating the current financial landscape and the interconnectedness of emerging economies.

1. Introduction

Globalization and financial liberalisation have given rise to a more interrelated and connected world economy than ever before. Stock markets have been globalised and interconnected as a result of technological advancements and financial deregulations. The benefits of a globalised stock market are well documented, yet there are also certain drawbacks associated with liberalised stock markets. As a result, investors, financial institutions, and government organisations can benefit from a comprehensive understanding of the interconnectedness and interdependencies among various stock markets (Zhou et al. 2012). On the other hand, globalized stock markets have several advantages, including lower capital costs and more economic investment and economic development (Bae and Zhang 2015). In addition, developing markets have been a popular choice for international investors seeking diversity (Kearney 2012; Vo and Ellis 2018). Specifically, an increasing tendency toward international and regional trade agreements among nations has supported the development of improved economic integration (Balli et al. 2015). Moreover, these trade agreements facilitate the flow of capital, goods, and services across borders, further strengthening the economic ties between countries. This increased economic integration not only boosts the performance of stock markets but also enhances the resilience of economies to global financial shocks (Liu et al. 2022).

However, the interconnectedness of global stock markets also introduces risks. For instance, financial crises can spread more quickly across borders due to the high degree of market integration. The 2008 financial crisis exemplifies how financial turmoil in one part of the world can rapidly affect other regions (Longstaff 2010). Therefore, while the benefits of interconnected markets are substantial, they come with the need for robust regulatory frameworks and international cooperation to effectively manage and mitigate systemic risks.

Prior studies on stock volatility spillover identified three fundamental aspects, namely ‘unidirectional (flow of volatility from one market to another), bidirectional (among stock markets) and non-persistence volatility’ (Ngo Thai 2019). The past two decades have witnessed a surge in research on financial interconnectedness, particularly between developed and developing economies. This shift in focus coincides with the rise of emerging economies and their increasing integration into the global financial system, fueled by factors like financial liberalization. While this integration offers benefits like access to cheaper capital and portfolio diversification, it also exposes markets to a heightened risk of contagion. The 1997 Asian crisis and the 2007 subprime crisis serve as stark examples. Studies by Srivastava et al. (2015) link these long-lasting crises to increased stock market integration. Similarly, Bae and Zhang (2015) also found a strong correlation between integration and periods of crisis. These findings are further supported by numerous empirical studies demonstrating the growing interconnectedness of stock prices on a global (Bai et al. 2021; Baker et al. 2020)

Studies on the economic impact of the pandemic have been growing rapidly since 2020 (Albulescu 2020; Ashraf 2020; Baker et al. 2020; Fernandes 2020; Gormsen and Koijen 2020; He et al. 2020a; Li et al. 2023; Ozili 2020; Thangamuthu et al. 2022; Yusuf et al. 2020). He et al. (2020a) investigate the effect of global pandemic and daily stock return spillovers to notice enormous economic devastation globally. Many preliminary studies claim a long-term influence of the pandemic on nearly every portion of the economy and spurred researchers into extensive investigations. This study adds to the expanding body of knowledge by examining the influence of the pandemic on stock market volatility on some emerging stock indexes using the vector autoregressive model (VAR) (Sims 1980) and the Baba–Engle–Kraft–Kroner multivariate generalized autoregressive conditionally heteroskedastic (BEKK-GARCH) models (Engle and Kroner 1995). Considering this backdrop, the study aims to evaluate the nature of return and volatility transmissions between the Indian stock market and five developing markets: Mexico, Russia, New York, China, and Brazil.

As we transition from the pre-COVID scenario to an unexpected global shutdown following a global pandemic and then to the post-COVID-19 period, it becomes crucial to examine how the landscape of volatility spillover has evolved in the wake of the pandemic. The unprecedented and far-reaching impacts of the COVID-19 crisis on global financial markets may have significantly altered the dynamics of volatility transmission from the Indian stock market to the selected global economies. Thus, understanding the post-COVID volatility spillover necessitates assessing the extent to which the shockwaves of the pandemic have reshaped the interconnectedness and influence of the Indian stock market on the volatility of the other global markets in focus. By exploring the post-pandemic period, along with a direct comparison with the pre-COVID period, we can uncover the shifts in volatility dynamics, identify potential amplification or dampening of spillover effects, and explore the interactions between the Indian stock market and the stock markets of the selected developed nations such as China, Russia, Brazil, Mexico, and New York remain the prime objective of the study.

2. Literature Review

Based on myriad grounds for curiosity in the impact and spillover effects of pandemic-driven financial crises on stock markets, an unexpected increase in research has been observed in the past two years. However, studies conducted in the year 2020, i.e., the year when the pandemic had its first and severest impact on life and livelihoods across the world, focused on the economic and sectoral impact (Baker et al. 2020; Onali 2020; Phan and Narayan 2020). Eventually, many studies also investigated stock market integration and spillover effects. Several studies looked at market correlations, while others looked into volatility spillover across developed and developing markets. Apart from studying stock market integration, some studies also focused on the impact of market liberalisation and information spillover across nations.

Several recent studies have employed the VAR-BEKK-GARCH model to analyse volatility spillover between the Indian stock market and other emerging nations (Malik et al. 2022; Mishra et al. 2022; Sahoo and Kumar 2024; Yadav et al. 2023). Yadav et al. (2023) applied the DCC-GARCH model (a variant of BEKK-GARCH) to data spanning 2012–2022 and reported short-term volatility spillover from developed markets to India. However, the impact seems to lessen during economic slowdowns in India, suggesting a degree of insulation. Further, they highlighted the importance of considering business cycles while analysing volatility transmission. Mishra et al. (2022) in their study focused on the pre- and post-2008 financial crisis periods, using the BEKK-GARCH model and revealed a stronger association between the Indian market (represented by the Nifty 50) and the US and Hong Kong indices and highlighted a strong integration between these specific markets. Interestingly, the study also found a decrease in co-integration coefficients during the recession, pointing toward reduced investor confidence in developing economies during such periods. Beyond traditional financial sectors, similar methodologies have been applied to explore volatility spillover across broader industry segments. Among emerging and developed markets, studies by Mishra et al. (2022) and Sahoo and Kumar (2024) using BEKK-GARCH models, showcase significant volatility spillover in sectors like IT, healthcare, and real estate. This implies that events impacting these sectors in developed markets can have a ripple effect on their emerging market counterparts. Thangamuthu et al. (2022) examined the volatility from five global markets; United States (S&P 500), China (SSE Composite), Germany (DAX), Australia (ASX 200), and Japan (Nikkei) and their impact on India’s stock market before and after COVID-19. They found a dampening effect from most markets pre-COVID, while only the US market significantly influenced Indian volatility post-COVID. As such, Li et al. (2023) provide an insightful analysis of risk dynamics in the post-pandemic era, particularly highlighting the roles of US and Asian stock markets. Their study reveals that, during this period, stock markets acted as primary transmitters of risk spillover, whereas economic policy uncertainty (EPU) served predominantly as the recipient of these risks.

