Evidence from the nonlinear autoregressive distributed lag model on the asymmetric influence of the first wave of the COVID-19 pandemic on energy markets

Abstract

The COVID-19 pandemic remained a global risk factor and integrated into various means in the functioning of companies, economies and financial markets. Therefore, this paper investigates how COVID-19 influences the energy market in the main financial markets (China, France, Germany, Italy, Spain and the United States), using time series from February 28, 2020, to November 3, 2020. The goal of this research is to investigate the asymmetric impact of COVID-19 from leading financial markets on energy commodities. In this regard, the non-linear auto-regressive distributed lag (NARDL) framework is employed to capture the long-run asymmetric reactions. The econometric design allows to explore the long-term asymmetric reactions of dependent variables through positive and negative partial sum decompositions of changes in the explanatory variables. The quantitative results show a significant long-run asymmetric interdependence between the number of new SARS-CoV-2 incidence and mortality and the daily percent change in close price of future contracts pertaining to Brent oil, crude oil WTI, carbon emissions, gasoline RBOB, heating oil, Chukyo kerosene, and natural gas. Furthermore, no asymmetry is found in the case of ethanol and fuel oil futures. The novelty of this article is the study of the impact of COVID-19 on the energy sector during the first two waves of COVID-19 by applying the NARDL model that allows to capture long-term asymmetric reactions. Certainly, further research on this topic is necessary due to the permanent shifts in the pandemic, as well as the availability of longer data periods on COVID-19.

Keywords

COVID-19 energy commodities NARDL model TGARCH

Introduction

The world is facing a unique shock on account of the novel coronavirus outburst,¹ being thrown into fear by the “evil hand” labeled COVID-19.² The disease has caused serious economic consequences due to population jobless and business bankruptcies, while financial markets have replied with striking fluctuations.³ Mazur, Dang⁴ claimed that many public corporations own profitable assets, along with substantial free cash flow potential, but the crash of quotations did not happen because of the bubble burst Current pandemic reveals the leading case of a market crash provoked by a medical emergency.⁵ In fact, stockholders, portfolio managers, and decision makers are confronted with a combined health, financial, and economic decline that ruins the whole economy and yields remarkable concerns relating to shock propagation amid assets.⁶

The COVID-19 pandemic troubled global supply and demand chains,⁷ but its influence is different for various kinds of commodities, as opposed to the 2008–09 global recession when nearly all commodity prices encountered great and lasting drops.⁸ In this regard, Nwosa⁹ documented that the ongoing health crisis had more effects on the fuel price, currency exchange, and stock market performance than the 2009 and 2016 global recessions. Moreover, Adekoya and Oliyide¹⁰ supported that the corona crisis has been substantially accountable for risk dispersal via varied commodity and financial markets. For instance, Gharib, Mefteh-Wali¹¹ found a joint contagion effect of bubbles in oil and gold markets over the pandemic. Wang, Yang¹² found that the risks associated with coal and WTI crude oil are distributed and interacted with one another. Amamou and Bargaoui¹³ reported that since the emergence of the second pandemic wave, the number of reported cases has lost explanatory capacity, but mortality caused by this virus has become a substantial explanatory factor.

The worldwide energy sector was affected by the coronavirus outbreak on account of the severe fall in oil prices and demand, energy supply shock, and out-of-work that conducted on an extensive energy crisis.¹⁴ Elavarasan, Shafiullah¹⁵ claimed that the electric sales dropped around 4.9 percent and 14.8 percent in France, Italy, Spain, Portugal, Belgium, the Netherlands, Germany, and the United Kingdom, whereas it dropped among 5.7 percent and 26 percent in the United States, India, Singapore, Australia, and China. Also, for the second quarter of 2020 compared to the corresponding period of 2019, Şahin, Ballı¹⁶ exhibited a decrease of electricity production from total non-renewables by 21%–25% for France, Germany, Spain, and the UK, as well as by 11% for Turkey. As well, Wang, Li¹⁷ proved that China's electricity declined by 29% and Wang, Li¹⁸ documented that the US petroleum consumption throughout January 2020 – March 2021 was almost 18.14% lesser than that under the common condition. Khan, Su¹⁹ noticed that the influence of COVID-19 on energy prices is negative across all the quantile, but the effect is greater on the oil prices as compared to the natural gas and the heating oil price. Besides, Wu, Wang²⁰ found that that social media supports the prediction of oil price, production, and consumption. In this regard, Norouzi, Rubens²¹ noticed that a 1% rise in the severeness index determined a reduction of 0.9% in oil demand alike Qin, Zhang²² that confirmed the negative impact of pandemics index on oil prices. The reduced oil prices are argued by the limited production and traveling, despite the lack of arrangement between Saudi Arabia and Russia, who have dissimilar budget requirements, fiscal deficits, and market perspectives.²³ Selmi, Bouoiyour²⁴ suggested that partnership is imperative between oil-producing states and oil-importing nations to achieve a reasonable price. In this vein, Brosemer, Schelly²⁵ ascertain the current pandemic as a public health, economic, and justice crisis which intensify contemporary crises in energy domination. In contrast to different assets, crude oil prices oscillated sharply by virtue of the disequilibrium of supply and demand throughout the pandemic, notably the WTI crude oil in April, 2019.²⁶ As such, states in full quarantine suffered a mean decline of 25% in energy demand per week, whereas 18% fall in case of nations in partial lockdown.²⁷ According to IEA,²⁸ the energy use decreased by 6% in 2020, seven times higher than the effect of the 2008 financial crisis on global energy demand, being the major slump in 70 years in percentage terms and the greatest ever in absolute terms. Therewith, the oil-exporting developing nations were punched due to their non-diversification and reliance on a sole commodity for export.²⁹ Tahir and Batool³⁰ advised that a new world order is set out consisting of exploiter and exploited, the world supreme authority being own by those nations who will outlive from COVID-19. Under such circumstances, Wang and Zhang³¹ argued that the spillover effect of China's economic restoration post-COVID-19 has the most noticeable effect on the increase in energy consumption in high-income nations, preceded by middle-income states. Reduced global energy demand imperils gas more than other fuel sources over the next decades because of prolonged coal supply and increasing renewables.³²

Oil demand is supposed to go up steadily as tourism and travel are restrained by health matters and as overall economic activity is predicted to go back to pre-pandemic levels in 2022.³³ As well, the related shock is temporary, although with very lengthy consequences.³⁴ Therefore, Okorie and Lin³⁵ advised that COVID-19 fractal contagion consequences vanish in the medium and long term and Szczygielski, Bwanya³⁶ supported the dissolution of the progressive effect of COVID-19 associated incertitude as the slump advances. As opposed, Seven and Yılmaz³⁷ cautioned that nations greatly dependent on natural resources and leisure industries will face a delayed revival owing to their negative connection with the recovery rate.

In addition to the negative effects of the COVID-19 pandemic, the positive effects are not long in coming, even in the most unfortunate situations. According to Wang and Su,³⁸ the NO2 concentration decreased significantly in the atmosphere. From a spatial perspective, the reduction of environmental pollutants initially appeared in areas with severe outbreaks, because they were the first to implement stringent restrictions, and along the way more and more regions adopted quarantine measures. As well, Wang, Li³⁹ supported that the pandemic has led to a greater reduction in greenhouse gas emissions from developing markets such as China and India. Thus, Wang, Wang⁴⁰ advised that boosting decarbonization or energy-saving systems directly contributes to a reduction in carbon intensity. However, Jiang and Chen⁴¹ claimed that the post-outbreak carbon market spillovers are roughly twice as large as the pre-COVID-19 period.

In context of such a multilayered issue and a sharp decline in energy demands, our study provides insights on how the first phase of the COVID-19 outbreak impacted energy markets. The oil industry was considered for its unique characteristics: it is unexpected, inconstant, and volatile.⁴² In this regard, Zhang and Hamori⁴³ claimed that the oil prices rest volatile as the COVID-19 pandemic persists. This paper contributes to the literature in several ways. First, prior papers that dealt with the impact of novel coronavirus on the energy sector covered either the COVID-19 global cases^44–48 or those outside China,^44,49 out of USA and Japan,⁵⁰ France and Cameroon,⁵¹ Turkey,^52,53 or from China.^44,49 To address this gap in the field, the current paper includes data on the number of novel SARS-CoV-2 incidence and mortality from the major global capital markets of China, France, Germany, Italy, Spain, and the United States. Second, unlike previous studies,^50,54–56 this study introduces the NARDL (nonlinear autoregressive distributed lags) methodology to quantify the asymmetric effect. Third, the paper analyzes the volatility behavior of the energy markets using the threshold autoregressive conditional heteroscedasticity (TGARCH) model. Not least, current study covers several energy commodities such as Brent oil, carbon emissions, crude oil WTI, ethanol, fuel oil, gasoline RBOB, heating oil, Chukyo Kerosene, natural gas.

The rest of the paper is laid out as follows. The previous literature is discussed in Section “Overview of the literature”. The dataset and econometric strategy are described in Section “Quantitative design”. The empirical investigation results and discussion are presented throughout Section “Empirical results”. The manuscript draws the conclusions within the final section.

Overview of the literature

Crude oil is one of the world's most valuable commodities that influence the worldwide economy, their prices being contingent on the economic circumstances, business cycle, and turning points,⁴⁵ while its price variability is a prominent gauge for stock market instability and financial distress.⁵⁷ The pandemic perturbed the operation of the worldwide crude oil market and several financial markets, hence generating risks to investors.⁵⁸ For this reason, Salisu and Adediran⁵⁹ suggested that infectious disease equity market volatility is a reliable indicator of electricity market fluctuations. Table 1 summarizes recent research on the effect of the COVID-19 outbreak on commodity prices.

Table 1.

Prior research on the influence of the COVID-19 disease outbreak on electricity sector.

Author(s)	Period	Variables	Methods	Outcomes
Christopoulos, Kalantonis⁶⁰	January 1, 2020 – May 13, 2021	CBOE crude oil implied volatility index, COVID-19-related deaths and COVID-19-related speed of death and infection growth, economic and market uncertainty index, day of the week effect	Fixed-effect panel models	The pandemic affected the volatility of the price of crude oil worldwide
Dmytrów, Landmesser⁶¹	January 2, 2020 – March 15, 2021	Brent crude oil, CO2 allowances, Heating oil, Palm oil, Ultra Low-Sulphur Diesel, Coal, Natural gas, Gasoline, Ethanol, Uranium	Dynamic Time Warping (DTW) distance method	Ultra-Low-Sulphur Diesel, heating oil, crude oil, and gasoline are poorly related with COVID-19 Natural gas, palm oil, CO2 allowances, and ethanol are strongly connected with the pandemic
Nyga-Łukaszewska and Aruga⁵⁰	January 21, 2020 – June 2, 2020	Crude oil, natural gas, WTI, Platts Dubai crude oil	Auto-Regressive Distributive Lag (ARDL)	Disease have a detrimental effect on petroleum products prices in US, but have a beneficial impact on price of gas. The crude oil market in Japan does seem to have an adverse impact
Si, Li⁶²	January 20, 2020 – April 20, 2021	Stock price volatility series of oil exploitation, coal mining, other mining, petrochemical, power supply equipment, electrical automation equipment, power, gas and optoelectronics sectors	High-dimensional and time-varying factor-augmented VAR model	Over January – June, 2020 the volatility spillover of the pandemic is not only the highest, but also remained prolonged for oil exploitation, power and gas sectors
Hemrit and Benlagha⁶³	January 3, 2005 – June 30, 2020	Stock volatility index (VIX), S&P 500 Index, Arca Tech 100 Index, Oil prices and quantities, Gold prices, Economic Policy Uncertainty, World Pandemic Uncertainty Index	Quantile regressions	Renewable energy index reacted positively to the pandemic-induced insecurity shock
Gharib, Mefteh-Wali⁶⁴	November 1, 2019 – December 31, 2020	Daily West Texas Light crude oil, North Sea Brent crude, diesel, gasoline	Log Periodic Power-Law Singularity (LPPLS) model	Over pandemic, West Texas Light and North Sea Brent crude oil prices are driven by bubbles, but no significant negative bubble was detected for gasoline and diesel prices
Bourghelle, Jawadi⁶⁵	January 2, 2014 – April 1, 2020	West Texas Index, Eonomic Policy Uncertainty Index, Equity Market Related EPU Index	Vector Autoregressive (VAR) framework	Oil price volatility reacted largely to the pandemic shock
Le, Le⁵⁴	January 17, 2020 – September 14, 2020	West Texas Intermediate, Brent, Economic Policy Uncertainty Index, Equity Market-related Economic Uncertainty Index, VIX, MSCI World Index, FTSE All-World index, S&P Global 100	Autoregressive distributed lag (ARDL) model	COVID-19 cases growth, US EPU, and the VIX drives the fall in WTI prices
Maneejuk, Thongkairat⁶⁶	December 29, 2019 – December 30, 2020.	Natural gas, Gasoil, Heating Oil, Coal, and Brent crude oil	Markov Switching Dynamic Copula	Favorable and unfavorable viral shocks have the same impact on the energy markets
Algamdi, Brika⁵⁵	January 22, 2020 – June 14, 2020	Oil prices in Saudi Arabia	ARDL (Autoregressive Distributed Lag) model	Death ratio negatively influence oil price fluctuation
Adedeji, Ahmed⁶⁷	March 20, 2020 – May 28, 2020	WTI, Brent, Daqing, Bonny light	Structural vector autoregressive (VAR) method	The effect of pandemic counted for smallest shares of shift in Bonny and Daqing oil prices, the effect being lower on BRENT and WTI
Atri, Kouki⁵⁶	January 23, 2020 – June 23, 2020	West Texas Intermediate (WTI) oil price	ARDL analysis	The pandemic crisis negatively influence the oil price

