Green stocks, crypto asset, crude oil and COVID19 pandemic: Application of rolling window multiple correlation.

The COVID-19 pandemic disrupted almost all spares of global social, psychological, and economic life. The emergence of various variants and corresponding variations in daily infection asymmetrically influenced economic indicators. This study extends the existing literature by exploring the hedging potential of crude oil, carbon efficiency index of green firms, and bitcoin during this pandemic. This objective is realized by employing the recently advanced rolling window multiple correlation of Polanco-Martínez (2020). This approach is based on the new p-value corrected method, which has advantages over other correlation methods. The sample observations are based on daily data from 1/22/2020 to 12/20/2021. In the bivariate case, we find a significant positive correlation between COVID-19 and CEI, while a negative impact is observed between COVID-19 and WTI. Similarly, we observe a significant and nonlinear association between COVID-19 and BTC. However, our findings show positive and significant correlations among variables in the multivariate case. The overall findings show that CEI and BTC can be safe havens for investors during this worse pandemic. The study's robust findings can be used to derive important policy implications worldwide during the COVID-19 pandemic.

Introduction

The coronavirus (COVID-19) has emerged as one of the most fatal pandemics in history in terms of human, social, and economic cost. The emergence of various variants of SARS-COV-2 results in continuous fluctuations in daily cases of new infections, see Fig. 1 and make it the gravest challenge to the world. It poses unprecedented challenges to public health, economy, and the food system. The pandemic induces economic lockdown, and travel restrictions severely impacted manufacturing activities, shipments, airlines (Fareed et al., 2022), travel and tourism (Yan et al., 2022), banking, insurance, stocks and cryptocurrency markets (Chen et al., 2022; Iqbal et al., 2021). As a result, global oil prices fell to multi-decade lows, and producers rushed to find space on land and sea to store oil rather than selling at a loss. However, Fig. 1 reveals that oil prices improved after initial plunge to negative territory in mid-April 2020. Furthermore, these economic uncertainties have considerably reduced the stock market returns (Sharma et al., 2019, 2021b). Therefore, investors are searching for alternative assets for hedging purposes. We explored the hedging capacity of green firms’ stocks, bitcoin, and oil market to realise this objective during the current pandemic. Several strains of arguments motivate this study.

Fig. 1

Time trend of raw sereis

Note: CEI, WTI, BTC, COVID19 denote carbon efficiency index, oil price, bitcoin, and worldwide COVID19 daily new cases, respectively.

First, the current pandemic has created economic chaos due to sudden and massive disruption in global production activities along with corresponding shocks in market returns in both developed and developing countries (Narayan et al., 2020). Many stock markets crashed across developed and developing during the initial phase due to chaotic panic and lockdowns. Mazur et al. (2021) noted that in March 2020 Dow Jones Industrial Average plunged to 6400 points in barely four trading days. Moreover, Topcu and Gulal (2020) have noted that stock markets of emerging countries received the relatively highest impact as compared with developed economies. Nevertheless, this negative impact on the stock return may not necessarily be equally detrimental to all firms and industries. Interestingly, Fig. 1 reveals that the return of the carbon efficiency index shows an increasing trend after the initial plunge. This positive trend may be due to increasing environmental concerns during the current pandemic situation. Several studies associated the emergence of the COVID-19 virus with the increase in global warming (Sasikumar et al., 2020; Sharma et al., 2021a). Moreover, Koçak et al. (2021) have also reported a significant positive impact of the COVID outbreak on green companies from 2nd January to October 5, 2020.

Second, the high volatility in oil and other commodity prices across the globe forced investors to diversify their portfolios toward relative safe-haven assets. According to some recent studies, bitcoin has the potential to be an important hedge and diversifier asset during the current chaotic situation (Urquhart and Zhang, 2019; Salisu et al., 2020; Vukovic et al., 2021). However, it has been neglected in the literature as a hedge asset due to its short-term volatilities (Sharma et al., 2019) Therefore, among the available literature, few studies have concluded good hedging behaviour of bitcoin as it improves the risk-return profile of portfolio investment (Bouri et al., 2017; Chkili et al., 2021; Su et al., 2020). At the same time, some studies are sceptical regarding the role of cryptocurrencies as an effective hedge, especially in the prolonged chaotic situation by concluding the asymmetric influence of COVID-19 shocks on the return of cryptocurrencies (Iqbal et al., 2021). Furthermore, Fig. 1 reveals that the price of Bitcoin has been volatile during the current pandemic, but the overall trend is positive for the sample period. Therefore, this modelled bitcoin to explore its hedging behaviour during this pandemic.