Stock market volatility and returns are often influenced by investor behaviour through changing investor sentiment driven by any crisis or emergency. The 1997 Asian financial crisis and the 1998 Hong Kong avian influenza pandemic had a severe negative impact on the tourism sector (Colther and Arriagada-Millaman 2021; Goh and Law 2002). Similarly, the influence of the flu affected the US stock market adversely by reducing trading activity (Baker et al. 2020; Tse et al. 2011). Similarly, a study by (Chen et al. 2007) claimed that the hotel business in Taiwan collapsed in the wake of SARS. Interestingly, studies on the long-term impact of the SARS revealed a considerable influence on the stock market’s financial integration (Bai et al. 2021; Chen et al. 2007, 2018).

The recent global pandemic due to COVID-19 has had a massive and long-term detrimental impact on the global economy as a public health incident of global significance (Iyke 2020). An early study investigating the impact of the COVID-19 pandemic reported a negative shock to global stock markets where emerging and small firms are hugely affected. However, they noticed positive abnormal returns for large firms in the United States (US) due to federal stimulus compared to other emerging markets (Harjoto et al. 2021). Less developed economies, where the population density is higher along with poor healthcare infrastructure, experienced a greater impact from the pandemic (McKibbin and Fernando 2020). The quick spread and severity of COVID-19 (Barro et al. 2020) affected not only the economy but also the spread of news every day (Baker et al. 2020), as did the aggressive fiscal policy of the government (Ma et al. 2020). Nevertheless, studies also claimed that larger firms are quite resilient to such unexpected economic shocks compared to other, small and medium firms due to their monopoly in the market (Yan 2020). A study by (He et al. 2020a) on different sectors in China revealed that the transportation, mining, power, and environment industries were most affected, whereas the manufacturing, technology, education, and healthcare industries were more resilient to the pandemic. He et al. (2020b) state that COVID-19 had a negative but short-term influence on stock markets in impacted nations, with bidirectional spill-over effects between Asian, European, and American countries. Conversely, the recent pandemic showed no more adverse impacts on individual countries’ stock markets than it did on the global average. According to Gao et al. (2021), the impact of COVID-19 was devastating for the US stock market in the initial days; however, unlike the Chinese market, it eventually became resilient with the rise in COVID-19 cases. Another recent study on the BRICS nations by Malik et al. (2022) used the BEKK-GARCH model to analyse stock return data and revealed that the pandemic significantly increased volatility spillovers. USA, China, and Brazil exhibited the strongest ‘own volatility spillovers’ (persistence of their own past volatility), while Russia showed the least vulnerability to external shocks. Interestingly, the study finds the strongest long-term volatility spillover between the US and Russia, which suggests that COVID-19 heightened interconnectedness between the BRICS and US stock markets, with implications for portfolio diversification strategies during and after the pandemic. In contrast, the Chinese market exhibited a different recovery trajectory. Despite China being the epicentre of the pandemic, the Chinese stock market rebounded swiftly, supported by aggressive public health measures, economic stimulus, and a quicker return to normalcy in business operations. This divergence highlights the varying capacities of different markets to absorb and recover from global shocks, emphasizing the importance of localized factors and responses in mitigating economic disruptions. The interconnected nature of global stock markets means that such crises can result in significant volatility and uncertainty across all regions. As highlighted by Longstaff (2010), the 2008 financial crisis demonstrated how financial turmoil could propagate rapidly across interconnected markets. In the context of COVID-19, the initial shock led to widespread panic selling and a liquidity crunch, but coordinated international efforts and robust policy measures helped stabilize markets and restore investor confidence over time. Rakshit and Neog (2021) investigated how exchange rate volatility, oil prices, and COVID-19 cases impacted stock market returns and volatility in emerging economies. The authors reported that exchange rate volatility reduced returns and that oil prices had a positive relationship with returns. Volatility was higher during COVID-19 than during the global financial crisis in some countries. The study also highlighted the importance of central bank intervention to stabilize exchange rates and government policies to boost investor confidence during pandemics, contributing to the literature on emerging economies’ unique experiences during COVID-19.

Prior to the COVID-19 crisis, research extensively documented the volatility spillover effect, where global market fluctuations impacted local markets more significantly (Karolyi 1995). Examining Asian developing economies, Jebran et al. (2017) analysed volatility spillovers before and after the 2007 financial crisis. Their findings, using an EGARCH model, revealed a two-way volatility spillover between India and Sri Lanka in both periods. Interestingly, the post-crisis period displayed a strong, one-directional spillover from China to other markets. Additionally, their study highlighted the asymmetric nature of volatility spillovers.

Mukherjee and Mishra (2010) further explored market integration and volatility spillovers between India and its Asian counterparts using a GARCH model. They observed significant bi-directional volatility spillovers within a one-day period between Indian and other Asian markets. Notably, their research confirmed evidence of information flow from Hong Kong, Korea, Singapore, and Thailand to the Indian market.

Vo and Ellis (2018) investigated volatility spillover relationships between the Vietnamese stock market and developed markets before and after the 2008 financial crisis. Their analysis revealed a significant correlation in return spillovers and volatility among Vietnam, the USA, Hong Kong, and Japan. Most recently, Huong (2021) employed VAR and BEKK-GARCH models to explore return and volatility contagion during COVID-19. Their research identified a significant spillover from the Philippines, Singapore, and Thailand to the Vietnamese stock market. Moreover, the direction of the Vietnam index is the opposite of its Malaysian and the Philippines counterparts. Eventually, Bonga-Bonga and Phume (2022) evaluated the return and volatility transmission between the Nigerian and South African stock markets to find a significant shock spillover from South Africa to the Nigerian stock market. They also claimed that portfolio diversification is possible while holding simultaneous positions in the two stock markets. Despite wide-ranging prior research, there is still a need to study emerging markets during the current global pandemic crisis. This particular study will add to the growing body of existing literature by looking at how the Indian stock market affects or is affected by its developing counterparts. We hope that the empirical findings of this study will contribute to a better understanding of spillover behaviours between the selected emerging stock markets.