Crude oil price variations have a substantial influence on the worldwide economy.²⁰ Weng, Zhang⁶⁸ proved that COVID-19 pandemic news has more predictive information, which is essential for temporary volatility prediction of crude oil futures. According to Niu, Liu,⁶⁹ outbreak reports acquired through panic index and country sentiment index could be used to estimate China's petroleum products fluctuations.

The first strand of the literature is focused on the relationship between oil and the stock market over the COVID-19 period. Kamaludin, Sundarasen⁷⁰ established a strong coherency between US stock exchanges and oil prices and Topcu and Gulal⁷¹ proved that a shock to real oil prices negatively influences stock markets. Prabheesh, Padhan⁷² noticed that decreasing oil prices behave as a detrimental prediction for the stock market. Therefore, investors possessing oil-derived assets are vulnerable to unfavorable oil price variability.⁷³ According to Kamdem, Essomba,⁵¹ an increase in SARS-CoV-2-related casualties relates to declining oil products and Brent pricing. exhibited that a surge in fatalities due to SARS-CoV-2 matches with descending prices for Crude oil and Brent. Nyga-Łukaszewska and Aruga⁵⁰ found that the aggregate volume of US COVID-19 occurrences had a detrimental impact on commodity prices, even though the effect for Japan was rather minimal. By considering the new infection cases day by day recorded globally and in China, Albulescu⁴⁴ documented a negative long-term association between oil prices and coronavirus figures, besides financial volatility and the US economic uncertainty, but no significant relation when new cases outside China were considered. In the same vein, Yilmazkuday⁴⁹ revealed a decrease by 0.4% in crude oil prices after a weekly growth of 1000 in daily COVID-19 cases in the rest of the world, although no significant effects when COVID-19 cases in China were examined. Liu, Wang⁴⁸ on the other hand, reported that the pandemic spread has a beneficial impact on petroleum products and stock market performance.

The energy markets react in a different way to undesirable announcements and beneficial media, thus exerting various outcomes on stock price returns and fluctuations.⁴² Hence, the second strand of literature is oriented on oil price's reaction to COVID-19 related news. Haroon and Rizvi⁷⁴ established that fear induced by mass media is related to intensified variability in global stock markets, being stronger for sectors worst affected by ongoing coronavirus, such as energy. Kwan and Mertens⁷⁵ noticed that the energy industry reveals more systematic risk and stockholders presume it to register a higher drop than the broader market. Therefore, another strand of literature deals with oil price reactions to pandemic news. Sharif, Aloui⁷⁶ asserted that news concerning oil charges and the pandemic are major determinants of the US stock market and Devpura and Narayan⁴⁶ exhibited that COVID-19 cases and deaths determine a rise of between 8% and 22% in oil price volatility. Furthermore, after a certain threshold of COVID-19 new cases (e.g. 84479), Narayan⁴⁷ noticed that the influence of pandemic on oil prices is higher. Salisu, Ebuh² found that traders in the oil and stock sectors during the COVID-19 pandemic may face increased preliminary repercussions than before the occurrence. Ahundjanov, Akhundjanov⁷⁷ revealed that the Dow Jones US Oil & Gas Total index and the New York Harbor Conventional Gasoline Regular spot price both dropped as a consequence of a spike in the global search for COVID-19. Meher, Hawaldar⁷⁸ documented that undesirable news such as SARS-CoV-2 dispersal yielded a greater effect on the fluctuation of crude oil. Sakurai and Kurosaki⁷⁹ concluded that following the outbreak of coronavirus disease, a positive (negative) oil shock is even an optimistic (hopeless) info for the stock market compared with a comparable shock prior the crisis. On the contrary, Salisu, Akanni⁸⁰ reported an upsurge of commodity returns as the global fear index increases. Akhtaruzzaman, Boubaker⁸¹ provided evidence that the ongoing pandemic can temper the link between shifts in oil prices and stock returns all over the world.

The third stream of research explores the influence of economic incertitude associated with overall pandemics on the volatility of energy commodities. As such, Borgards, Czudaj⁸² claimed the peculiarity of crude oil given that it registers an increased number of adverse overreactions than positive ones over the pandemic and Kartal⁵³ highlighted that VIX index has the leading impact on oil prices. Hence, Yang, Wei⁸³ supported that the bad volatility of China's crude oil futures is highly sensitive to worldwide financial unpredictability than good volatility. Yang, Ma⁸⁴ proved that Brent and WTI exert the foremost contribution in risk transfer within market. Güngör, Ertuğrul⁵² remarked that gasoline usage volatility heightened substantially after the outbreak determined by authorities’ rules to limit the virus widespread (lockdowns, travel restrictions), by social conduct (interpersonal distance, remote work, virtual meetings), and by market presumptions (hoarding). Lahmiri and Bekiros⁸⁵ revealed that the oil prices’ reduction attributed to Russia-Saudi Arabia oil price competition driven a rise in reciprocal evidence between WTI market and investor fear index measured by VIX. Nonetheless, Fasanya, Oyewole⁸⁶ proved that the impact of volatility due to the pandemic on the energy futures returns’ hinges on the extent of market performance.

Because the SARS-CoV-2 infection spread swiftly, it taken a toll on a number of countries and stock exchanges simultaneously,³⁵ with petroleum being among the primary transmitters of realized volatility.⁸⁷ However, Foglia and Angelini⁸⁸ concluded that COVID-19 pandemic and the termination of the OPEC production deal changed the status of oil market from volatility issuer to volatility recipient. As such, other studies explored volatility spillovers in oil markets. Mensi, Al-Yahyaee⁸⁹ confirmed that crude oil futures lead to the risk of other markets throughout the oil crisis, COVID-19, and Brexit period relative to different terms, whereas Urom, Ndubuisi⁹⁰ reinforced that either strong and weak shocks from the electricity market boost stock market fluctuation in the long term. In this sense, Gunay⁹¹ reported that the shockwave of the COVID-19 in the overall volatility spillover is nearly eight times higher in contrast to the global financial crisis. Shehzad, Xiaoxing⁹² also proved that financial markets in the United States, Germany, and Italy have more conditional variance during the COVID-19 timeframe than during the global financial meltdown. Nevertheless, the common flight-to-safety phenomenon occurred and Corbet, Hou⁹³ have shown persistent volatility spillovers from the coronavirus and influenza indices to Chinese gold and oil futures markets. Moreover, argued by the insecurity around the scale and time span of the novel coronavirus, Atukeren, Çevik⁵⁷ found Granger-causal feedback link between the volatility of WTI and Brent crude oil prices. Corbet, Goodell⁹⁴ pointed out that decreased fuel prices have beneficial repercussions both in clean energy and coal sectors. Chang, McAleer⁹⁵ emphasized that US conventional energy sectors exhibit strong intersectoral side effects and risk spillovers through any electricity market. In contrast, Lin and Su⁹⁶ remarked merely a narrow impact of the pandemic on the pairwise connectedness between energy commodities. In addition, Samadi, Owjimehr⁹⁷ underlined that the COVID-19 breakout had no effect on the cross-shift of oil and stock prices, as well as fuel prices and exchange rates.

Quantitative design

Data

Daily data for all the investigated variables were gathered for the first stage of COVID-19 pandemic, similar Nyga-Łukaszewska and Aruga,⁵⁰ from February 28, 2020 to November 3, 2020. Table 2 lists the measures that have been selected.

Table 2.

The definition of variables.

Variables	Description	Source
SARS-CoV-2-related variables
CH-NC	New pandemic infections in China	Our World In Data
CH-ND	New pandemic fatalities in China	Our World In Data
FR_NC	New pandemic infections in France	Our World In Data
FR_ND	New pandemic fatalities in France	Our World In Data
GR_NC	New pandemic infections in Germany	Our World In Data
GR-ND	New pandemic fatalities in Germany	Our World In Data
IT-NC	New pandemic infections in Italy	Our World In Data
IT-ND	New pandemic fatalities in Italy	Our World In Data
USA-NC	New pandemic infections in USA	Our World In Data
USA-ND	New pandemic fatalities in USA	Our World In Data
SP-NC	New pandemic infections in Spain	Our World In Data
SP-ND	New pandemic fatalities in Spain	Our World In Data
Variables towards the electricity marketplace
BRENT	The percentage variation in the close price of Brent Oil Futures	Investing.com
CARBON	The percentage variation in the close price of Carbon Emissions Futures	Investing.com
WTI	The percentage variation in the close price of Crude Oil WTI Futures	Investing.com
ETHANOL	The percentage variation in the close price of Ethanol Futures	Investing.com
FUEL_OIL	The percentage variation in the close price of Fuel Oil Futures	Investing.com
GASOLINE	The percentage variation in the close price of Gasoline RBOB Futures	Investing.com
HEATING_OIL	The percentage variation in the close price of Heating Oil Futures	Investing.com
KEROSENE	The percentage variation in the close price of Chukyo Kerosene Futures	Investing.com
NATURAL_GAS	The percentage variation in the close price of Natural Gas Futures	Investing.com

Source: Authors’ own work.

There were included a variety of indicators to guide us achieve our goals, including energy commodity indices, the new incidences, and the new death toll associated with COVID-19 in China, the United States, France, Italy, Spain, and Germany.

Econometric framework

The goal of this research is to come to a robust and conclusive finding on the link among energy metrics and the COVID-19 outbreak. Evaluating the unit roots of the investigated data series is one of the initial procedures in econometric analysis. In this research, the Dickey-Fuller unit root test (ADF) will be used to study the nonstationary hypothesis alike Wang, Li,¹⁷ Atri, Kouki,⁵⁶ Maneejuk, Thongkairat,⁶⁶ Adedeji, Ahmed.⁶⁷ ADF testing involves estimating the following equation:

Δ y_{t} = α + β t + q y_{t} + \sum_{j = 1}^{k} γ_{j} Δ y_{t - j} + ε_{t}, t = 1, \dots, T

(1)

where t is a time trend, T is the length of the sample, and k measures the length of the lag in the dependent variable. The null hypothesis assumes that the variable has a unit root, and the alternative is that the variable is generated by a stationary process.¹⁸

In this analysis, daily returns were calculated as continuous compound returns, which is the first difference in the logarithm of closing prices of stock indices, as in.^47,72,74,92 The used formula is shown below:

R_{t} = \ln (\frac{P_{t}}{P_{t - 1}})

(2)

where

P_{t}

represents the price of the stock index at time t,

P_{t - 1}

depicts the price of the stock index at time t − 1, and

R_{t}

is the logarithmic yield of the stock indices. According to the prior literature, logarithmic yields are preferred, which should follow a normal distribution.