Therefore, this study contributes to existing literature in many folds. First, it is a novel contribution that examines the hedging behavior of bitcoin, carbon efficient companies and the oil market to examine their hedging capacity during pandemic. Second, this study employed a recent advanced rolling window multiple correlation approach that can address the correlation throughout the sample period and provide a complete picture of fluctuations. Furthermore, this novel approach provides more efficient estimates by addressing the asymmetric nature of relationships. Third, the findings of this study revealed that although selected assets are showing good hedging potential, but hedging capacity of the carbon efficiency index and bitcoin is far better and stable during the peak of pandemics. Therefore, investors can consider these assets for their portfolio investment during current pandemic. Reviewed studies have generalized their proposition concerning the impact of the COVID-19 pandemic on oil prices, stock market, gold prices, and bitcoins based on estimation techniques that measure aggregated behavior during the sample period. Since the COVID-19 cases, along with the behavior of stock markets and other assets, fluctuate on a daily basis. Therefore, the aggregated response can lead to biased generalization.

The rest of this study is organized as follows; section 2 reviews some selected literature regarding the impact of COVID-19 on various assets and stock markets; section 3 presents a methodological framework; section 4 discusses estimated results, whereas section 5 concludes the study with policy recommendations.

Literature review

This historical decline in oil prices and economic lockdown associated with the COVID-19 pandemic have a multifaceted impact on economic aggregates. This section critically reviews selected literature to highlight the impact of the COVID-19 pandemic on the oil market, carbon efficiency index, and bitcoin prices.

The crude oil prices witnessed historical fluctuations during the COVID-19 pandemic due to considerable oil demand distortion and geopolitical oil-price war between OPEC and Russian Federation. Therefore, Le et al. (2021) explored the factors responsible for historical oil price fluctuation during the Covid-19 pandemics by employing the ARDL bounds testing approach with a cross-sectional break on daily series from 17 January to September 14, 2020. The findings reveal that increase in Covid-19 pandemic cases, US economic policy uncertainty, and expected stock market volatility continue to fall WTI crude oil price, whereas the fall in the global stock markets appears to reduce the fall significantly. Furthermore, Russia-Saudi Arabian oil price war and speculation on oil futures have a critical impact on the collapse of the oil markets. In contrast, Dutta et al. (2020) explore the time-varying correlation between gold, bitcoin, and oil market volatility to see whether gold can be a safe-haven asset in the case of international oil market volatility during the COVID-19 period. The estimated result of the DCC-GARCH model reveals that the gold market is a good hedge against global crude oil market fluctuations, whereas bitcoins act only as diversifiers.

Furthermore, Mensi et al. (2020) examine the effect of the COVID-19 pandemic on the multifractality of oil and gold prices based on their upward and downward trends by employing the Asymmetric Multifractal Detrended Fluctuation Analysis (A-MF-DFA) approach on 15-min interval intraday data. The findings revealed strong evidence concerning an increase in asymmetric multifractality scale. Furthermore, multifractality is higher in a downward trend for Brent oil and an upward trend for gold. Moreover, this asymmetry has been accentuated during the COVID-19 pandemic. They conclude that during the COVID-19 pandemic, both gold and oil markets have been inefficient, and their efficiency is sensitive to scales and market sentiments. Similarly, Salisu et al. (2020) explore the impact of the COVID-19 pandemic and oil stock. The finding of panel vector autoregressive (PVAR) model estimates reveals that both oil and stock market may experience a greater initial and prolonged impact of own and cross shock during the pandemic as compared to the situation before it; whereas the logit model result suggests that the probability of negative oil and stock returns during the COVID-19 pandemic may be due to uncertainties associated with these markets. In comparison, Jia et al. (2021) employed a computable general equilibrium (CGE) model with multiple sectors and multi-households along with considering six scenarios to investigate the impact of the COVID-19 pandemic on crude oil prices, economic performance, and carbon intensity during 2020. This study disaggregated the impact of the COVID-19 pandemic on the changes in factor inputs and changes in consumer preferences. The finding shows that the decline in factors input is a major cause of the economic downturn during the COVID-19 pandemic.