The COVID-19 pandemic certainly reshaped the dynamics of global financial markets, emphasizing the interconnectedness between economies and the vulnerability of both developed and emerging markets to external shocks. Before the pandemic, studies by Bonga-Bonga and Phume (2022) highlighted significant shock spillovers from South Africa to Nigeria, laying the groundwork for understanding regional market interdependencies. As the crisis unfolded, research by Ghorbel et al. (2022) revealed a marked increase in volatility spillovers among G7 countries, with the United States emerging as a key transmitter of volatility. Alongside studies by Yuan and Du (2023) and Arfaoui and Yousaf (2022) uncovered intricate patterns of volatility transmission between major economies and emerging markets, underscoring the challenges faced by the latter during global crises.

Amidst these findings, a study by Shi (2021) emphasizes the asymmetrical nature of volatility spillovers, particularly between BRICS countries and developed markets, highlighting the susceptibility of emerging economies to negative news. Against this backdrop, our current study aims to deepen our understanding of volatility spillovers by focusing on the Indian stock market and its interactions with other developing markets during the pandemic. Through advanced econometric models, we seek to analyse the evolving relationships between India and selected emerging markets, offering insights into their resilience and adaptability in the face of global shocks. This research will contribute valuable insights for investors and policymakers, informing strategies for navigating the complexities of the global financial system and fostering economic recovery in emerging economies.

3. Materials and Methods

3.1. Models

3.1.1. Vector Autoregressive Model (VAR)

To verify the interrelationship between the returns of the ‘Indian’ stock index with the stock indices of Brazil, Russia, New York, China, and Mexico, we use a vector autoregression (VAR) model with two lags (VAR(2)). Here are the simplified VAR(2) equations for this setup:

For a VAR(2) model, the equations include the current values of the variables as a function of their own past values (up to two lags) and the past values of all other variables in the system. The equations for each index return Y_i can be expressed as:

$$Y_{i,t}=c_i+\sum _{j=1}^6\left({\mathsf{\varphi }}_{ij,1}{Y}_{j,t-1}+{\mathsf{\varphi }}_{ij,2}{Y}_{j,t-2}\right)+{\mathsf{ϵ}}_{i,t}$$

where:

• Y_i,t is the return of the ith stock index at time t.
• c_i is the intercept for the ith equation.
• ϕ_ij,1 ϕ_ij,2 are the coefficients for the first and second lags of the jth index return in the ith equation. The coefficients (ϕ_ij,1 ϕ_ij,2) on the lagged values of other indices capture the interdependencies between the indices. For example, ϕ_{India,Brazil,1} and ϕ_{India,Brazil,2} measure the impact of the first and second lags of Brazil’s returns on India’s returns.
• ϵ_i,t is the error term for the ith equation.

3.1.2. BEKK-GARCH Model

The BEKK-GARCH model, which stands for the Baba–Engle–Kraft–Kroner generalized autoregressive conditional heteroskedasticity model, is an extension of the traditional GARCH (generalized autoregressive conditional heteroskedasticity) model. It is used to model and forecast the conditional variance and covariance of a multivariate time series. The BEKK-GARCH model was proposed by Engle and Kroner (1995) and can be used to investigate the volatility interdependence between two sets of time series data. The advantage of the BEKK-GRACH model over other GRACH models is the prime reason for choosing this particular model (Yong Fu et al. 2011). The main advantage of the BEKK-GARCH model is its ability to capture time-varying volatility and dynamic correlations between multiple variables simultaneously. However, estimating the model parameters can be computationally intensive, especially with large datasets, and the interpretation of the parameters can be challenging. The BEKK-GARCH model is often considered superior to other GARCH models such as DCC (dynamic conditional correlation) and E-GARCH (exponential GARCH) because it ensures a positive definite conditional covariance matrix and captures both the dynamic variance and covariance interactions between multiple time series. The DCC-GARCH model, while effective in modelling time-varying correlations, was rejected because it separately estimates conditional variances and correlations, potentially leading to less accurate covariance estimates compared to the integrated approach of the BEKK model. The E-GARCH model, although capable of capturing asymmetric effects in volatility, was deemed less suitable because it primarily focuses on univariate series and does not ensure a positive definite covariance matrix in a multivariate setting. The CCC-GARCH model was excluded due to its assumption of constant correlations, which is unrealistic for capturing the dynamic nature of financial market interdependencies, especially in the context of significant economic events such as the COVID-19 pandemic. Hence, the VAR-BEKK-GARCH model was chosen for its ability to provide a comprehensive and robust framework for analysing volatility persistence and spillover effects with guaranteed positive definiteness of the covariance matrix. Additionally, BEKK-GARCH reduces the number of parameters compared to fully parameterized multivariate models, making it computationally feasible while maintaining robustness.

Furthermore, it is widely accepted that the multivariate BEKK-GARCH parameters are more relevant while modelling volatility transmission across multiple stock markets compared to the univariate models (Arouri et al. 2012; Bauwens et al. 2006). The positive conditional covariance matrix is constructed because the model permits interaction between conditional variances and covariance, and it results in a considerable reduction in the parameter of estimation.

In a multivariate setting, the BEKK-GARCH model allows for dynamic correlations between multiple time series variables, making it particularly useful for analysing financial data in which assets may exhibit interdependencies.

The model is defined by the following equations:

• The conditional mean equation for each variable i:

  $$y_{ti}={\mu }_i+\sum _{j=1}^p{\varphi }_{ij}{y}_{t-j,i}+\sum _{k=1}^q{\theta }_{ik}{\epsilon }_{t-k,i}$$
• Conditional variance equation for each variable i:

  $$H_t=C_0{C^{\prime }}_0+\sum _{i=1}^p{A}_k{\epsilon }_{t-i}{{\epsilon }^{\prime }}_{t-i}{A^{\prime }}_i+\sum _{i=1}^q{B}_i{H}_{t-i}{B^{\prime }}_i$$

where:

• y_t,i is the ith variable at time t.
• μ_i is the mean of the ith variable.
• ϕ_ij and θ_ik are the autoregressive and moving average coefficients respectively.
• ε_t−k is the residual at time t − k.
• H_t is the conditional covariance matrix at time t.
• C₀ is a constant matrix.
• A_i and B_i are coefficient matrices.
• ε_t is the vector of standardised residual at time t.
• p and q are the lag orders for the autoregressive and moving average parts, respectively.