In the study of financial data series, ARCH / GARCH (Autoregressive conditional heteroskedasticity / Generalized Autoregressive conditional heteroskedasticity) approaches are being used (e.g. sales price development, rate of return on financial assets, or exchange rates). These models simultaneously evaluate and test the yield and volatility processes. The importance of these models results from the difference between conditional and unconditional variances. Unconditional variance should be time-independent, while conditional variances should depend on past events that are included in the multitude of information at time t − 1.

Engle⁹⁸ proposed the ARCH representations, while Bollerslev⁹⁹ outlined the GARCH variants. The conditional variation in a GARCH model is reliant on the past lag. By incorporating an MA (Moving-Average) process to the AR process (Autoregressive), GARCH models turn the AR process from the ARCH model into an ARMA process (Autoregressive Moving Average). The GARCH specification show the following form^36,57,84:

\begin{aligned} y_{t} = & μ + ε_{t} \sim N (0, σ^{2}) \\ σ^{2} = & ω + α_{1} ε_{t - 1}^{2} + \dots + α_{q} ε_{t - q}^{2} + β_{1} σ_{t - 1}^{2} + \dots β_{p} σ_{t - p}^{2} \end{aligned}

(3)

where ω>0 and

α_{i} \geq 0

β_{i} \geq 0

The conditional variation of random perturbations is reliant on both the historical values of the shocks and the values of the past variation, as shown by the aforementioned formulae. The coefficients $σ_{t - p}^{2}$ show the volatility's permanence, whereas the coefficients $ε_{t - q}^{2}$ indicate the volatility's speed of the reaction to capital market disturbances. The p parameter emphasizes the GARCH terms’ order, whereas q highlights the ARCH terms’ order.

A GARCH(1,1) model is deemed to be suitable for representing the evolution of volatility,⁵⁷ but we will examine different variants of the GARCH model throughout our study. A GARCH model (1,1) corresponds to an ARCH form (2), whilst the GARCH model (p, q) relates to an ARCH specification (p + q).

TGARCH (Threshold GARCH) version was pioneered by Glosten, Jagannathan.¹⁰⁰ For the TGARCH model (1,1), the variation equation can be written as below:

σ_{t}^{2} = ω + α ε_{t - 1}^{2} + β σ_{t - 1}^{2} + γ ε_{t - 1}^{2} I_{t - 1}

(4)

where I_t₋₁ is a dummy variable, namely an indicator for negative innovation: I_t₋₁ = 1 if ε_t₋₁< 0 (“negative news” has an impact on α) and I_t₋₁ denotes that “positive news” has an impact of α + γ. If γ > 0, we note that there is a leverage effect and if γ ≠ 0, the impact of the news is asymmetric.

The main objective was to identify an econometric model to model the volatility of selected stock indices. The analysis was performed using models from the GARCH family, very useful tools applied in financial economics. The TARCH model is an asymmetrical ARCH model just like the EGARCH. Asymmetric models are based on the fact that a negative news has a considerably greater impact on volatility than a positive news. According to the literature, the TGARCH model works well in the post-crisis period in capturing stock market volatility.

The nonlinear autoregressive distributed lag (NARDL) model will be used to gain insight into the linkages in energy commodities during the COVID-19 pandemic breakout. The NARDL model will be approached to test the occurrence of asymmetry in long-run relationships among the selected variables.

The NARDL model is an asymmetric extension of the linear autoregressive distributed lag (ARDL) model developed by.¹⁰¹ As a corollary, the linear ARDL technique overlooks the possibility that the independent variables’ desirable and undesirable variances have distinct effect on a dependent variable.

The general form of an ARDL model is shown below:

Δ y_{t} = μ + q y_{t - 1} + θ x_{t - 1} + \sum_{j = 1}^{p - 1} α_{j} Δ y_{t - j} + \sum_{j = 0}^{q - 1} π_{j} Δ x_{t - j} + ε_{t}

(5)

where Δ represents the first difference,

y_{t}

is the dependent variable in period t, µ denotes the intercept, x is a k × 1 vector of regressors, whereas q and θ denote the long-run coefficients. Furthermore,

α_{j}

and

π_{j}

are the short-run coefficients, p and q represent the optimal lags for the dependent variable and the independent variables, respectively, and finally,

ε_{t}

depicts the error term at time t. However, the ARDL model does not consider the possible asymmetric link among the variables. Hence, it implicitly assumes that both positive and negative fluctuations of the predictor factors have the same effect on dependent variable. As such, we are going to apply the nonlinear version of ARDL which allows asymmetric relationships. According to,¹⁰² the NARDL model is based on the asymmetric long-run equilibrium relation shown below:

y_{t} = β^{+} x_{t}^{+} + β^{-} x_{t}^{-} + u_{t}

(6)

where it can be regarded that the y-x equilibrium link is divided into two parts: positive (

β^{+} x_{t}^{+}

) and negative (

β^{-} x_{t}^{-}

) effects, plus the error term (

u_{t}

) representing possible deviations from the long-equilibrium held on deposit.

As shown in equation (6), the effect of the variable x can be decomposed into two parts, positive and negative:

x_{t} = x_{0} + x_{t}^{+} + x_{t}^{-}

(7)

where

x_{0}

represents the random initial value and

x_{t}^{+} + x_{t}^{-}

denote partial sum processes which accumulate positive and negative changes,⁹⁰ respectively, and are defined as below:

x_{t}^{+} = \sum_{j = 1}^{t} Δ x_{j}^{+} = \sum_{j = 1}^{t} max (Δ x_{j}, 0)

(8)

x_{t}^{-} = \sum_{j = 1}^{t} Δ x_{j}^{-} = \sum_{j = 1}^{t} min (Δ x_{j}, 0)

(9)

By combining equation (6) with the linear ARDL(p, q) model depicted in equation (5), the asymmetric error correction model is obtained as following:

\begin{aligned} Δ y_{t} = & μ + ρ y_{t - 1} + θ^{+} x_{t - 1}^{+} + θ^{-} x_{t - 1}^{-} + \sum_{j = 1}^{p - 1} α_{j} Δ y_{t - j} \\ + \sum_{j = 0}^{q - 1} (π_{j}^{+} Δ x_{t - j}^{+} + π_{j}^{-} Δ x_{t - j}^{-}) + ε_{t} \end{aligned}

(10)

The NARDL approach allows using a combination of variables I(0) and I(1). It is important to check the integration order of variables as the NARDL model is not valid for measures that are I(2).

To our knowledge, existing studies on the impact of COVID-19 on the energy market during the first two waves of COVID-19, did not address this type of asymmetric study of relationships, the novelty of the article constantly deepening this NARDL model that will allow the analysis of long-term relationships between selected variables.

The innovations of this research include the following aspects. First, this study measured the impact of the COVID-19 pandemic on the energy market during the first two pandemic waves. Second, this research integrated advanced econometric models to obtain detailed results on the long-term asymmetric relationships between COVID-19 variables and the energy market. Third, this study will help fill the gap in the literature and will be a focus for future energy market research during the COVID-19 pandemic.

Empirical results

Descriptive analysis, correlations and unit root test

Table 3 shows the descriptive analysis for the variables. All COVID-19 variables, as well as the majority of the commodities included, have positively skewed distributions. Hence, a positive skewness indicates an asymmetric distribution on the right, similar Güngör, Ertuğrul,⁵² Fasanya, Oyewole.⁸⁶ The most noteworthy feature is kurtosis, which measures the size of the extremes. A kurtosis higher than the value of 3 suggests that the returns of the indices show heavy tails than the normal distribution. Therefore, the probability of extreme returns is higher than the probability that they are below the normal distribution. This feature is titled leptokurtic or simply heavy tail. Except for the incidence rate owing to SARS-CoV-2 in the United States, all variables in our investigation had a kurtosis value larger than 3, indicating that they are distributed leptokurtically, consistent with prior studies.^52,103

Table 3.

Descriptive statistics of the variables.

Variables	Mean	Std. Dev.	Skewness	Kurtosis	Jarque-Bera	Jarque-Bera Prob.
CH_NC	45.84663	53.62013	2.539914	11.47407	662.9652	0
CH_ND	8.803681	100.9862	12.6434	160.9062	173,688.6	0
FR_NC	5997.736	10,232.19	2.85892	11.36543	697.328	0
FR_ND	150.1472	228.8304	2.86865	13.88885	1028.826	0
GR_NC	2333.589	3214.085	2.851985	11.84176	751.9185	0
GR_ND	48.17178	69.65866	1.965435	6.249394	176.6532	0
IT_NC	2864.552	5140.487	3.306718	14.27568	1160.55	0
IT_ND	149.2822	215.9165	1.630698	4.482463	87.167	0
SP_NC	7027.761	9031.022	2.502211	11.30124	638.1096	0
SP_ND	168.2209	292.1932	−1.18006	18.90003	1754.839	0
USA_NC	38,824.06	17,812.11	0.522062	2.592643	8.531239	0.014043
USA_ND	989.319	661.2234	1.903615	10.1744	448.0264	0
BRENT	0.002868	0.045226	−0.68317	13.7119	791.9874	0
CARBON	0.002882	0.033561	0.434799	3.770651	9.169455	0.010207
WTI	0.006431	0.062826	1.110102	10.96215	464.0414	0
ETHANOL	0.003815	0.030848	−0.57479	5.705675	58.69497	0
FUEL_OIL	0.000705	0.031099	−0.01076	7.290842	125.0467	0
GASOLINE	0.003215	0.058073	−1.70205	18.96426	1809.608	0
HEATING_OIL	0.001022	0.038904	−0.70427	7.428274	146.6567	0
KEROSENE	0.000226	0.036461	−2.38568	24.11044	3181.328	0
NATURAL_GAS	0.003961	0.045883	0.826096	5.224802	52.15647	0

Source: Authors’ own calculations. Notes: For the definition of variables, please see Table 2.

Furthermore, the Jarque-Bera findings, like earlier studies^{7,42,52,66,86} show that the selected series are not normally distributed, except for the newly diagnosed cases related to SARS-CoV-2 in the United States and the daily percentage change in the close price of Carbon Emissions Futures.

Figure 1 depicts the progression of new infections owing to COVID-19, whereas Figure 2 displays the expansion of new deaths attributed to COVID-19. During the studied timeframe, the United States, France, and Spain had the largest values in this respect.

Figure 1.

Daily infections’ advancement. Source: Authors’ own work. Notes: For the definition of variables, please see Table 2.

Figure 2.

Daily death toll expansion. Source: Authors’ own work. Notes: For the definition of variables, please see Table 2.

Figure 3 shows the evolution of the returns corresponding to the indicators that describe energy commodities. Periods of high volatility are quickly observed in March – April 2020, as well as in September – October 2020, when it coincides with the second lockdown. Moreover, Gasoline RBOB Futures, Crude Oil WTI Futures, Chukyo Kerosene Futures, and Natural Gas Futures record the most significant volatilities.

Figure 3.

Energy commodities progression. Source: Authors’ own work. Notes: For the definition of variables, please see Table 2.

From Figures 1–3 we may see a “volatility clustering” phenomenon, similar to Selmi, Bouoiyour,²⁴ as well as an oscillation amongst times of low volatility and periods of high volatility. A large autocorrelation of returns is also implied by volatility clustering.

Table 4 reveals the correlations among variables. There are acknowledged positive correlations between energy commodity indices. There are no strong association between the incidence rate and deaths caused by COVID-19 in China, the United States, Spain, France, Germany, and Italy, and the rest of variables.

Table 4.

Correlation matrix.