Stock markets are highly vulnerable to external shocks related to the economic and political situation. COVID-19 induced economic lockdowns, and chaos has crashed many stock markets across emerging and developed countries. Narayan et al. (2020) investigated the relationship between currency deviation due to pandemics on stock market performance in Japan. This objective by using several variants of econometric tools and empirical specifications. The findings conclude that the sharp decline in depreciation of the Yen vis-à-vis dollar can lead to gains in stock returns. This study concludes that one standard deviation depreciation of the Japanese Yen during the COVID-19 pandemic improves stock market return by 71% on average from January 2020 to August 2020. Topcu and Gulal (2020) attempted to explore the impact of COVID-19 on the stock market returns of emerging countries from March 10 to April 30, 2020, by using the Driscoll-Kraay estimation technique. The findings show a significant negative impact of the COVID-19 pandemic on the emerging market stock. Furthermore, this impact is highest in Asian emerging markets, whereas emerging markets in Europe are experiencing the lowest impact. They, therefore, urge the official response on time with stimulus packages to offset the effect of the pandemic.

Furthermore, Ali et al. (2020) investigate the reaction of global financial markets in terms of decline and volatility associated with the COVID-19 pandemic. The findings reveal that China has stabilized while other global markets have witnessed considerable freefall, especially in later phases of the spread. Furthermore, when the pandemic reached the USA, even the relatively safer commodities suffered considerable volatility. Whereas Sharma et al. (2021c) examine the time-frequency relationship between COVID-19 cases and stock market return in top 15 most affected countries. They employed wavelet coherence and partial wavelet coherence on daily data from February 1, 2020 to 13 May 2020. The findings reveal strong co-movement between COVID-19 cases and stock market return of sampled countries. Similarly, Mazur et al. (2021) investigated the impact of the COVID-19 pandemic on US stock market performance during the crash of March 2020. The findings reveal that the COVID pandemic natural gas, food, healthcare, and software stock earned high positive returns, whereas petroleum, real estate, hospitality, and entertainment sectors fall dramatically. Whereas Koçak et al. (2021) investigated aggregated cointegration between the COVID-19 pandemic and stock returns of major environmentally friendly companies from 2nd January to October 5, 2020. The findings reveal that the Standard & Poor's (S&P) 500 Carbon Efficiency Index (CEI) significantly gains from the COVID-19 outbreak.

The pandemic induced uncertain and disastrous market conditions, along with a plunge in the oil market, forex market, and stock market returns, initiated the search for alternatively safer assets. Therefore, the crypto market becomes a research hotspot after the pandemic outbreak. Demir et al. (2020) examined the impact of COVID-19 cases/deaths on major cryptocurrencies such as bitcoin, Ethereum, and Ripple using wavelet coherence analysis on daily data. The objective is to figure out whether cryptocurrencies can serve as a hedge asset during the pandemic. The finding reveals the negative of the COVID-19 pandemic on bitcoin initially; however, this relationship becomes positive during the later periods. Similarly, the findings of Ethereum and ripple also exhibited similar patterns with weaker interactions. This study, therefore, supported the hedging property of cryptocurrencies.

Similarly, Zhang et al. (2021) explored the impact of the COVID-19 pandemic on bitcoin prices by focusing on dynamic correlation and volatility spillover of the bitcoin futures to spot prices by employing vector autoregression-dynamic correlation coefficient-generalized autoregressive conditional heteroskedasticity (VAR-DCC-GARCH) model and vector autoregression-Baba, Engle, Kraft, and Kroner-generalized autoregressive conditional heteroskedasticity (VAR-BEKK-GARCH) models. The findings reveal that spot and futures markets of Bitcoin are highly connected and a bi-directional volatility spillover exists. It implies that bitcoin futures can indeed hedge against risks in the bitcoin spot market. Moreover, COVID-19 has increased the hedging ability of bitcoin futures.