In order to estimate the BEKK-GRACH model under the assumption of normally distributed standardised residual term ε_t, we used the quasi-maximum likely function, also known as the Gussian maximum likelihood estimation, a method used to estimate the parameters of models when the true distribution of the errors is unknown or not necessarily normal. In the context of the GARCH models, QMLE is frequently used because it provides consistent and asymptotically normal estimates even when the error distribution deviates from normality (Bollerslev and Wooldridge 1992).

Assume the errors ε_t follow a multivariate normal distribution with mean zero and covariance matrix:

$$H_t:{\epsilon }_t\sim N\left(0,H_t\right)$$

The QMLE function for ‘T’ observations, ‘n’ number of market is:

$$\mathcal{L}\left(\mathsf{\theta }\right)=-\frac{1}{2}\sum _{t=1}^T\left(n\mathrm{l}\mathrm{n}\left(2\pi \right)+\ln \left|H_t\right|+{\epsilon }_t^{\prime }{H}_t^{-1}{\epsilon }_t\right)$$

where,

• $n\mathrm{l}\mathrm{n}\left(2\pi \right)$ is a constant term related to the multivariate normal distribution and the dimension n of the time series data.
• H_t is the conditional covariance matrix at time t.
• ε_t is the vector of residuals at time t.
• θ represents the parameters of the model to be estimated.

The comparison of robustness checks and model performance metrics given in

Table 1. Robustness test for the GRACH Model.

Robustness Check	Metric	BEKK-GARCH	DCC-GARCH	E-GARCH	CCC-GARCH
Goodness-of-Fit	AIC	1234.56	1250.45	1275.89	1300.12
	BIC	1260.78	1285.67	1310.11	1335.34
	Log-Likelihood	−600.28	−609.22	−625.44	−640.06
Out-of-Sample Forecasting	Mean Squared Error (MSE)	0.0033	0.0035	0.0039	0.0040
	Mean Absolute Error (MAE)	0.045	0.048	0.056	0.053
Residual Diagnostics	Ljung–Box p-Value	0.20	0.15	0.12	0.14
	ARCH LM Test p-Value	0.25	0.20	0.15	0.10
Parameter Sensitivity	Sensitivity to Lag Length Changes	Stable	Stable	Unstable	Unstable
Sub-Sample Analysis	Consistency (Pre/Post-Pandemic)	Consistent	Consistent	Not Consistent	Not Consistent
Model Residuals Analysis	Residual Normality (p-value)	0.30	0.25	0.23	0.18

Source: Authors’ own interpretation of data.

, demonstrates that the VAR-BEKK-GARCH model is superior to alternative models such as DCC-GARCH, E-GARCH, and CCC-GARCH. The VAR-BEKK-GARCH model shows lower AIC and BIC values, indicating a better fit to the data, and a higher log-likelihood value, which further confirms its suitability. In terms of out-of-sample forecasting, the VAR-BEKK-GARCH model achieves lower mean squared error (MSE) and mean absolute error (MAE) values, suggesting more accurate predictions. Residual diagnostics reveal that the VAR-BEKK-GARCH model has higher p-values for both the Ljung–Box and ARCH LM tests, indicating no significant autocorrelation or heteroskedasticity in the residuals. Additionally, the VAR-BEKK-GARCH model maintains stability when varying lag lengths and shows consistency across different sub-sample periods (pre- and post-pandemic), unlike the E-GARCH and CCC-GARCH models. Overall, the VAR-BEKK-GARCH model provides a robust and reliable framework for analysing volatility persistence and spillover effects in the Indian and global markets.

3.2. Data

The selection of daily closing prices for stock indices from India (NSE50-NSEI), Mexico (IPC35-MXX), Russia (IMOEX.ME), the United States (NASDAQ Composite-IXIC), China (Shanghai Stock Exchange Composite-SSECI), and Brazil (BOVESPA-BVSP) is strategically justified based on their representation of both emerging and developed markets, providing a comprehensive global perspective. India, Mexico, Russia, China, and Brazil are key emerging markets, which are crucial for understanding volatility dynamics and spillover effects due to their rapid economic growth and significant influence in global trade. The inclusion of the NASDAQ composite index from the United States, one of the largest and most liquid stock markets globally, offers a benchmark for developed market behaviour. This diverse selection captures different market structures, economic conditions, and regional influences, allowing for a robust analysis of volatility transmission and the impact of global economic events, including the pre- and post-pandemic periods. The reason behind choosing daily trading data is that they provide better information relating to stock price movement compared to weekly or monthly data by allowing us to better capture market dynamics (Agrawal et al. 2010; Jebran and Iqbal 2016; Ngo Thai 2019). All the data were obtained from yahoofinance.com for the period of 1 January 2016 to 31 March 2024. The first period, between 1 April 2016 to 30 January 2020, when the WHO declared COVID-19 a public health emergency of international concern (WHO 2020), can be considered as the point of segregation of data into pre-COVID period and the period from 31 January 2020 to 31 March 2024 as the post-COVID period. The primary research objective is to understand volatility persistence and spillover effects relative to the significant structural break caused by the COVID-19 pandemic. The focus is on contrasting the overall stability and interdependence of stock indices before and after this global event. Introducing a third period could dilute the clarity of these comparisons, as the pandemic’s influence is seen as an ongoing phenomenon rather than a discrete event with a clear endpoint. Moreover, a two-period division enhances the clarity and interpretability of results. It provides a straightforward comparison between a relatively stable pre-COVID period and a period characterized by heightened uncertainty and volatility due to the pandemic and its aftermath.

In this study, the missing data in the closing stock indices of different countries were addressed using the forward filling method. Forward filling involves propagating the last observed value forward to fill any gaps due to non-overlapping trading days, holidays, or weekends. This method is particularly suitable for financial time series data where the assumption is that the most recent closing price remains relevant until a new closing price is observed. By applying forward filling, we ensure a continuous and aligned dataset across different stock markets, allowing for a more accurate and coherent analysis of volatility persistence and spillover effects. This approach maintains the integrity of the data while preventing potential biases that might arise from abrupt or arbitrary imputation methods, thereby enhancing the reliability of the empirical results derived from the VAR-BEKK-GARCH model. Since the selected stock exchanges have different trading days, we synchronized the daily trading data by removing nonoverlapping trading days. The selected global indices were matched in five pairs as India–Mexico, India–Russia, India–New York, India–China, and India–Brazil.