Variables	1	2	3	4	5	6	7	8	9	10	11
1 CH_NC	1
2 CH_ND	0.457	1
3 FR_NC	−0.131	−0.029	1
4 FR_ND	0.086	0.209	0.073	1
5 GR_NC	−0.006	0.027	0.852	0.321	1
6 GR_ND	0.036	0.284	−0.012	0.893	0.219	1
7 IT_NC	−0.056	0.014	0.901	0.254	0.948	0.166	1
8 IT_ND	0.150	0.145	−0.027	0.761	0.306	0.727	0.235	1
9 SP_NC	−0.058	−0.029	0.802	−0.027	0.589	−0.135	0.696	−0.062	1
10 SP_ND	0.122	0.117	0.097	0.599	0.305	0.554	0.232	0.702	0.121	1
11 USA_NC	0.256	−0.035	0.528	−0.210	0.412	−0.266	0.416	−0.430	0.377	−0.197	1
12 USA_ND	0.005	0.153	−0.155	0.592	−0.058	0.705	−0.098	0.294	−0.284	0.215	−0.064
13 BRENT	−0.047	0.014	−0.068	−0.018	−0.049	0.038	−0.053	0.038	−0.061	0.020	−0.083
14 CARBON	0.029	0.078	−0.071	0.088	−0.011	0.035	−0.023	0.105	−0.061	0.031	−0.105
15 WTI	−0.146	−0.112	−0.079	0.053	−0.074	0.166	−0.058	0.085	−0.088	0.037	−0.110
16 ETHANOL	−0.074	0.044	−0.012	0.013	−0.060	0.040	−0.029	−0.059	0.016	−0.011	−0.020
17 FUEL_OIL	−0.056	−0.059	−0.057	−0.092	−0.059	−0.080	−0.113	−0.034	−0.080	−0.042	−0.024
18 GASOLINE	−0.083	0.005	−0.056	0.043	−0.054	0.099	−0.054	0.060	−0.044	0.050	−0.058
19 HEATING_OIL	−0.040	0.021	−0.057	−0.107	−0.063	−0.087	−0.043	−0.067	−0.059	−0.114	−0.080
20 KEROSENE	−0.003	0.062	−0.051	−0.154	−0.076	−0.171	−0.071	−0.142	−0.046	−0.155	−0.030
21 NATURAL_GAS	0.067	0.061	−0.013	−0.029	0.014	−0.063	0.014	−0.024	0.001	−0.079	0.056
Variables	12	13	14	15	16	17	18	19	20	21
12 USA_ND	1
13 BRENT	0.001	1
14 CARBON	0.012	0.275	1
15 WTI	0.107	0.762	0.217	1
16 ETHANOL	0.037	0.156	0.207	0.172	1
17 FUEL_OIL	0.019	0.180	0.032	0.085	0.072	1
18 GASOLINE	0.049	0.576	0.343	0.484	0.327	0.048	1
19 HEATING_OIL	−0.107	0.777	0.304	0.594	0.179	0.090	0.541	1
20 KEROSENE	−0.079	0.091	0.078	−0.118	0.068	0.408	−0.069	0.137	1
21 NATURAL_GAS	−0.025	0.061	0.167	0.053	0.125	−0.101	0.093	0.145	−0.062	1

Source: Authors’ own calculations. Notes: For the definition of variables, please see Table 2.

The ADF stationarity test is used to check for the stationarity of the specified variables. This test is most typically used to confirm the pattern in which data series are integrated. It is imperative to notice that nonstationary variables lead to inappropriate results, which entails results that are not meaningful.

Table 5 shows the outcomes of the stationarity investigation.

Table 5.

The outcomes of the augmented Dickey-Fuller test.

Variables	Level Prob	1st Difference Prob.	Integration Order
CH-NC	0.1709	0	I(1)
CH-ND	0	—	I(0)
FR_NC	1	0.9987	I(2)
FR_ND	0.0612	0	I(1)
GR_NC	0.9272	0.9891	I(2)
GR-ND	0.1523	0	I(1)
IT-NC	0.2367	0.0199	I(1)
IT-ND	0	—	I(0)
USA-NC	0.8466	0.0243	I(1)
USA-ND	0.1418	0.0001	I(1)
SP-NC	0.9875	0.0004	I(1)
SP-ND	0.2982	0	I(1)
BRENT	0	—	I(0)
CARBON	0	—	I(0)
WTI	0	—	I(0)
ETHANOL	0.0024	—	I(0)
FUEL_OIL	0	—	I(0)
GASOLINE	0	—	I(0)
HEATING_OIL	0	—	I(0)
KEROSENE	0.0001	—	I(0)
NATURAL_GAS	0	—	I(0)

Source: Authors’ own calculations. Notes: Null Hypothesis: has a unit root. For the definition of variables, please see Table 2.

All energy commodity indexes are stationary at the level, hence they have an integration order of I(0), identical to Mensi, Sensoy.⁴² The indicators for COVID-19 evolution in the most affected regions, namely the United States, China, France, Germany, Spain, and Italy, reveal a mixed integration order (I(0), I(1), and I(2)).

In addition, the Q-Q graph gives a convenient method for comparing two distributions (see Figure 4). It depicts the relationship between an actual and theoretical distribution graph (normal distribution).

Figure 4.

Q-Q plot for variables that describe energy commodities. Source: Authors’ own work. Notes: For the definition of variables, please see Table 2.

If the empirical distribution is normal, the associated Q-Q graph should be the first bisector, although the distribution in our study is anything but normal, consistent with Selmi, Bouoiyour.²⁴ In addition, this fact is also reinforced by the graphical histogram (see Figure 5).

Figure 5.

Density of data series for variables that describe energy commodities. Source: Authors’ own work. Notes: For the definition of variables, please see Table 2.

Volatility behavior examination

Prior estimating GARCH models, several preliminary checks should be carried out to detect ARCH effects, as shown in Table 6. Autocorrelation (AC), partial autocorrelation (PAC), and the Q test were used to study heteroskedasticity. For all series data, a number of 20 lags have been used.

Table 6.

Autocorrelation (AC), partial autocorrelation (PAC) and Q statistics.

Variables	Lag	AC	PAC	Q-Stat	Prob
BRENT	20	−0.014	−0.056	50.6	0
CARBON	20	0.076	0.111	18.859	0.531
WTI	20	−0.016	−0.043	126	0
ETHANOL	20	0.019	−0.034	43.381	0.002
FUEL_OIL	20	0.053	0.103	27.482	0.122
GASOLINE	20	0.334	0.287	41.765	0.003
HEATING_OIL	20	0.021	−0.079	74.233	0
KEROSENE	20	0.007	−0.115	64.613	0
NATURAL_GAS	20	−0.035	−0.023	27.944	0.111

Source: Authors’ own calculations. Notes: For the definition of variables, please see Table 2.

The existence of the series’ correlation heteroskedasticity is confirmed by the outcomes of the Q test (the probability value being less than 5 percent, which consists in rejecting the null assumption). With the exception of the indices Carbon Emissions Futures and Fuel Oil Futures, the null hypothesis of no serial connection up to lag 20 can be rejected in this circumstance. As a consequence, the data series exhibit heteroskedasticity, which may be addressed using the GARCH model. Heteroskedasticity is a prerequisite for using GARCH models with financial time series.

The volatility of the main energy indices is shown in Figure 6. It was generated using the historical volatility series, based on the TGARCH(1,1) equation similar Güngör, Ertuğrul,⁵² Ertuğrul, Güngör.¹⁰³

Figure 6.

Volatility of indicators that capture energy commodities using an asymmetric TGARCH volatility model. Source: Authors’ own work. Notes: For the definition of variables, please see Table 2.

The beginning of 2020 was marked by growth, but due to the quick spread of COVID-19 globally, stock markets experienced significant episodes of volatility as a result of the effects of contagion on financial markets and forecasts of the global economic downturn. The outcomes support Maneejuk, Thongkairat⁶⁶ which found that the linkages among pandemic disturbances and oil markets are substantial in 2020. The high volatility of Brent oil and WTI was driven by lower prices. In the short term, COVID-19 had a significant impact on the functioning of the global energy market, a result also highlighted by Wang, Yang.¹² The second lockdown imposed in many countries has contributed to considerable uncertainty among investors. Consequently, the level of stress in the European financial system, according to the composite indicator calculated by the European Central Bank increased significantly in March 2020. In addition, in March 2020, trading conditions were significantly deteriorated due to the effects of COVID-19, but since the end of April there has been a return of international markets. However, since October, there has been a deterioration in conditions in the US and Eurozone financial markets, according to the indices calculated by Bloomberg. International stock markets reacted violently, with the highest volatility in March, when all the world's economies began to impose restrictions to reduce the spread of the virus, which negatively impacted the evolution of global stock market indices, also being influenced by investors’ sense of fear.

The second lockdown that began in September 2020 led to a new wave of panic and uncertainty among investors. Volatility also manifested itself, but in this case its magnitude was much smaller than at the beginning of 2020 when the first lockdown was “unknown” to everyone, especially since it was not acknowledged how the economic situation would evolve.

Outcomes of the NARDL estimation

After studying the stationary of the data series and due to the mixed outcomes, we assert that the ARDL model is the most adequate in uncovering the relationships among variables, particularly regarding how each specific energy commodity measure is influenced by the advancement of the reports related to the recent number of instances, as well as the new mortality rate attributable to SARS-CoV-2 in main global capital markets.

Nevertheless, the existence of asymmetry in financial markets has been argued by prior literature, and the ARDL approach does not consider the possible asymmetric link between variables. In other words, it implicitly assumes that the positive and negative variations of the explanatory variables have the same effect on the dependent variable. It was considered that this was not the appropriate method for investigating the relation between energy markets and COVID-19 since all these variables may have asymmetric ties.

Moreover, the NARDL model not only allows the detection of the existence of asymmetric effects that independent variables can have on the dependent variable, but also allows the testing of cointegration in a single framework of equations. Moreover, the NARDL model has several advantages compared to other commonly used cointegration techniques, such as its flexibility in terms of the order of integration of the variables involved (in the present study we have an order of integration such as I(0), I(1), and I(2)), the possibility of testing for hidden cointegration, avoiding the omission of any relationship that is not visible in a conventional linear framework, and a better performance in small samples.

Table 7 provides the outcomes of cointegration examination. The null hypothesis of no cointegration is rejected if the F- statistic is greater than the critical value of the upper limit. If the F-stat is less than the lower limit, the null hypothesis cannot be rejected, indicating that there is no cointegration at that level of confidence. Because we could reject the null hypothesis at 1% of the significance level, this finding suggests the existence of a long-term connection between energy indicators and the variables that define COVID-19. This outcome is concordant with Nyga-Łukaszewska and Aruga.⁵⁰

Table 7.

The results of the NARDL bounds test for the energy indices & COVID-19.

	Significance	I0 Bound	I1 Bound
	10%	2.45	3.52
	5%	2.86	4.01
	2.50%	3.25	4.49
	1%	3.74	5.06
BRENT	ETHANOL			HEATING_OIL
COVID-19 countries	F-stat	COVID-19 countries	F-stat	COVID-19 countries	F-stat
China	20.28748	China	27.11568	China	39.26083
France	21.52285	France	27.77183	France	29.33117
Germany	19.29097	Germany	27.86456	Germany	28.16412
Italy	20.65146	Italy	27.74679	Italy	18.65331
Spain	15.19041	Spain	27.62586	Spain	28.17716
USA	18.52978	USA	26.44492	USA	29.78095
WTI	FUEL_OIL	KEROSENE
COVID-19 countries	F-stat	COVID-19 countries	F-stat	COVID-19 countries	F-stat
China	23.06983	China	21.01698	China	11.03373
France	15.44308	France	22.8626	France	22.94222
Germany	19.14176	Germany	20.84626	Germany	9.646938
Italy	18.82517	Italy	19.29825	Italy	10.1344
Spain	8.780741	Spain	18.97299	Spain	9.554035
USA	19.68427	USA	19.34965	USA	29.1323
CARBON	GASOLINE			NATURAL_GAS
COVID-19 countries	F-stat	COVID-19 countries	F-stat	COVID-19 countries	F-stat
China	44.22136	China	42.1895	China	27.03402
France	38.72413	France	50.39329	France	25.21857
Germany	40.52925	Germany	42.33656	Germany	26.59801
Italy	39.38211	Italy	46.03537	Italy	22.88571
Spain	44.84697	Spain	33.35345	Spain	25.31298
USA	17.42637	USA	42.42165	USA	25.31298

Source: Authors’ own calculations. Notes: For the definition of variables, please see Table 2.