Furhermore, Vukovic et al. (2021) examine the safe-haven property of the crypto market during the turmoil of the COVID-19 pandemic by developing a global composite index that measures time-varying movements on each day. The findings of OLS, quantile, and robust regression reveal no statistically significant influence of the COVID-19 crisis on the crypto market in the first wave, whereas in later periods, positive spillover is observed from risky assets (S&P 500) on the crypto market. Whereas Iqbal et al. (2021) investigate the impact of the COVID-19 pandemic measured by daily infections worldwide on daily returns of the top ten cryptocurrencies based on market capitalization by using the Quantile-on-Quantile approach. The findings reveal that the changing intensity of the COVID-19 pandemic affects the Bearish and the Bullish market asymmetrically. They conclude that most cryptocurrencies can absorb small shocks but fail to resist huge changes except bitcoin, ADA, CRO, and Ethereum.

The review of the above discussed and other empirical literature concerning the impact of COVID-19 on economic growth and volatilities across various assets markets and stock markets has mostly employed estimation techniques that provide average impact throughout the sample period. Given changing intensities of the COVID-19 pandemic along with the stock market and prices of other assets, these findings can lead to a superficial and sometimes misleading conclusion. The severity of these fluctuations urges a technique that can model daily fluctuation to provide better generalization. Therefore, the disagreements among many scholars concerning the impact of COVID-19 on returns of the stock market and other major assets can be generalized to this methodological issue. Therefore, there is a need for a study that can model the daily response of markets and assets to fluctuations in the COVID-19 pandemic.

Date, source and methodology

Data

The sample consists of daily observations from January 24, 2020 to December 20, 2021. The daily new additions of COVID-19 cases worldwide are collected from the website of our world in data.1 We use daily WTI (West Texas Intermediate) oil price data in US dollars per barrel. The daily bitcoin data in USD is sourced from yahoo finance.2 We chose bitcoin since it is the most extensively traded cryptocurrency on the planet. Similarly, we use the daily data of the Carbon Efficiency Index (CEI) for S&P 500. The S&P 500 Carbon Efficient Index is intended to assess the performance of S&P 500 firms while underweighting or overweighting those that emit less or more carbon per unit of revenue. All variables are collected in raw format daily and then converted to return series using the formula below.where is return series, is the natural log of the current price of each series, whereas is the natural log of lagged price of each series. Fig. 2 shows the graphical representation of returns for all series. As seen in the graph, there are many variations in each series at the commencement period of the COVID19 outbreak. This means that all series are subjected to significant COVID-19 shocks.

Fig. 2

Returns series over time

Note: CEI, WTI, BTC, and COVID19 denote carbon efficiency index, oil price, bitcoin, and worldwide COVID19 daily new cases, respectively. All series are in the form of log returns.

Table 1 shows the summary statistics for both raw and returns series. The average value of COVID-19 daily cases is the highest, followed by BTC, CEI, and WTI for both raw and return series. Similarly, we observed the highest degree of dispersion in COVID-19 daily cases, followed by BTC, CEI and WTI. Due to COVID-19 lockdown at the peak of viral dissemination, WTI's minimum stat is -.36.96. This is the first time in history when the oil prices went negative due to the severe COVID-19 economic shock. Jarque-Bera (JB) stats are highly significant, meaning that all variables are not normally distributed. Such evidence validates that we can employ the non-parametric method. Hence, rolling window multiple correlation analysis is a good choice. Interestingly we found a negative correlation between COVID-19 and WTI in the returns series. Hence, rolling window multiple correlation is used to investigate the finding further.

Table 1

Descriptive Statistics (raw data).

Variable	Obs.	Mean	Std. Dev.	Min	Max	JB	Correlation
CEI	484	484.795	77.419	289.14	612.07	20.07***	0.763***
WTI	484	53.131	17.782	−36.98	85.64	16.24***	0.597***
BTC	484	29548.098	19653.553	4970.79	67566.797	50.93***	0.719***
COVID19	484	414777.47	246662.93	0	905842	24.56***	–
Descriptive Statistics (returns series)
CEI	482	.001	.017	−.131	.092	4397***	−0.021***
WTI	482	.001	.084	−1.322	.601	3521***	−0.009***
BTC	482	.003	.046	−.465	.172	1240***	0.066***
COVID19	482	.076	.258	−1.586	3.59	1306***	–