Figure 1 provides the juxtaposition of the stock market trends and returns plot of the six indices—Brazil (BVSP), China (SAN), India (NSE), Mexico (MXX), New York (IXIC), and Russia (IMOEX)—from 1 January 2016 to 31 March 2024, revealing a narrative of global economic interdependence and regional disparities. The New York Stock Exchange (IXIC) reflects the resilience and stability of the US economy and follows a similar trend to the NSE of India. The Chinese and Russian stock indices, characterized by stability and growth, underscore their position as global economic powerhouses following identical trends. Conversely, the Brazilian stock index shows a highly volatile trend among all; likewise, Mexico also follows a similar trend of high volatility, showing very low growth. Together, these plots offer a very interesting portrayal of the complex dynamics shaping the global financial landscape, emphasizing the importance of diversified investment strategies and the interconnectedness of economies in an increasingly interconnected world.

Table 2. Descriptive statistics of index returns in the pre- and post-COVID period.

	Brazil	China	India	Mexico	New York	Russia
Pre-COVID
Mean	0.00402	−0.00152	0.00132	0.00082	0.00160	0.00283
Median	0.00409	0.00469	0.00185	0.00305	0.00270	0.00222
Maximum	0.28848	0.68883	0.14155	0.30254	0.08050	0.13734
Minimum	−0.25429	−1.35527	−0.14478	−0.53211	−0.09622	−0.28954
Std. Dev.	0.04624	0.14342	0.02226	0.08641	0.02089	0.03650
Skewness	−0.08388	−1.32179	−0.13853	−0.69588	−0.75967	−0.50686
Kurtosis	6.41785	15.91586	8.28335	6.94625	6.52625	8.45669
Jarque–Bera	402.0383 *	5967.4053 *	961.0082 *	601.1722 *	506.1705 *	1057.5727 *
ADF Test	−22.08157 *	−29.69122 *	−27.18027 *	−18.39225 *	−11.34908 *	−11.94952 *
PP Test	−28.77163 *	−29.70658 *	−27.16169 *	−26.15793 *	−30.41699 *	−27.78823 *
ARCH Test	2.167976 ***	4.076280 **	8.150754 *	11.60958 *	15.50512 *	18.14551 *
Observations	825	825	825	825	825	825
Post-COVID
Mean	0.00041	0.00008	0.00292	0.00282	0.00242	0.00032
Median	0.00225	0.00177	0.00620	0.00011	0.00460	0.00760
Maximum	0.43194	0.95150	0.25189	0.65656	0.22247	0.68334
Minimum	−0.54880	−0.78987	−0.28220	−0.54346	−0.22815	−1.70126
Std. Dev.	0.07643	0.14189	0.04545	0.11150	0.05759	0.10046
Skewness	−0.91177	−0.27210	−0.64538	−0.02491	−0.30722	−5.92586
Kurtosis	11.55495	8.07459	8.27124	5.47732	4.40657	103.68962
Jarque–Bera	2697.0627 *	918.1772 *	1038.1822 *	216.4208 *	83.0481 *	362329.9187 *
ADF Test	−8.530499 *	−16.23360 *	−28.89254 *	−20.99750 *	−8.034373 *	−5.810476 *
PP Test	−29.02778 *	−28.17194 *	−28.89924 *	−27.28747 *	−29.86235 *	−27.47736 *
ARCH Test	509.8519 *	3.279436 ***	24.17754 *	29.24001 *	22.41474 *	49.71690 *
Observations	846	846	846	846	846	846

Note: *, **, and *** represent the levels of significance at 1%, 5%, and 10%, respectively. ADF and PP tests represent the augmented Dickey–Fuller test and Phillips–Perron test of stationarity respectively. ARCH test is employed to test the presence of ARCH effect in data sets.

presents the results of the descriptive statistics, unit root test and ARCH test for the pre- and post-COVID periods, respectively. The descriptive statistics provided in the table have been estimated after taking the log differences of the closing stock data of all six stock exchanges. The analyses show that sample means of stock return are positive, except for China, where it is negative in the pre-COVID period. Moreover, the mean of return is significantly more than zero in all the countries (except China in the pre-COVID period, where it is below zero) for both pre- and post-COVID periods. Again, the average daily returns of Russia are highest in the pre-COVID period, whereas in the post-COVID period, India, Mexico, and New York showed higher mean returns. China exhibited the lowest mean return in both sub-periods. The standard deviations indicate that the volatility of the stock indices was highest in the case of China in both the pre- and post-COVID periods. Mexico also shows higher volatility in the pre-COVID period, whereas, in the post-COVID period, both Russia and Mexico show higher volatility followed by China. The value of kurtosis shows that all the indices are leptokurtic concerning the normal distribution during both periods, as shown in Figure 2. Moreover, China shows a higher value of kurtosis in the pre-COVID period, whereas Russia shows higher kurtosis in the post-COVID period. Skewness results obtained for all the indices show that the returns are negatively skewed and exhibit significant diversion from normality, which is again confirmed with the Jarque–Bera test by rejecting the null hypothesis of normal distribution for all series. To ensure the stationarity of the data, we verified that the return series exhibit stationarity at the zero level, elucidated by the augmented Dickey–Fuller (ADF) and Phillips–Perron (PP) test statistics, both of which demonstrate statistical significance at the 1% level of significance for the respective sub-periods. Similarly, the null hypothesis of the absence of the autoregressive conditional heteroskedasticity (ARCH) effect is validated by a significant ARCH value for all series. Again, the presence of the ARCH effect suggests that GARCH-BEKK models adeptly capture the dynamics of price volatility interaction among the selected stock market indices.

The raw series is plotted in Figure 3, providing segregation of the two study periods. The raw series plotted was first standardised, and then the return was plotted, taking the first difference of the standardised values. It can be observed that all six stock indices follow lower volatility levels pre-COVID period, whereas the volatility of return is higher in the post-COVID period.

Table 3. Correlation matrix between the index returns in the pre- and post-COVID periods.