The study estimated 54 NARDL models, which is a substantial amount. The goal was to look at the relationships between energy measures, as well as metrics that track the pandemic. The presence of cointegration is confirmed in all 54 estimated NARDL models (see Table 7), with the F-stat much greater than the critical values of I(0) and I(1). As a result, the variables investigated are cointegrated and will move in sync over time. We will additionally focus at the asymmetries in the connection among energy variables and pandemic indicators throughout selected timeframe. Tables 8–12 reports the outcomes in this regard.

Table 8.

Long-term coefficients for the daily percentage change in brent oil futures and crude oil WTI futures close prices.

BRENT					WTI
Variables	Coeff	Std. Error	t-Stat	Prob.	Variables	Coeff	Std. Error	t-Stat	Prob.
China					China
CH_NC⁺	−3.7 × 10⁻⁵	0.000084	−0.44284	0.6585	CH_NC⁺	−8.1 × 10⁻⁵	0.000064	−1.27717	0.2036
CH_NC⁻	−1.5 × 10⁻⁵	0.000085	−0.17883	0.8583	CH_NC⁻	−4.8 × 10⁻⁵	0.000065	−0.74239	0.459
CH_ND⁺	−0.00319	0.004773	−0.66809	0.5051	CH_ND⁺	−3.7 × 10⁻⁵	0.000033	−1.11978	0.2646
CH_ND⁻	−0.0032	0.004766	−0.67061	0.5035	CH_ND⁻	−6.9 × 10⁻⁵	0.000033	−2.07318	0.0399
France					France
DFR_NC⁺	0.000001	0.000002	0.556525	0.5787	DFR_NC⁺	−1 × 10⁻⁶	0.000001	−0.65176	0.5156
DFR_NC⁻	0.000001	0.000002	0.621531	0.5352	DFR_NC⁻	−1 × 10⁻⁶	0.000001	−0.60961	0.5431
FR_ND⁺	0.000088	0.000066	1.32107	0.1886	FR_ND⁺	−9 × 10⁻⁶	0.000087	−0.09953	0.9209
FR_ND⁻	0.000082	0.000059	1.375918	0.171	FR_ND⁻	−0.00001	0.000077	−0.13383	0.8937
Germany					Germany
DGR_NC⁺	0.000033	0.000016	2.118155	0.0359	DGR_NC⁺	−4 × 10⁻⁶	0.000004	−0.98036	0.3286
DGR_NC⁻	0.000032	0.000016	2.070567	0.0402	DGR_NC⁻	−3 × 10⁻⁶	0.000004	−0.79653	0.427
GR_ND⁺	−0.00018	0.00023	−0.78436	0.4341	GR_ND⁺	−0.00019	0.00021	−0.92414	0.357
GR_ND⁻	−0.00013	0.000204	−0.63619	0.5257	GR_ND⁻	−0.00022	0.000194	−1.13228	0.2594
Spain					Spain
SP_NC⁺	0	0.000001	−0.06634	0.9472	SP_NC⁺	0	0.000001	−0.56687	0.5717
SP_NC⁻	0	0.000001	−0.02048	0.9837	SP_NC⁻	0	0.000001	−0.58061	0.5624
SP_ND⁺	0.000001	0.000019	0.027456	0.9781	SP_ND⁺	−1.7 × 10⁻⁵	0.000019	−0.85814	0.3922
SP_ND⁻	0.000001	0.000017	0.050923	0.9595	SP_ND⁻	−1.3 × 10⁻⁵	0.000018	−0.71666	0.4747
USA					USA
USA_NC⁺	0	0	−0.62944	0.53	USA_NC⁺	0	0	−0.80763	0.4206
USA_NC⁻	0	0	−0.48685	0.6271	USA_NC⁻	0	0	−0.4729	0.637
USA_ND⁺	−1.3 × 10⁻⁵	0.000012	−1.12324	0.2632	USA_ND⁺	0.000024	0.000012	1.980801	0.0495
USA_ND⁻	−1.4 × 10⁻⁵	0.000011	−1.23593	0.2185	USA_ND⁻	0.000022	0.000011	1.973475	0.0504
Italy					Italy
IT_NC⁺	0.000003	0.000005	0.767946	0.4437	IT_NC⁺	−4 × 10⁻⁶	0.000002	−1.65594	0.0999
IT_NC⁻	0.000004	0.000009	0.45654	0.6487	IT_NC⁻	−1.9 × 10⁻⁵	0.000007	−2.61825	0.0098
IT_ND⁺	0.000243	0.000154	1.581713	0.1158	IT_ND⁺	−0.00025	0.000142	−1.77326	0.0783
IT_ND⁻	0.000156	0.000098	1.592333	0.1134	IT_ND⁻	−5.2 × 10⁻⁵	0.000093	−0.56194	0.575

Source: Authors’ own calculations. Notes: For the definition of variables, please see Table 1.

Table 12.

Natural gas futures long-term coefficients for the daily percentage change in close price.

NATURAL_GAS
Variables	Coeff	Std. Error	t-Stat	Prob.
China
CH_NC⁺	0.000041	0.000059	0.691293	0.4905
CH_NC⁻	0.000028	0.000061	0.462137	0.6447
CH_ND⁺	0.000002	0.00003	0.056212	0.9552
CH_ND⁻	0.000008	0.000031	0.2496	0.8032
France
DFR_NC⁺	−1 × 10⁻⁶	0.000001	−1.27941	0.2027
DFR_NC⁻	−1 × 10⁻⁶	0.000001	−1.28302	0.2015
FR_ND⁺	−1.7 × 10⁻⁵	0.000022	−0.75252	0.4529
FR_ND⁻	−1.6 × 10⁻⁵	0.000018	−0.8505	0.3964
Germany
DGR_NC⁺	0	0.000003	0.141967	0.8873
DGR_NC⁻	0	0.000003	0.10884	0.9135
GR_ND⁺	0.000504	0.000191	2.640708	0.0092
GR_ND⁻	0.00049	0.000179	2.738315	0.0069
Spain
SP_NC⁺	0	0.000001	−0.37978	0.7046
SP_NC⁻	0	0.000001	−0.46827	0.6403
SP_ND⁺	−1.2 × 10⁻⁵	0.000014	−0.84962	0.3969
SP_ND⁻	−1.1 × 10⁻⁵	0.000013	−0.86284	0.3896
USA
USA_NC⁺	0	0	1.930449	0.0555
USA_NC⁻	0	0	1.679539	0.0951
USA_ND⁺	0.000004	0.000005	0.702171	0.4837
USA_ND⁻	0.000004	0.000005	0.8667	0.3875
Italy
IT_NC⁺	0.000001	0.000001	0.488465	0.6259
IT_NC⁻	0.000005	0.000004	1.117649	0.2655
IT_ND⁺	0.00011	0.00011	1.005387	0.3163
IT_ND⁻	0.000041	0.000068	0.60371	0.5469

Source: Authors’ own calculations. Notes: For the definition of variables, please see Table 2.

Nevertheless, equally positive and negative shocks to the number of additional SARS-CoV-2 infections in Germany influence energy use in the long-term (see Table 8). This finding implies that positive shocks involving additional SARS-CoV-2 infections in Germany affect Brent Oil Futures more than adverse shocks. Both positive and negative changes have a long-term positive effect on Brent Oil Futures evolution. Alike Adedeji, Ahmed,⁶⁷ this result is consistent with the oil market theory which suppose that oil demand factors positively influence oil price oscillation, but supply factors adversely strike oil price instability. However, if they are different, it can be determined by applying an asymmetry test, which practically tests whether the coefficients are equal or not. If they are equal, then there is no asymmetry, and if they are not, then there is evidence of asymmetry. To test for long-term asymmetry, we use the Wald test According to the results, the null hypothesis of equality is accepted as p-value is greater than 0.05. The Wald test suggests that the long-term impact of new cases attributable to COVID-19 in Germany on Brent Oil Futures is not asymmetric.

The coefficient of negative shock of new cases of death in China owing to COVID-19 is statistically significant at 1%, according to the predicted Crude Oil WTI Futures equation (see Table 8). This result indicates that the impact of the number of new deaths in China on the evolution of the WTI price is asymmetric in the long run, contrary to Adedeji, Ahmed⁶⁷ which claimed that the influence of COVID-19 infections is temporary and leave a slightest effect on economies. Furthermore, the findings contradict Maneejuk, Thongkairat⁶⁶ which claimed a symmetric effect of COVID-19 shocks on energy markets. In China, the estimated long-term coefficient for the negative shock of new fatalities caused by COVID-19 is negative. In line with Atri, Kouki,⁵⁶ the gravity of the pandemic urged the implementation of preventive measures (isolation, social distancing, travel suspension, stoppage of factories) which conducted to a decline in demand for oil. Moreover, long-term asymmetry relations between the number of new deaths caused by COVID-19 in the United States, as well as the number of new cases of COVID-19 in Italy and the expansion of the WTI price have been established. Only positive shocks in the number of new deaths due to COVID-19 impact the WTI in the United States, while negative shocks have no effect. In contrast to the previous model, only the negative shock of new COVID-19 instances has an impact on WTI in the model that includes COVID-19 variables from Italy. Specifically, a reduction in the number of new COVID-19 cases in Italy will lead to an increase in the WTI price in the long term. Consistent with Nyga-Łukaszewska and Aruga,⁵⁰ considering that crude oil exerts an essential role in energizing cars and jets, the lower number of individuals using vehicles and aircrafts following the surge in the COVID-19 occurrences diminished crude oil demand, adversely impacting crude oil prices. Also, Le, Le⁵⁴ argued that increased incertitude in the economy involved higher risks that detrimentally affected business operation, investors’ prospects, and corporate decisions towards investment and output plans. In turn, related oil value was lowered since petroleum is a vital input in the production process.

In terms of COVID-19's impact on future carbon emissions, Table 9 shows that in the long run, the number of additional fatalities caused by COVID-19 in the United States affects carbon emissions, with this conclusion being verified for both positive and negative shocks. However, there is no asymmetry since the coefficients’ values are equal. Additionally, the Wald test showed that there is no asymmetry. Besides, long-term asymmetric linkages in the number of reported deaths caused by COVID-19 in Italy have been found. Carbon Emissions are solely affected by positive shocks in the number of new fatalities owing to COVID-19. Therefore, the increase in the number of additional fatalities in Italy due to pandemic will lead to an increase in carbon emissions in the long run.

Table 9.

Coefficients for the daily percentage change in the close price of carbon emissions futures and ethanol futures over the long run.