***p < 0.01, **p < 0.05, *p < 0.1.

Research methodology

Rolling window multiple correlation

The inflation of the Type I error is one of the collateral effects of multiple testing when the statistical significance of the rolling window correlation coefficients is calculated (Lehmann and Romano, 2006). Multiple testing will result in false significance levels, leading to erroneous and biased interpretations and decisions about the link between two or more time series under investigation (Polanco-Martínez, 2020). Therefore, a solution for overcoming the problem of multiple testing (Type I error) in bivariate or multivariate correlation is desperately needed. The Bonferroni3 technique is perhaps the most widely used method for dealing with the comparison problem or multiple testing. However, the “Bonferroni adjustment” approach is highly cautious because overlapping windows are dependent (Polanco-Martínez, 2019; Telford, 2013). To solve this flaw, Polanco-Martínez (2020) introduced a rolling window multiple correlation approach which is based on the false discovery system (FDS) of Benjamini and Hochberg (1995). The Benjamini and Hochberg (BH)4 adjustment approach performs substantially better and is less conservative (in terms of statistical power) than Bonferroni and other comparable correction methods, and is best appropriate when a large number of comparisons are made (Benjamini and Hochberg, 1995; Chen et al., 2018; Polanco-Martínez, 2020). There are also several p-value correction methods which are purely based on the sequential goodness of fit measurement (SGOF), the family-wise error rate (FWER), and false discovery rate (FDR) (these corrective methods are based on parametric estimations). The other p-value corrective method of Telford (2013) is purely based on non-parametric estimations. However, the R package5 “RolWinMulCor” of Polanco-Martínez (2020) makes it simple for potential users to take advantage of all p-value corrective methods mentioned above.

Example

In contrast, in Polanco-Martínez (2020), we generated the same periodical event (P1 = 21) in two synthetic time series (COVID-19 and CEI; Fig. 3b) depending on the nature of the data. It is contaminated by Gaussian noise with a mean zero and a standard deviation of 1 for all series and time intervals as under;In above Eq. (2), two-time series are divided into time intervals based on the total number of observations as under;

Fig. 3b

Rolling window correlation between COVID19 and FDS. P-values were corrected by “BH” method. Horizontal continuous black line shows zero correlation while horizontal grey, and doted grey lines indicate p-values at 5% and 10%. Rolling window size is 21 days.

Results and discussions

We estimate the bivariate and multiple correlation heatmap graphs between COVID19, carbon efficiency index, oil price, and bitcoin. In each heatmap, we can observe the strength of the correlation between the variables by the horizontal color bar. The darker red color shows a strong positive correlation while the light red color shows the weak positive correlation. Similarly, the dark blue color indicates a strong negative correlation whereas light blue color shows the weak negative correlation among the variables. Furthermore, the transparent green color means the insignificant relationship between variables.

Bivariate case

Fig. 3a depicts the correlation between COVID-19 daily cases and the carbon efficiency index in the bivariate heatmap. It is clearly evident from the figure that there is a significant weak negative correlation between COVID-19 cases and CEI, as indicated by light red colour at the start of the COVID-19 pandemic (during March 2020 and September 2020). The remaining graph shows the moderate positive relationship between COVID-19 and CEI. The findings are in line with Koçak et al. (2021) confirm that the COVID19 pandemic positively influenced green firms. However, the overall graph shows the positive (negative) and significant relationship between COVID-19 and CEI. These findings come from the fact that COVID19 transmission rapidly increased, and lockdown measures were implemented worldwide at the start of the pandemic. Later, the lockdown brings shutdown to the whole sectors of the countries, especially the industry and banking sectors. This brings a huge fluctuation in the stock returns of the carbon efficiency index. Specifically, during the lockdown, all industry sector was closed, and there seemed to be negative correlations between COVID-19 and CEI.

Fig. 3a

Rolling window correlation between COVID19 and CEI (Bivariate Heat Map). Transparent green shows an Insignificant correlation (>0.05). Line counters show similar values to the correlation coefficients. P-values were corrected by “BH” method. The Rolling window size is 21 days.

Fig. 3b shows the dynamic correlation coefficients on the left y-axis and p-values on the right y-axis. We can observe weak/strong coefficients and significant/insignificant levels throughout the sample period. It shows the asymmetric behaviour between the COVID-19 and stock returns of the carbon efficiency index. COVID19 and stock prices have asymmetric correlations because investor mood is influenced by good and bad news in the market.