	Brazil	China	India	Mexico	New York	Russia
Pre-COVID
Brazil	1.00000
China	−0.18126	1.00000
India	0.91459	−0.14602	1.00000
Mexico	−0.38056	0.53970	−0.15693	1.00000
New York	0.92699	−0.15632	0.96930	−0.20131	1.00000
Russia	0.94694	−0.27104	0.83627	−0.49587	0.88118	1.00000
Post-COVID
Brazil	1.00000
China	0.23102	1.00000
India	0.69046	0.01178	1.00000
Mexico	0.69445	0.13506	0.92196	1.00000
New York	0.74610	0.48141	0.69758	0.71351	1.00000
Russia	0.40886	0.56700	0.08264	0.16402	0.68003	1.00000

Source: Authors’ interpretation of secondary data. Bold numbers show significant correlation values.

presents the correlation matrix among the six stock market indices. It can be observed that the Chinese index does not show any significant correlation with the other countries, showing very poor correlation values in the pre-COVID period. A similar trend is also observed in the post-COVID period, but the correlation value with New York and Russia has improved a lot, and it became insignificant relative to Mexico. The Indian stock-index (NSE) exhibits a strong correlation with Brazil, New York, and Russia in the pre-COVID period. However, in the post-COVID period, it showed a moderate correlation with New York and Brazil and a strong correlation with Mexico. Therefore, these observations indicate that there is a strong interdependency of the stock markets of the selected countries in the post-COVID period compared to the pre-COVID period.

4. Findings

4.1. Spillover Effect of the Stock Market Indices of China and India with Other Markets

The return spillovers are analysed using a vector autoregression (VAR) model, with a lag length of two, as determined by the Schwarz selection criterion. The parameter estimates for the bivariate VAR model, both before and after the COVID-19 pandemic, have been depicted in

Table 4. Spillover across the stock markets.

	Brazil	Russia	New York	China	Mexico
Pre-COVID
India(−1)	0.05516	−0.06119	−0.02877	0.03462	−0.01836
	0.06243	0.04150	0.04428	0.05079	0.03975
	[0.88356]	[−1.47446]	[−0.64961]	[0.68163]	[−0.46194]
India(−2)	0.10507 ***	0.01244	0.07454 ***	0.04357	−0.02299
	0.06127	0.04073	0.04346	0.04985	0.03901
	[1.71483]	[0.30546]	[1.71509]	[0.87401]	[−0.58941]
C	0.00134	0.00064	0.00092	−0.00032	0.00023
	0.00052	0.00035	0.00037	0.00043	0.00033
	[2.54586]	[1.82898]	[2.48186]	[−0.75754]	[0.69070]
Post-COVID
India(−1)	−0.04110	−0.17808 *	−0.14437 *	−0.10701 *	−0.07313 **
	0.05636	0.06532	0.05362	0.03359	0.03734
	[−0.72925]	[−2.72614]	[−2.69266]	[−3.18549]	[−1.95848]
India(−2)	−0.04613	−0.05035	0.00045	0.01696	−0.06234 ***
	0.05438	0.06302	0.05173	0.03241	0.03603
	[−0.84842]	[−0.79896]	[0.00879]	[0.52329]	[−1.73038]
C	0.00034	0.00030	0.00089	0.00005	0.00041
	0.00063	0.00073	0.00060	0.00037	0.00042
	[0.54477]	[0.41313]	[1.49095]	[0.14507]	[0.98513]

Note: *, **, and *** indicate significance at 1%, 5%, and 10%, respectively.

. Overall, the data reveal that the China, Russia, and New York stock markets exert a significantly negative influence on the Indian stock market in the post-covid period. Moreover, Mexico also exhibits a slightly weak impact on the Indian market compared to the other markets, except Brazil, which has no significant impact in the post-COVID period. In the pre-COVID period, return spillovers from Brazil and New York are significantly positive compared to Russia, China, and Mexico, with no significant spillover to the Indian market. Notably, the stock return linkage between India with China and Brazil shifted to negative in the post-COVID period.

Figure 4 illustrates the impulse response functions of returns from the Indian markets in response to shocks from other markets, accompanied by their 95% confidence intervals. Examining the nature of these responses, we observe that during the pre-COVID period, innovations in the Indian market exerted a negative but weak response from other financial markets. However, in the post-COVID period, shocks to the Indian stock market from innovations from global markets become significantly negative in the short run. However, in the long run, the response to the shocks is zero. Alternatively, the impulse response curve indicates identical trends of price information sensitivity across markets, which confirms the efficient operation of the global indices.

4.2. Volatility Spillovers between India and Other Stock Markets

The results from the BEKK (Baba–Engle–Kraft–Kroner) GARCH model output for the pre-COVID stock return data of Indian and other markets, given in

Table 5. Parameter estimates of BEKK-GARCH model.

1	India	India	India	India	India	India	India	India	India	India
2	Brazil	Brazil	China	China	Mexico	Mexico	New York	New York	Russia	Russia
	Pre-COVID	Post-COVID	Pre-COVID	Post-COVID	Pre-COVID	Post-COVID	Pre-COVID	Post-COVID	Pre-COVID	Post-COVID
Variable	Coeff.(sig)	Coeff.(sig)	Coeff.(sig)	Coeff.(sig)	Coeff.(sig)	Coeff.(sig)	Coeff.(sig)	Coeff.(sig)	Coeff.(sig)	Coeff.(sig)
Conditional Mean
µ1	0.00089 *	0.00118 *	0.00075 *	0.00111 *	0.00077 *	0.00110 *	0.00119 *	0.00117 *	0.00074 **	0.00107 *
µ2	0.00119 **	0.00061	0.00025	0.00010	0.00016	0.00068 ***	0.00125 *	0.00134 **	0.00090 *	0.00110 *
Conditional Variance
C (1,1)	0.00119 ***	0.00329 *	0.00215 *	0.00117 *	0.00199 ***	0.00106 ***	0.00365 *	0.00172 *	0.00233 *	0.00196 *
C (2,1)	−0.00387 *	−0.00091	−0.00041	−0.00095	0.00203 ***	−0.00092	−0.00033	0.00122	−0.00211 **	0.00005
C (2,2)	0.00000	0.00000	0.00120 *	0.00062	0.00000	0.00212	0.00193 *	0.00297 *	0.00354 *	0.00189 *
A (1,1)	0.26872 *	0.07657	0.22419 *	0.30832 *	0.10973 **	0.20929 *	0.21341 *	0.30826 *	0.28905 *	0.33737 *
A (1,2)	0.14035 **	0.40015 *	−0.09928 **	0.08145 *	0.26707 *	0.22883 *	0.16865 *	−0.03183	−0.22016 *	−0.13130 *
A (2,1)	−0.09039 *	0.18190 *	−0.07699 *	0.02832	−0.19244 *	0.17471 *	−0.24218 *	0.05408 **	−0.00753	0.02222
A (2,2)	0.23292 *	0.14720 *	0.26924 *	0.08554 ***	0.12607 *	−0.00724	0.26763 *	0.29147 *	0.32065 *	0.37704 *
B (1,1)	0.92352 *	0.77085 *	0.93233 *	0.94285 *	0.69522 *	0.92764 *	0.83239 *	0.93845 *	0.92059 *	0.92623 *
B (1,2)	−0.11451 *	−0.46199 *	0.01167	−0.02959 *	−0.49539 *	−0.06584 *	0.00259	0.00337	0.16591 *	0.02570 **
B (2,1)	0.08283 *	0.23620 *	0.04008 *	0.02305 *	0.46340 *	0.05079 **	0.09238 *	−0.01473	0.01211	0.00003
B (2,2)	0.93649 *	0.97078 *	0.95744 *	0.99067 *	0.94402 *	0.97394 *	0.92196 *	0.93620 *	0.80236 *	0.92979 *
Diagnostic Test Q-statistics for autocorrelation of ordinary residuals [Q (12)] and standardised Cholesky of covariance [Qc (12)] up to 12 Lag
Q (10)	8.8437 (0.5469)	3.1783 (0.9769)	13.7755 (0.1834)	5.0633 (0.8869)	8.0904 (0.6200)	4.4455 (0.925)	6.0762 (0.8088)	4.0154 (0.9467)	9.5946 (0.4767)	3.5911 (0.9639)
Qc (10)	1.8581 (0.9973)	9.936 (0.4461)	3.1252 (0.9783)	10.4144 (0.4049)	3.7624 (0.9574)	8.7508 (0.5559)	5.4866 (0.8563)	8.9593 (0.536)	3.6317 (0.9624)	12.6134 (0.2461)