CARBON					ETHANOL
Variables	Coeff	Std. Error	t-Stat	Prob.	Variables	Coeff	Std. Error	t-Stat	Prob.
China					China
CH_NC⁺	−2.8 × 10⁻⁵	0.000046	−0.62556	0.5326	CH_NC⁺	−5 × 10⁻⁶	0.000059	−0.08169	0.935
CH_NC⁻	−1.9 × 10⁻⁵	0.000047	−0.40518	0.6859	CH_NC⁻	−2 × 10⁻⁶	0.00006	−0.03225	0.9743
CH_ND⁺	0.00003	0.000023	1.308517	0.1927	CH_ND⁺	0.000066	0.000047	1.4143	0.1594
CH_ND⁻	0.000035	0.000024	1.493189	0.1375	CH_ND⁻	0.000063	0.000046	1.37204	0.1721
France					France
DFR_NC⁺	−1 × 10⁻⁶	0.000001	−0.71157	0.4779	DFR_NC⁺	0	0.000001	−0.21795	0.8278
DFR_NC⁻	−1 × 10⁻⁶	0.000001	−0.69583	0.4876	DFR_NC⁻	0	0.000001	−0.07956	0.9367
FR_ND⁺	0.000005	0.000031	0.17545	0.861	FR_ND⁺	0.000078	0.00004	1.956449	0.0523
FR_ND⁻	0.000007	0.000027	0.25861	0.7963	FR_ND⁻	0.000071	0.000036	1.982824	0.0492
Germany					Germany
DGR_NC⁺	−3 × 10⁻⁶	0.000003	−1.18484	0.238	DGR_NC⁺	−2 × 10⁻⁶	0.000002	−1.00501	0.3165
DGR_NC⁻	−3 × 10⁻⁶	0.000003	−1.14054	0.2559	DGR_NC⁻	−2 × 10⁻⁶	0.000002	−0.92787	0.3549
GR_ND⁺	0.000034	0.000057	0.594526	0.5531	GR_ND⁺	0.000066	0.000063	1.05889	0.2913
GR_ND⁻	0.000037	0.000047	0.793222	0.4289	GR_ND⁻	0.000054	0.000052	1.032362	0.3035
Spain					Spain
SP_NC⁺	0	0	0.47219	0.6375	SP_NC⁺	−1 × 10⁻⁶	0.000001	−1.04565	0.2974
SP_NC⁻	0	0	0.560263	0.5761	SP_NC⁻	−1 × 10⁻⁶	0.000001	−1.02222	0.3083
SP_ND⁺	0.000001	0.00001	0.091012	0.9276	SP_ND⁺	0.000004	0.000012	0.358841	0.7202
SP_ND⁻	0.000001	0.000009	0.151467	0.8798	SP_ND⁻	0.000004	0.000011	0.403253	0.6873
USA					USA
USA_NC⁺	0	0	0.144812	0.8851	USA_NC⁺	0	0	−0.41871	0.676
USA_NC⁻	0	0	−0.03794	0.9698	USA_NC⁻	0	0	−0.4194	0.6755
USA_ND⁺	−1.1 × 10⁻⁵	0.000005	−2.16801	0.0318	USA_ND⁺	−2 × 10⁻⁶	0.000006	−0.39798	0.6912
USA_ND⁻	−1.1 × 10⁻⁵	0.000005	−2.1521	0.0331	USA_ND⁻	−2 × 10⁻⁶	0.000006	−0.44061	0.6601
Italy					Italy
IT_NC⁺	0.000002	0.000002	1.140999	0.2557	IT_NC⁺	0	0.000001	−0.08077	0.9357
IT_NC⁻	0.000007	0.000004	1.579158	0.1164	IT_NC⁻	−2 × 10⁻⁶	0.000005	−0.4575	0.648
IT_ND⁺	0.000167	0.000082	2.047535	0.0424	IT_ND⁺	−1.8 × 10⁻⁵	0.0001	−0.18085	0.8567
IT_ND⁻	0.000081	0.000051	1.588928	0.1142	IT_ND⁻	0.000002	0.000062	0.039247	0.9687

Source: Authors’ own calculations. Notes: For the definition of variables, please see Table 2.

In regards of COVID-19's influence on Ethanol Futures (see Table 9), the results reveal that COVID-19 has no long-term impact on energy use. This assertion is valid for both positive and negative shocks. The coefficient of the negative shock of the number of new deaths attributable to COVID-19 in France is statistically significant, according to the predicted Fuel Oil Futures equation (see Table 10). Hence, the long-term impact of the number of new deaths in France related to SARS-CoV-2 on Fuel Oil Futures is asymmetric. According to Gharib, Mefteh-Wali,⁶⁴ the pandemic engendered a decline in worldwide manufacturing, thus triggering smaller oil consumption which diminished the barrel price. As well, Bourghelle, Jawadi⁶⁵ argued that the oil sector was affected by decreased worldwide demand for crude oil due to heightened insecurity, as well as by oil deal conflict among the main oil-producing countries. This finding entails that decreases in the number of new deaths caused by COVID-19 in France lead in a rise in the price of Fuel Oil Futures in the long run. Also, the results of Table 10 estimations demonstrate that the impact of new SARS-CoV-2 mortality in France on Gasoline RBOB Futures is asymmetric in the long run. Thus, only negative shocks from new COVID-19 fatalities in France affect Gasoline RBOB Futures, whereas positive shocks have no effect. This result is consistent with Christopoulos, Kalantonis⁶⁰ which found that that oil instability is influenced by COVID-19 mortalities, emphasizing that pandemic is a recent factor that has deepened the market risk. Also, the outcomes support Algamdi, Brika⁵⁵ which reported that pandemic deaths cases adversely influence oil price changes.

Table 10.

Coefficients for the daily percentage change in the close price of fuel oil futures and gasoline RBOB futures over the long run.

FUEL_OIL					GASOLINE
Variables	Coeff	Std. Error	t-Stat	Prob.	Variables	Coeff	Std. Error	t-Stat	Prob.
China					China
CH_NC⁺	−2.5 × 10⁻⁵	0.000047	−0.53254	0.5951	CH_NC⁺	−3.7 × 10⁻⁵	0.000066	−0.5556	0.5793
CH_NC⁻	−1.7 × 10⁻⁵	0.000048	−0.35016	0.7267	CH_NC⁻	−1.4 × 10⁻⁵	0.000068	−0.20385	0.8387
CH_ND⁺	−1.1 × 10⁻⁵	0.000024	−0.47115	0.6382	CH_ND⁺	−7 × 10⁻⁶	0.000034	−0.19503	0.8456
CH_ND⁻	−1.6 × 10⁻⁵	0.000025	−0.64517	0.5198	CH_ND⁻	−1 × 10⁻⁶	0.000035	−0.03849	0.9693
France					France
DFR_NC⁺	0	0.000001	−0.472229	0.6375	DFR_NC⁺	−1 × 10⁻⁶	0.000001	−1.7184	0.0878
DFR_NC⁻	0	0.000001	−0.388651	0.6981	DFR_NC⁻	−1 × 10⁻⁶	0.000001	−1.62082	0.1072
FR_ND⁺	−0.000062	0.000033	−1.868.368	0.0637	FR_ND⁺	0.000089	0.000046	1.903462	0.0589
FR_ND⁻	−0.000059	0.00003	−2.001.338	0.0472	FR_ND⁻	0.00009	0.000041	2.174833	0.0312
Germany					Germany
DGR_NC⁺	0.000003	0.000005	0.651043	0.5161	DGR_NC⁺	−1 × 10⁻⁶	0.000003	−0.17023	0.8651
DGR_NC⁻	0.000004	0.000005	0.877605	0.3816	DGR_NC⁻	0	0.000003	−0.07778	0.9381
GR_ND⁺	−4.9 × 10⁻⁵	0.000079	−0.61372	0.5404	GR_ND⁺	0.00001	0.000167	0.058462	0.9535
GR_ND⁻	−6.8 × 10⁻⁵	0.000066	−1.04034	0.2999	GR_ND⁻	0.000018	0.000152	0.117341	0.9068
Spain					Spain
SP_NC⁺	0	0	−0.36837	0.7131	SP_NC⁺	0	0.000001	−0.02643	0.9789
SP_NC⁻	0	0	−0.23966	0.8109	SP_NC⁻	0	0.000001	0.032762	0.9739
SP_ND⁺	−1 × 10⁻⁶	0.000011	−0.06309	0.9498	SP_ND⁺	0.000009	0.000021	0.45203	0.6519
SP_ND⁻	−2 × 10⁻⁶	0.00001	−0.18672	0.8521	SP_ND⁻	0.000009	0.000019	0.497977	0.6192
USA					USA
USA_NC⁺	0	0	−1.72635	0.0864	USA_NC⁺	0	0	−0.51985	0.604
USA_NC⁻	0	0	−0.98813	0.3248	USA_NC⁻	0	0	−0.60138	0.5485
USA_ND⁺	0.000002	0.000007	0.228823	0.8193	USA_ND⁺	−2 × 10⁻⁶	0.000009	−0.26145	0.7941
USA_ND⁻	0	0.000007	−0.02957	0.9764	USA_ND⁻	−1 × 10⁻⁶	0.000008	−0.16258	0.8711
Italy					Italy
IT_NC⁺	−2 × 10⁻⁶	0.000001	−1.712	0.0889	IT_NC⁺	−1 × 10⁻⁶	0.000002	−0.59046	0.5558
IT_NC⁻	−2 × 10⁻⁶	0.000003	−0.50367	0.6152	IT_NC⁻	−0.00001	0.000006	−1.5686	0.119
IT_ND⁺	0.000139	0.000076	1.829364	0.0693	IT_ND⁺	−0.0003	0.00017	−1.75949	0.0806
IT_ND⁻	0.00009	0.000049	1.827511	0.0696	IT_ND⁻	−0.00012	0.000104	−1.15673	0.2493

Source: Authors’ own calculations. Notes: For the definition of variables, please see Table 2.

The impact of additional fatalities in Italy owing to COVID-19 on Heating Oil Futures indicates that new deaths affect energy use in the long run (see Table 11). This applies to both positive and negative shocks. The outcome is consistent with Khan, Su,¹⁹ being argued by great incertitude at the commencement of the outbreak, acute lockdown and international moving constraints which dropped the energy demand. As well, the findings of Dmytrów, Landmesser⁶¹ are reinforced since heating oil is poorly related with COVID-19. This finding means that positive shocks from new fatalities in Italy because of COVID-19 have a greater influence on Heating Oil Futures than negative shocks. As well, a positive shock in the number of new SARS-CoV-2 cases in Italy has been found. Table 11 further shows that the coefficient of the positive shock of the number of new cases owing to COVID-19 in the United States is statistically significant, as determined by the estimated Chukyo Kerosene Futures equation. This result supports that the impact of the number of new instances due to COVID-19 in the United States on Chukyo Kerosene Futures is asymmetric in the long run. As such, an increase in the number of new cases of SARS-CoV-2 in the United States leads to an increase in Chukyo Kerosene Futures prices in the long run.

Table 11.

Heating oil futures and Chukyo Kerosene futures long-term coefficients for the daily percentage change in close price.

HEATING_OIL					KEROSENE
Variables	Coeff	Std. Error	t-Stat	Prob.	Variables	Coeff	Std. Error	t-Stat	Prob.
China					China
CH_NC⁺	−3.8 × 10⁻⁵	0.000061	−0.61995	0.5362	CH_NC⁺	−1.2 × 10⁻⁵	0.000034	−0.34565	0.7301
CH_NC⁻	−2.3 × 10⁻⁵	0.000063	−0.37214	0.7103	CH_NC⁻	0	0.000035	0.00342	0.9973
CH_ND⁺	0.000188	0.003551	0.052935	0.9579	CH_ND⁺	0.000025	0.000018	1.368767	0.1732
CH_ND⁻	0.000172	0.003547	0.048462	0.9614	CH_ND⁻	0.000009	0.000019	0.475812	0.6349
France					France
DFR_NC⁺	0.000001	0.000001	0.549194	0.5837	DFR_NC⁺	0	0	−0.85355	0.3948
DFR_NC⁻	0.000001	0.000002	0.632176	0.5283	DFR_NC⁻	0	0	−0.59242	0.5545
FR_ND⁺	−9 × 10⁻⁶	0.000052	−0.17132	0.8642	FR_ND⁺	−1.8 × 10⁻⁵	0.000031	−0.59833	0.5506
FR_ND⁻	−1.3 × 10⁻⁵	0.000047	−0.27167	0.7863	FR_ND⁻	−2.3 × 10⁻⁵	0.000028	−0.83005	0.4079
Germany					Germany
DGR_NC⁺	−1 × 10⁻⁶	0.000003	−0.18002	0.8574	DGR_NC⁺	0.000001	0.000002	0.401367	0.6887
DGR_NC⁻	0	0.000003	−0.02452	0.9805	DGR_NC⁻	0.000001	0.000002	0.660388	0.5101
GR_ND⁺	0.000167	0.000122	1.366826	0.1738	GR_ND⁺	−4.5 × 10⁻⁵	0.000068	−0.66544	0.5068
GR_ND⁻	0.000138	0.000108	1.281159	0.2022	GR_ND⁻	−6.5 × 10⁻⁵	0.000062	−1.04866	0.2961
Spain					Spain
SP_NC⁺	0	0.000001	−0.51394	0.608	SP_NC⁺	0	0	−0.03392	0.973
SP_NC⁻	0	0.000001	−0.51502	0.6073	SP_NC⁻	0	0	−0.02119	0.9831
SP_ND⁺	−1.8 × 10⁻⁵	0.000015	−1.18094	0.2395	SP_ND⁺	−2.8 × 10⁻⁵	0.000013	−2.21625	0.0282
SP_ND⁻	−1.7 × 10⁻⁵	0.000014	−1.26029	0.2095	SP_ND⁻	−2.7 × 10⁻⁵	0.000011	−2.37938	0.0186
USA					USA
USA_NC⁺	0	0	−1.44588	0.1503	USA_NC⁺	0	0	−2.2296	0.0273
USA_NC⁻	0	0	−0.95761	0.3398	USA_NC⁻	0	0	−0.93052	0.3537
USA_ND⁺	0	0.000006	−0.02717	0.9784	USA_ND⁺	0.000002	0.000005	0.34434	0.7311
USA_ND⁻	−2 × 10⁻⁶	0.000005	−0.40616	0.6852	USA_ND⁻	−1 × 10⁻⁶	0.000005	−0.13706	0.8912
Italy					Italy
IT_NC⁺	0.000006	0.000003	2.013406	0.046	IT_NC⁺	0	0.000001	0.129901	0.8968
IT_NC⁻	0.000004	0.000005	0.734586	0.4638	IT_NC⁻	0.000001	0.000003	0.524356	0.6008
IT_ND⁺	−0.00036	0.000142	−2.49551	0.0137	IT_ND⁺	−1.6 × 10⁻⁵	0.000075	−0.20608	0.837
IT_ND⁻	−0.00024	0.000089	−2.64481	0.0091	IT_ND⁻	−3.3 × 10⁻⁵	0.000047	−0.7032	0.483

Source: Authors’ own calculations. Notes: For the definition of variables, please see Table 1.