Fig. 4a shows the correlation between COVID-19 daily cases and oil prices in the bivariate heatmap. We observe the negative correlation between COVID19 and oil price at the start of the COVID pandemic. However, in the later period, we see a strong negative and significant correlation between COVID-19 and oil prices. These findings show that COVID-19 shocks cause severe oil prices slump worldwide. Overall findings suggest an asymmetric association between COVID19 and oil prices over the different time intervals.

Fig. 4a

Rolling window correlation between COVID19 and WTI (Bivariate Heat Map). Transparent green shows an Insignificant correlation (>0.05). Line counters show similar values to the correlation coefficients. P-values were corrected by “BH” method. The Rolling window size is 21 days.

Fig. 4b shows the dynamic correlation coefficients on the left y-axis and p-values on the right y-axis. We observe the strong/weak coefficients and significant/insignificant p values during the time interval. The findings suggest a nonlinear relationship between COVID-19 and oil prices.

Fig. 4b

Rolling window correlation between COVID19 and WTI. P-values were corrected by “BH” method. Horizontal continuous black line shows zero correlation while horizontal grey, and doted grey lines indicate p-values at 5% and 10%. Rolling window size is 21 days.

Fig. 5a shows the correlation between COVID-19 cases and bitcoin returns in the bivariate heatmap. The bivariate heat graph shows a strong negative correlation between COVID-19 cases and BTC returns from the start of the sample period to mid of the sample period. However, from mid sample period to the end sample period, we got a strong positive correlation between COVID-19 and BTC. The overall results show an asymmetric association between COVID-19 and BTC. The findings show BTC was a safe haven for investors from November to September 2021.

Fig. 5a

Rolling window correlation between COVID19 and BTC (Bivariate Heat Map). Transparent green shows an Insignificant correlation (>0.05). Line counters show similar values to the correlation coefficients. P-values were corrected by “BH” method. The Rolling window size is 21 days.

Fig. 5b shows the dynamic correlation coefficients on the left y-axis and p-values on the right y-axis. We observe the strong/weak coefficients and significant/insignificant p-values at the different time intervals. The findings suggest a nonlinear relationship between COVID-19 and BTC. Overall findings in all bivariate cases showed asymmetric associations of COVID-19 with CEI, WTI and BTC.

Fig. 5b

Rolling window correlation between COVID19 and BTC. P-values were corrected by “BH” method. Horizontal continuous black line shows zero correlation while horizontal grey, and doted grey lines indicate p-values at 5% and 10%. Rolling window size is 21 days.

Multivariate case (tetravariate)

Fig. 6a shows the combined correlation impact of COVID-19, WTI and BTC on CEI in tetravariate heat map graph. We find weak negative and significant correlation impact of COVID-19 on WTI, while a weak positive impact is observed with BTC on CEI throughout the sample period.

Fig. 6a

Rolling window correlation between CEI (dependent variable) and COVID19, WTI & BTC (independent variables) (multivariate Heat Map). Transparent green shows Insignificant correlation (>0.05). Line counters show similar values of the correlation coefficients. P-values were corrected by “BH” method. Rolling window size is 21 days.

Fig. 7a shows the combined correlation impact of COVID-19, CEI and BTC on WTI in tetravariate heat map graph. Unlike the previous findings in Fig. 6a, we find a significant and stronger positive impact of COVID19, CEI, and BTC on WTI throughout the sample period. The findings suggest that CEI and BTC return cancelling the COVID-19 shocks on oil price in a greater strength compared to Fig. 6a.

Fig. 7a

Rolling window correlation between WTI (dependent variable) and COVID19, CEI & BTC (independent variables) (multivariate Heat Map). Transparent green shows Insignificant correlation (>0.05). Line counters show similar values of the correlation coefficients. P-values were corrected by “BH” method. Rolling window size is 21 days.

Fig. 8a shows the combined correlation impact of COVID-19, WTI and CEI on BTC in tetravariate heat map graph. Unlike the previous findings in Figs. 6a and 7a, we find a significant and strong/weak positive impact of COVID-19, CEI, and WTI on BTC in the whole sample period. The findings suggest CEI and WTI return, cancelling the COVID-19 shocks on bitcoin. We find an overall positive and significant association in all tetravariate heatmap graphs. This implies that CEI, BTC and WTI are alternatively safe havens for the investors during the COVID-19 pandemic. During the COVID-19 epidemic, investors should diversify their risk by building a portfolio and investing in CEI, BTC, and WTI stocks.