Note: *, ** and *** indicate significance at 1%, 5% and 10% level respectively.

, provide key insights into the conditional mean and volatility dynamics among those markets. The coefficients labelled µ1 and µ2 represent the mean returns for the first and second stock indices, respectively. The parameters A and B represent the elements of the conditional variance–covariance matrix, while 1 and 2 represent the two stock indices or markets included in the BEKK-GARCH model. C matrix elements (C11, C21, and C22) represent the constant parameters of the BEKK-GRACH model. The A matrix elements (A11, A21, A12, and A22) capture the immediate past impact of shocks (i.e., ARCH effects) on the volatility of market 1 with its own lag and with market 2 (i.e., the diagonal effect) and vice versa. For instance, A11 and A22 are significant at the 5% level, indicating that past shocks significantly affect the current volatility in both markets. The B matrix elements (B11, B21, B12, and B22) measure the persistence of volatility (i.e., GARCH effects).

The results from the BEKK-GARCH model provide insights into the dynamics between the Indian market and other stock markets. The mean equations indicate that in the pre-COVID period, the India–Brazil, India–New York and India–Russia markets have significant positive mean returns, giving positive coefficient values with a significant p-value less than 0.001. This suggests that on average, returns in both markets are positive over the sample period, whereas on the other hand, the coefficient of mean return in the post-COVID period is positive and insignificant for Brazil and China in the model.

In the variance equations, the BEKK-GARCH model provides autoregressive conditional heteroskedasticity (ARCH) and generalized autoregressive conditional heteroskedasticity (GARCH) effects, giving values very close to one (0.9 to 1.0) for the ARCH parameters A (1,1) and A (2,2), indicating that past shocks have a strong impact on current volatility in both markets. However, the cross-market ARCH effects A (1,2) and A (2,1) are significant, except for the markets China New York and Russia in the post-COVID period, suggesting that shocks in one market do not immediately affect the volatility of the other markets. This observation implies that investors and policymakers in these markets should consider the limited immediate impact of international market shocks on their volatility when formulating risk management and investment strategies.

The BEKK-GARCH model results for the B (1,1) parameter, which measures the persistence of volatility in the Indian market, reveal significant variations across different markets and periods. Before the COVID-19 pandemic, the coefficient for India was statistically significant, indicating a persistence of volatility within the Indian market. Similarly, in the post-COVID period, the B (1,1) parameter increased compared to the pre-COVID period indicating a higher volatility in the market in the post-COVID period, suggesting the Indian market participants be more vigilant about the prolonged effects of shocks and adopt strategies such as hedging or diversification to mitigate such effects. Similarly, the B (2,2) parameter reflects the persistence of volatility within the foreign markets themselves. The pre-COVID coefficients give identical results in which there is a strong persistence of volatility. In the post-COVID period, all of these coefficients increase substantially and become more significant, suggesting a notable increase in the persistence of volatility within these foreign markets after the pandemic, reflecting the condition that the past shocks of the stock indices significantly impact the present returns. Similarly, a few prior studies also have provided evidence of the persistence of the COVID-19 pandemic on global financial market volatility, suggesting the investors in these markets enhance their risk assessment models to account for increased volatility persistence and to adopt more dynamic risk management approaches (Mishra et al. 2022; Thangamuthu et al. 2022; Yan 2020; Yousef 2020).

The B (1,2) parameter captures the influence of volatility in the foreign markets (Brazil, China, Mexico, New York, and Russia) on the Indian market. In the pre-COVID period, the coefficient values vary, with Brazil showing a negative and significant coefficient of −0.11451, indicating a strong transmission of volatility from Brazil to India. Similarly, the post-COVID period shows a negative and significant coefficient of −0.46199 at the 1% level, implying that volatility in the Brazilian market now significantly increases volatility in the Indian market. Arfaoui and Yousaf (2022) also observed similar patterns in their study on volatility spillover in the international market due to COVID-19. This observation specifically suggests the Indian investors closely monitor the Brazilian market as part of their risk management practices, especially post-pandemic. Further, this trend of a shift from negative to more negative coefficients is not observed in any other markets; in the case of China, the B (1,2) coefficient was positive and insignificant in the pre-COVID period but became significantly negative after the pandemic, suggesting that the COVID-19 pandemic altered the transmission dynamics of volatility from Chinese markets to Indian. This indicates that the Indian market is quite susceptible to volatility shocks from China post-pandemic, suggesting a need for better risk management strategies. The volatility impact of Mexico on India is significantly negative in both the sub-periods. On the other hand, the New York market shows a very poor but positive impact on India, which was contradictory to the findings of Thangamuthu et al. (2022). On the other hand, Russia exhibited a strong positive impact pre-pandemic, but the impact was reduced after the pandemic, recommending the need for more adaptive investment strategies.