The coefficient of the positive and negative shock of the number of new deaths attributable to COVID-19 in Germany is statistically significant in the case of the estimated equation for Natural Gas Futures (see Table 12). However, there is no imbalance because the coefficients’ values are equal. Additionally, according to the Wald test, there is no asymmetry. This outcome confirms Maneejuk, Thongkairat⁶⁶ which emphasize the symmetric influence of COVID-19 shocks on energy markets.

Conclusions

The global economic context was and is marked by uncertainty, the need to adapt to new conditions and structural changes that affected all segments of today's society remained the essential attributes of the current reality, representing both challenges and opportunities for both business and scientific research.

The goal of current research was to investigate if the initial wave of the COVID-19 pandemic has an impact on energy commodities. We examined daily data for the following markets, from February 28, 2020, to November 3, 2020: the United States, China, Spain, Italy, France, and Germany. Brent Oil Futures, Carbon Emissions Futures, Crude Oil WTI Futures, Ethanol Futures, Fuel Oil Futures, Gasoline RBOB Futures, Heating Oil Futures, Chukyo Kerosene Futures, and Natural Gas Futures, as well as the new number of cases of illness and deaths due to SARS-CoV-2 in major international financial markets have been included to achieve our purpose.

We concluded that the NARDL model is the most appropriate in exploring the cointegration between variables after studying the stationary of the data series and due to the mixed results. However, we focused on the asymmetrical aspects of the existing relationships. The NARDL model was used to find out if there was any imbalance in the long-term relationships between the variables. Pesaran, Shin¹⁰¹ set up the NARDL cointegration model, which is an asymmetric extension of the autoregressive distributed lag (ARDL) cointegration model. Negative and positive variations of independent variables may have divergent effects on the dependent variable, fact which is not considered in the linear ARDL approach.

The whole variables selected as proxies for the energy sector, apart from Ethanol Futures, have long-term connections with COVID-19 variables. According to the actual results of the NARDL model, the energy market has an asymmetric long-term connection with the novel coronavirus related variables in the vast majority of cases.

WTI showed the strongest long-term asymmetrical connections with COVID-19 variables of all the indicators chosen to examine energy commodities. The estimated Crude Oil WTI Futures equations revealed that the number of new deaths caused by pandemic in China has an asymmetric impact on the evolution of the WTI price in the long term. This finding means that a decrease in COVID-19 deaths in China will result in an increase in WTI in the long run. In addition, long-term asymmetry links between the number of new deaths in the United States and the number of new cases in Italy and the evolution of the WTI price have been found. Besides, only positive shocks in the number of new deaths affect the WTI in the United States, while negative shocks have no effect. In the model that includes COVID-19 variables from Italy, merely the negative shock of new pandemic occurrences has an impact on WTI, specifically a decrease in the number of new COVID-19 cases in Italy will result in a long-term increase in the WTI price.

Accordingly, the NARDL model empirical findings revealed the existence of an asymmetric long-term association between energy commodities and COVID-19 variables in China, the United States, France, Spain, Italy, and Germany.

As previously mentioned, the novelty of this scientific paper consists both in the chosen topic of study: the impact of COVID-19 on the energy sector during the first two waves of COVID-19, and by the statistical methods applied, the NARDL model that allows capturing asymmetric reactions in term long.

Further research on this topic in the near future is imminent due to the ongoing changes in the pandemic, as well as the availability of longer data on COVID-19. These findings should be taken into account by investors and policy makers, as they argue that the relationship between energy goods and COVID-19 is dynamic rather than linear.

Unlike 2020, when the initial shock was strong, determined by the novelty of this type of disturbance, 2021 assimilated the information and progress made to remedy the health situation, with the advent of vaccines and the development of management strategies improving the current situation, with a focus on both population health and economic support and, implicitly, on avoiding destabilization or crises. The start of future research that would include in the analysis the period of 2021 will provide an added value to the existing specialized literature.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Jean Vasile Andrei

Camelia Catalina Joldes

References

Harjoto

Rossi

Lee

, et al. How do equity markets react to COVID-19? Evidence from emerging and developed countries. J Econ Bus 2021. 115: 105966. DOI: 10.1016/j.jeconbus.2020.105966

Salisu

Ebuh

Usman

. Revisiting oil-stock nexus during COVID-19 pandemic: some preliminary results. Int Rev Econ Financ 2020; 69: 280–294.

Zhang

. Financial markets under the global pandemic of COVID-19. Financ Res Lett 2020; 36. DOI: 10.1016/j.frl.2020.101528.

Mazur

Dang

Vega

. COVID-19 and the march 2020 stock market crash. Evidence from S&P1500. Financ Res Lett 2020; 38: 101690. DOI: 10.1016/j.frl.2020.101690.

Smales

. Investor attention and global market returns during the COVID-19 crisis. Int Rev Financ Anal 2021; 73. DOI: 10.1016/j.irfa.2020.101616.

Bouri

Cepni

Gabauer

, et al. Return connectedness across asset classes around the COVID-19 outbreak. Int Rev Financ Anal 2020; 73: 101646. DOI: 10.1016/j.irfa.2020.101646.

Ezeaku

Asongu

Nnanna

. Volatility of international commodity prices in times of COVID-19: effects of oil supply and global demand shocks. Extr Ind Soc 2021; 8(1): 257–270. DOI: 10.1016/j.exis.2020.12.013.

World Bank. Commodity markets outlook. 2020.

Nwosa

. Oil price, exchange rate and stock market performance during the COVID-19 pandemic: implications for TNCs and FDI inflow in Nigeria. Trans Corpor Rev 2020; 13(1): 125–137. DOI: 10.1080/19186444.2020.1855957.

10.

Adekoya

Oliyide

. How COVID-19 drives connectedness among commodity and financial markets: evidence from TVP-VAR and causality-in-quantiles techniques. Resour Policy 2021; 70: 101898. DOI: 10.1016/j.resourpol.2020.101898.

11.

Gharib

Mefteh-Wali

Jabeur

. The bubble contagion effect of COVID-19 outbreak: evidence from crude oil and gold markets. Financ Res Lett 2021; 38: 101703.

12.

Wang

Yang

. The impact of the COVID-19 pandemic on the energy market – A comparative relationship between oil and coal. Energy Strategy Rev 2022; 39: 100761.

13.

Amamou

Bargaoui

. Energy markets responds to COVID-19 pandemic. Resour Policy 2022; 76: 102551.

14.

Chofreh

Goni

Klemeš

, et al. COVID-19 shock: development of strategic management framework for global energy. Renewable Sustainable Energy Rev 2021; 139: 110643. DOI: 10.1016/j.rser.2020.110643.

15.

Elavarasan

Shafiullah

Raju

, et al. COVID-19: impact analysis and recommendations for power sector operation. Appl Energ 2020; 279: 115739.

16.

Şahin

Ballı

Chen

. Forecasting seasonal electricity generation in European countries under COVID-19-induced lockdown using fractional grey prediction models and machine learning methods. Appl Energ 2021; 302: 117540.

17.

Wang

Jiang

. Uncovering the impact of the COVID-19 pandemic on energy consumption: new insight from difference between pandemic-free scenario and actual electricity consumption in China. J Clean Prod 2021; 313: 127897.

18.

Wang

Zhang

, et al. Impact of COVID-19 pandemic on oil consumption in the United States: a new estimation approach. Energy 2022; 239: 122280.

19.

Khan

C-W

Zhu

. Examining the behaviour of energy prices to COVID-19 uncertainty: a quantile on quantile approach. Energy 2021; 239: 122430. DOI: 10.1016/j.energy.2021.122430.

20.

Wang

, et al. Forecasting the U.S. oil markets based on social media information during the COVID-19 pandemic. Energy 2021; 226: 120403.

21.

Norouzi

Rubens

GZd

Choupanpiesheh

, et al. When pandemics impact economies and climate change: exploring the impacts of COVID-19 on oil and electricity demand in China. Energy Res Soc Sci 2020; 68: 101654.

22.

Qin

Zhang

Y-C

C-W

. The essential role of pandemics: a fresh insight into the oil market. Energy Res Lett 2020; 1. DOI: 10.46557/001c.13166.

23.

Kingsly

Henri

. COVID-19 and oil prices. 18 Mar 2020 2020.

24.

Selmi

Bouoiyour

Hammoudeh

, et al. The energy transition, trump energy agenda and COVID-19. Int Econ 2021; 165. DOI: 10.1016/j.inteco.2020.12.010.

25.

Brosemer

Schelly

Gagnon

, et al. The energy crises revealed by COVID: intersections of indigeneity, inequity, and health. Energy Res Soc Sci 2020; 68. DOI: 10.1016/j.erss.2020.101661.

26.

Guo

. Tail risk contagion between international financial markets during COVID-19 pandemic. Int Rev Financ Anal 2021; 73. DOI: 10.1016/j.irfa.2020.101649.

27.

IEA. Exploring the impacts of the COVID-19 pandemic on global energy markets, energy resilience, and climate change. 2020.

28.

IEA. Global energy review 2020. 2020.

29.

OECD. The impact of coronavirus (COVID-19) and the global oil price shock on the fiscal position of oil-exporting developing countries. 2020.

30.

Tahir

Batool

. COVID-19: healthy environmental impact for public safety and menaces oil market. Sci Total Environ 2020; 740: 140054.

31.

Wang

Zhang

. What does the China’s economic recovery after COVID-19 pandemic mean for the economic growth and energy consumption of other countries? J Clean Prod 2021; 295: 126265.

32.

Global SP. How is COVID-19 impacting the energy transition? 2020.

33.

World Bank. Impact of COVID-19 on commodity markets heaviest on energy prices; lower oil demand likely to persist beyond 2021. 2020.

34.

Gil-Alana

Monge

. Crude Oil prices and COVID-19: persistence of the shock. Energy Res Lett 2020; 1. DOI: 10.46557/001c.13200.

35.

Okorie

Lin

. Stock markets and the COVID-19 fractal contagion effects. Financ Res Lett 2021; 38: 101640.

36.

Szczygielski

Bwanya

Charteris

, et al. The only certainty is uncertainty: an analysis of the impact of COVID-19 uncertainty on regional stock markets. Financ Res Lett 2021; 43: 101945. DOI: 10.1016/j.frl.2021.101945.

37.

Seven

Yılmaz

. World equity markets and COVID-19: immediate response and recovery prospects. Res Int Bus Fin 2021; 56: 101349.

38.

Wang

. A preliminary assessment of the impact of COVID-19 on environment – A case study of China. Sci Total Environ 2020; 728. DOI: 10.1016/j.scitotenv.2020.138915.

39.

Wang

, et al. Underestimated impact of the COVID-19 on carbon emission reduction in developing countries – A novel assessment based on scenario analysis. Environ Res 2022; 204: 111990.