Fig. 8a

Rolling window correlation between BTC (dependent variable) and COVID19, WTI & CEI (independent variables) (multivariate Heat Map). Transparent green shows Insignificant correlation (>0.05). Line counters show similar values of the correlation coefficients. P-values were corrected by “BH” method. Rolling window size is 21 days.

Figs. 6b and 7b, and 8b show dynamic correlation coefficients on the left y-axis and p-values on the right y-axis. We find strong values of coefficients during the whole sample period. However, the significant levels are varied during the whole periods. In all tetrvariate cases, we find a nonlinear association between variables. Because both positive and bad market news influences investors' emotions, non-linear tendencies among variables were expected.

Fig. 6b

Rolling window correlation between CEI (dependent variable) and COVID19, WTI & BTC (independent variables). P-values were corrected by “BH” method. Horizontal continuous black line shows zero correlation while horizontal grey, and doted grey lines indicate p-values at 5% and 10%. Rolling window size is 21 days.

Fig. 7b

Rolling window correlation between WTI (dependent variable) and COVID19, CEI & BTC (independent variables). P-values were corrected by “BH” method. Horizontal continuous black line shows zero correlation while horizontal grey, and doted grey lines indicate p-values at 5% and 10%. Rolling window size is 21 days.

Fig. 8b

Rolling window correlation between BTC (dependent variable) and COVID19, WTI & CEI (independent variables). P-values were corrected by “BH” method. Horizontal continuous black line shows zero correlation while horizontal grey, and doted grey lines indicate p-values at 5% and 10%. Rolling window size is 21 days.

Conclusion

The emergence of the COVID-19 pandemic has severely disrupted nearly all spares of social and economic activities. The seasonal emergence of various variants made this virus the gravest challenge globally by posing unprecedented challenges to public health, employment, food security, and psychological stresses. The emergence of vaccine-resistant Omicron variant has considerably increased daily infections and destroyed hopes for normalization. The economic lockdowns and travel restrictions to contain virus spread has severely impacted airlines, shipping, travel, and manufacturing sectors and dramatically declined demand for crude oil along with stock market crashes across the globe. Admit this turmoil. This study aims to explore the hedging properties of various assets. This objective is realized by investigating the impact of COVID-19 on the crude oil, carbon efficiency index, and bitcoin using the recently advanced rolling window multiple correlation approach proposed by Polanco-Martínez (2020). This method is based on the new corrected p-value method and has advantages over the other correlation techniques.

The study's findings can be summarised as follows; (A) In the bivariate case; (1) COVID-19 significantly and positively affected the carbon efficiency index. (2) COVID-19 significantly and negatively affected oil prices. (3) COVID-19 has a significant and asymmetric impact on bitcoin (B) In all tetra-variate cases, we find a positive and significant correlation between BTC and CEI while a significant negative impact on WTI is observed. The robust findings corroborate important policy implications for the investors. Investors may alternatively diversify their portfolio from WTI to BTC and CEI during the pandemic type crisis. In other words, low-carbon company shares and crypto markets provide a safe harbour for investors against COVID-19 shocks.

The rolling window multiple correlation explores the nature of the causal relationship throughout the sample period. It cannot provides estimates for the short-term and long-term response of dependent variable to change in explanatory variables. Therefore, future research can model the short-term and long-term impact of COVID-19 on the behavior of selected assets. Moreover, the inclusion of gold, silver, and other minerals can extend the scope of the current investigation. Furthermore, future research can address the behavior of other financial assets such as green bonds, green equities, and green innovations that could help ensure long-term sustainability.

Funding

•Jiangsu Modern Finance and Taxation Collaborative Innovation Center, grant no.20WTB005.

•R&D project DATA4LOWDENSity COLab, (UIDP/04011/2020).

Author contributions

Zeeshan Fareed Methodology, Software, Formal analysis, Writing - Review & Editing Shujaat Abbas: Data Curation, Conceptualization, Writing Original Draft, Writing - Review & Editing, Resources. Livia Madureira: Writing - Review & Editing, Investigation, Supervision. Zhenkun Wang: Writing - Original Draft, Writing - Review & Editing, Investigation.

Uncited references

Fattouh and Imsirovic, 2020; Ma et al., 2021; Worldometer, 2021.