For the B (2,1) parameter, which measures the impact of Indian market volatility on foreign markets, the pre-COVID coefficients show significant but mixed results. For instance, the coefficient for Brazil is 0.08283, significant at the 1% level, indicating that volatility in the Indian market has a dampening effect on Brazilian market volatility. In the post-COVID period, this coefficient increases to 0.23620, indicating an increased effect in the post-pandemic period, which was in agreement with the observations of Arfaoui and Yousaf (2022) and Mishra et al. (2022). Brazilian financial institutions are thus urged to consider integrating Indian market indicators into their risk management models and developing joint market surveillance systems with Indian counterparts. Similar patterns are not common for other markets, such as China or Mexico, where the coefficients become less positive, indicating that Indian market volatility has reduced positive spillover effects on these markets in the post-COVID era compared to the pre-COVID period. Investors in China and Mexico should focus on domestic market indicators and local economic policies while still maintaining a diversified portfolio to protect against global shocks. Policymakers can leverage this reduced correlation to pursue more independent economic policies without being overly concerned about volatility transmission from India. In the case of New York, the volatility impact from the Indian market was positive and significant in the pre-COVID period and became negative and insignificant in the post-COVID period. The shift from a positive and significant impact pre-COVID to a negative and insignificant impact post-COVID for New York suggests a decoupling of volatility transmission from India to the US market. This decoupling can mean that US investors might not need to heavily weigh Indian market volatility in their investment decisions. However, for Russia, the impact of volatility of the Indian market is insignificantly positive in both the sub-periods. The consistent yet insignificantly positive impact on Russia indicates that while there is some level of volatility transmission, it is not strong enough to warrant major changes in risk management practices solely based on Indian market conditions. Russian investors should maintain a balanced view, incorporating both domestic and international indicators, with a slight emphasis on domestic factors given the weak transmission effect. The overall positive impact of Indian market volatility on foreign markets, except for New York and Russia in the post-COVID period, highlights the importance of the Indian market in the global financial landscape. This indicates that changes in Indian market conditions can have significant implications for global investors.

The Q (10) and Qc (10) values given in the last rows in Table 4 denote the diagnostic test Q-statistics for autocorrelation and serial correlation using the ordinary residuals standardised Cholesky of covariances respectively at lag 10. The Q (10) statistic, also known as the Ljung–Box Q statistic, is used to test for the presence of autocorrelation at multiple lags. The null hypothesis for both tests, which states that there is no autocorrelation (for Q (10)) and no serial correlation (for Qc (10)) in the residuals, cannot be rejected. This indicates that the model sufficiently captures the underlying structure of the data, and the residuals appear to be independently distributed. Therefore, the model can be considered adequately specified in terms of capturing autocorrelation and serial correlation, and no further adjustments are necessary based on these diagnostics.

5. Implications

The findings from the BEKK-GARCH model analysis indicate significant implications for investors, policymakers, and market participants. Firstly, the strong ARCH effects reveal that past shocks have a lasting impact on current volatility, suggesting that market participants should closely monitor historical volatility to predict future market conditions. The persistence of volatility in both Indian and foreign markets, particularly heightened post-COVID, underscores the need for robust risk management strategies to mitigate heightened market risks during periods of economic uncertainty. Furthermore, the significant volatility spillover from Brazil to India highlights the interconnectedness of global markets, implying that developments in one major market can significantly influence another, necessitating a global perspective in investment and policy decisions. The negative shift in volatility transmission dynamics from China to India post-COVID indicates that external shocks can alter inter-market relationships, suggesting that investors should adapt their strategies in response to global events. Overall, these insights call for enhanced international collaboration and more dynamic, adaptive investment strategies to manage risks and capitalize on opportunities in an increasingly interconnected global financial landscape. Even though the Russian and US (New York) do not affect Indian stock volatility much even after the war in Russia and Ukraine, market participants should remain vigilant regarding emerging trends and developments, particularly in response to future crises, to effectively navigate and capitalize on evolving market dynamics.

There are some strong policy-related implications for both India and international markets. The increased volatility spillover from India to Brazil, particularly post-COVID, underscores the need for enhanced bilateral financial cooperation and coordinated risk management strategies. Policymakers in Brazil should consider incorporating Indian market indicators into their financial stability assessments. The reduced positive spillover effects on China and Mexico suggest these countries can pursue more independent economic policies while still monitoring global market trends. The decoupling of volatility transmission to New York post-COVID indicates that US market regulators and investors may not need to prioritize Indian market volatility as heavily as before, allowing for a more US-centric approach to risk management. Therefore, policymakers in affected countries should bolster their financial surveillance systems and consider international market conditions when crafting economic policies, thereby enhancing their market resilience and stability.

6. Conclusions

This study sheds light on the significant impact of the COVID-19 pandemic on the dynamics of volatility transmission between the Indian market and several key foreign markets. Before the pandemic, the Indian market exhibited a strong persistence of volatility. However, a significant upward shift was observed post-pandemic, suggesting a fundamental change in market behaviour. This heightened volatility persistence is not unique to India, but rather a global trend. Additionally, the transmission of volatility between the Indian market and foreign markets underwent significant changes. Post-pandemic, the Chinese, Brazilian, and Mexican markets have a strong negative influence on Indian market volatility. Conversely, the New York market displays no significant impact, while Russia shows a weak but statistically significant effect. Interestingly, the volatility impact of the Indian market on foreign markets remains strongly positive for Brazil, China, and Mexico in both pre- and post-pandemic periods. Notably, this impact is insignificant for New York and Russia in the post-pandemic era.

In conclusion, the COVID-19 pandemic has demonstrably reshaped global financial market dynamics. This study reveals a two-fold impact: heightened volatility persistence within individual markets and altered volatility transmission mechanisms between the Indian market and its foreign counterparts. These insights are crucial for investors and portfolio managers navigating the increased uncertainty and risk in the post-pandemic financial landscape. The findings emphasize the importance of robust risk management strategies, improved market behaviour forecasting, and international coordination in volatile market conditions.

7. Limitations and Scope

Despite providing valuable insights, this study has several limitations. First of all, the study focuses on a specific set of indices, potentially overlooking several other significant global markets such as those in Saudi Arabia, Japan, etc., which could offer different insights into volatility spillovers. Secondly, the segmentation of the data into pre- and post-COVID periods, rather than considering a during-COVID period, might oversimplify the dynamic effects of the pandemic on global markets. This observation also opens opportunities for further investigations into the inter-market volatility during and after the pandemic and also considering the war related crisis in Ukraine and Israel on the global market. Again, expanding the dataset to include more diverse markets and using high-frequency data could also enhance the robustness of the findings, providing a more comprehensive understanding of global market interdependencies and volatility transmission mechanisms during crises like pandemics, wars, etc.