40.

Wang

Jiang

X-t

. Preventing a rebound in carbon intensity post-COVID-19 – lessons learned from the change in carbon intensity before and after the 2008 financial crisis. Sustain Prod Consump 2021; 27: 1841–1856.

41.

Jiang

Chen

. The time-frequency connectedness among carbon, traditional/new energy and material markets of China in pre- and post-COVID-19 outbreak periods. Energy 2022; 246. DOI: 10.1016/j.energy.2022.123320.

42.

Mensi

Sensoy

, et al. Impact of COVID-19 outbreak on asymmetric multifractality of gold and oil prices. Resour Policy 2020; 69: 101829.

43.

Zhang

Hamori

. Crude oil market and stock markets during the COVID-19 pandemic: evidence from the US, Japan, and Germany. Int Rev Financ Anal 2021; 74: 101702.

44.

Albulescu

. Coronavirus and oil price crash. 2020.

45.

Aloui

Goutte

Guesmi

, et al. COVID 19’s Impact on crude oil and natural gas S&P GS indexes. 2020.

46.

Devpura

Narayan

. Hourly oil price volatility: the role of COVID-19. Energy Res Lett 2020; 1. DOI: 10.46557/001c.13683.

47.

Narayan

. Oil price news and COVID-19—is there any connection? Energy Res Lett 2020; 1. DOI: 10.46557/001c.13176.

48.

Liu

Wang

E-Z

Lee

C-C

. Impact of the COVID-19 pandemic on the crude oil and stock markets in the US: a time-varying analysis. Energy Res Lett 2020; 1. DOI: 10.46557/001c.13154.

49.

Yilmazkuday

. Coronavirus disease 2019 and the global economy. June 26, 2020. 2020.

50.

Nyga-Łukaszewska

Aruga

. Energy prices and COVID-immunity: the case of crude oil and natural gas prices in the US and Japan. Energies 2020; 13: 6300.

51.

Kamdem

Essomba

Berinyuy

. Deep learning models for forecasting and analyzing the implications of COVID-19 spread on some commodities markets volatilities. Chaos Solitons Fractals 2020; 140: 110215.

52.

Güngör

Ertuğrul

Soytaş

. Impact of COVID-19 outbreak on Turkish gasoline consumption. Technol Forecast Soc 2021; 166: 120637. DOI: 10.1016/j.techfore.2021.120637.

53.

Kartal

. The effect of the COVID-19 pandemic on oil prices: evidence from Turkey. Energy Res Lett 2020; 1. DOI: 10.46557/001c.18723.

54.

T-H

H-C

. The historic oil price fluctuation during the COVID-19 pandemic: what are the causes? Res Int Bus Fin 2021; 58: 101489.

55.

Algamdi

Brika

SKM

Musa

, et al. COVID-19 deaths cases impact on oil prices: probable scenarios on Saudi Arabia economy. Front Public Health 2021; 9: 620875.

56.

Atri

Kouki

Gallali

. The impact of COVID-19 news, panic and media coverage on the oil and gold prices: an ARDL approach. Resour Policy 2021; 72: 102061.

57.

Atukeren

Çevik

Eİ

Korkmaz

. Volatility spillovers between WTI and brent spot crude oil prices: an analysis of granger causality in variance patterns over time. Res Int Bus Fin 2021; 56: 101385.

58.

Adekoya

Oliyide

Oduyemi

. How COVID-19 upturns the hedging potentials of gold against oil and stock markets risks: nonlinear evidences through threshold regression and markov-regime switching models. Resour Policy 2020; Available online 18 November 2020, 2021; 70: 101926. DOI: 10.1016/j.resourpol.2020.101926.

59.

Salisu

Adediran

. Uncertainty due to infectious diseases and energy market volatility. Energy Res Lett 2020; 1. DOI: 10.46557/001c.14185.

60.

Christopoulos

Kalantonis

Katsampoxakis

, et al. COVID-19 and the energy price volatility. Energies 2021; 14: 6496.

61.

Dmytrów

Landmesser

Bieszk-Stolorz

. The connections between COVID-19 and the energy commodities prices: evidence through the dynamic time warping method. Energies 2021; 14: 4024.

62.

D-K

X-L

, et al. The risk spillover effect of the COVID-19 pandemic on energy sector: evidence from China. Energ Econ 2021; 102: 105498.

63.

Hemrit

Benlagha

. Does renewable energy index respond to the pandemic uncertainty? Renew Energ 2021; 177: 336–347.

64.

Gharib

Mefteh-Wali

Serret

, et al. Impact of COVID-19 pandemic on crude oil prices: evidence from econophysics approach. Resour Policy 2021; 74: 102392.

65.

Bourghelle

Jawadi

Rozin

. Oil price volatility in the context of COVID-19. Int Econ 2021; 167: 39–49.

66.

Maneejuk

Thongkairat

Srichaikul

. Time-varying co-movement analysis between COVID-19 shocks and the energy markets using the Markov switching dynamic copula approach. Energy Rep 2021; 7: 81–88.

67.

Adedeji

Ahmed

Adam

. Examining the dynamic effect of COVID-19 pandemic on dwindling oil prices using structural vector autoregressive model. Energ Environ-Uk 2021; 230: 120813.

68.

Weng

Zhang

Yang

. Volatility forecasting of crude oil futures based on a genetic algorithm regularization online extreme learning machine with a forgetting factor: the role of news during the COVID-19 pandemic. Resour Policy 2021; 73: 102148.

69.

Niu

Liu

Gao

, et al. The role of coronavirus news in the volatility forecasting of crude oil futures markets: evidence from China. Resour Policy 2021; 73: 102173.

70.

Kamaludin

Sundarasen

Brahim

. COVID-19, Dow Jones and equity market movement in ASEAN-5 countries: evidence from wavelet analyses. Helyion 2021; 7: e05851.

71.

Topcu

Gulal

. The impact of COVID-19 on emerging stock markets. Financ Res Lett 2020; 36: 101691.

72.

Prabheesh

Padhan

Garg

. COVID-19 and the oil price – stock market nexus: evidence from net oil-importing countries. Energy Res Lett 2020; 1. DOI: 10.46557/001c.13745.

73.

Dutta

Das

Jana

, et al. COVID-19 and oil market crash: revisiting the safe haven property of gold and bitcoin. Resour Policy 2020; 69: 101816.

74.

Haroon

Rizvi

SAR

. COVID-19: media coverage and financial markets behavior—A sectoral inquiry. J Behav Exp Finance 2020; 27. DOI: 10.1016/j.jbef.2020.100343.

75.

Kwan

Mertens

. Market assessment of COVID-19. FRBSF Econ Lett 2020; 14: 1–5.

76.

Sharif

Aloui

Yarovaya

. COVID-19 pandemic, oil prices, stock market, geopolitical risk and policy uncertainty nexus in the US economy: fresh evidence from the wavelet-based approach. Int Rev Financ Anal 2020; 70: 101496.

77.

Ahundjanov

Akhundjanov

Okhunjanov

. Risk perception and oil and gasoline markets under COVID-19. J Econ Bus 2021; 15: 105979. DOI: 10.1016/j.jeconbus.2020.105979.

78.

Meher

Hawaldar

Mohapatra

, et al. The impact of COVID-19 on price volatility of crude oil and natural gas listed on multi commodity exchange of India. Int J Energy Econ Policy 2020; 10: 422–431.

79.

Sakurai

Kurosaki

. How has the relationship between oil and the US stock market changed after the COVID-19 crisis? Financ Res Lett 2020; 37: 101773.

80.

Salisu

Akanni

Raheem

. The COVID-19 global fear index and the predictability of commodity price returns. J Behav Exp Finance 2020; 27: 100383.

81.

Akhtaruzzaman

Boubaker

Chiah

, et al. COVID−19 and oil price risk exposure. Financ Res Lett 2020; 2021; 42: 101882. DOI: 10.1016/j.frl.2020.101882.

82.

Borgards

Czudaj

VanHoang

. Price overreactions in the commodity futures market: an intraday analysis of the COVID-19 pandemic impact. Resourc Policy 2021; 71: 101966.

83.

Yang

Wei

, et al. Global financial uncertainties and China’s crude oil futures market: evidence from interday and intraday price dynamics. Energ Econ 2021; 96: 105149. DOI: 10.1016/j.eneco.2021.105149.

84.

Yang

Y-R

, et al. Extreme risk spillover between Chinese and global crude oil futures. Financ Res Lett 2020; Available online 30 August 2020, 101743. DOI: 10.1016/j.frl.2020.101743. Online ahead of print.

85.

Lahmiri

Bekiros

. Renyi entropy and mutual information measurement of market expectations and investor fear during the COVID-19 pandemic. Chaos Solitons Fractals 2020; 139: 110084.

86.

Fasanya

Oyewole

Odei-Mensah

. Infectious diseases-energy futures nexus: a quantile-on-quantile approach. Energy Res Lett 2021; 1. DOI: 10.46557/001c.18267.

87.

Bouri

Lucey

Saeed

, et al. The realized volatility of commodity futures: interconnectedness and determinants. Int Rev Econ Financ 2021; 73: 139–151.

88.

Foglia

Angelini

. Volatility connectedness between clean energy firms and crude oil in the COVID-19 era. Sustainability-Basel 2020; 12: 9863.

89.

Mensi

Al-Yahyaee

, et al. Dynamic spillover and connectedness between oil futures and European bonds. North Am J Econ Financ 2021; 56: 101342. DOI: 10.1016/j.najef.2020.101342.

90.

Urom

Ndubuisi

Ozor

. Economic activity, and financial and commodity markets’ shocks: an analysis of implied volatility indexes. Int Econ 2021; 165: 51–66.

91.

Gunay

. Comparing COVID-19 with the GFC: a shockwave analysis of currency markets. Res Int Bus Financ 2021; 56: 101377.

92.

Shehzad

Xiaoxing

Kazouz

. COVID-19’s disasters are perilous than global financial crisis: a rumor or fact? Financ Res Lett 2020; 36: 101669.

93.

Corbet

Hou

, et al. Pandemic-related financial market volatility spillovers: evidence from the Chinese COVID-19 epicentre. Int Rev Econ Financ 2021; 71: 55–81.

94.

Corbet

Goodell

Günay

. Co-movements and spillovers of oil and renewable firms under extreme conditions: new evidence from negative WTI prices during COVID-19. Energ Econ 2020; 92. DOI: 10.1016/j.eneco.2020.104978.

95.

Chang

C-L

McAleer

Wang

Y-A

. Herding behaviour in energy stock markets during the global financial crisis, SARS, and ongoing COVID-19*. Renewable Sustainable Energy Rev 2020; 134: 110349.

96.

Lin

. Does COVID-19 open a Pandora’s box of changing the connectedness in energy commodities? Res Int Bus Finan 2021; 56. DOI: 10.1016/j.ribaf.2020.101360.

97.

Samadi

Owjimehr

Halafi

. The cross-impact between financial markets, COVID-19 pandemic, and economic sanctions: the case of Iran. J Policy Model 2021; 43, 1: 34-55. DOI: 10.1016/j.jpolmod.2020.08.001.

98.

Engle

. Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom inflation. Econometrica 1982; 50: 987–1007.

99.

Bollerslev

. Generalized autoregressive conditional heteroskedasticity. J Econometr 1986; 31: 307–327.

100.

Glosten

Jagannathan

Runkle

. On the relation between the expected value and the volatility of the nominal excess returnon stocks. J Finance 1993; 48: 1779–1801.

101.

Pesaran

Shin

Smith

. Bounds testing approaches to the analysis of level relationships. J Appl Econometr 2001; 16: 289–326. DOI: 10.1002/jae.616.

102.

Shin

Greenwood-Nimmo

. Modelling asymmetric cointegration and dynamic multipliers in a nonlinear ARDL framework. In: Sickles

Horrace

(eds) Festschrift in honor of peter Schmidt econometric methods and applications. New York: Springer Science & Business Media, 2014.

103.

Ertuğrul

Güngör

Soytaş

. The effect of the COVID-19 outbreak on the turkish diesel consumption volatility dynamics. Energy Res Lett 2020; 1. DOI: 10.46557/001c.17496.