key: cord-0782319-5zjbn6ke authors: Khan, Khalid; Su, Chi-Wei; Zhu, Meng Nan title: Examining the behaviour of energy prices to COVID-19 uncertainty: A quantile on quantile approach date: 2021-10-29 journal: Energy (Oxf) DOI: 10.1016/j.energy.2021.122430 sha: b48c9508d537e82172758c2d9660148602fc1b03 doc_id: 782319 cord_uid: 5zjbn6ke The energy market is extremely vulnerable to the uncertainty caused by the pandemic and leads to global lockdowns and stagnant economic activity. This study is important because energy prices (EPs) experience a dramatic decline due to the pandemic, which has negative consequences for the global economy. We aim to analyze EPs behaviour to coronavirus (COVID-19) from 2020:01 to 2021:05. The finding shows that EPs are extremely vulnerable to the uncertainty produced by the pandemic in the short run. The COVID-19 has a negative effect on EPs in the medium to upper quantile, which suggests that higher uncertainty caused by the pandemic results in rapid decline. However, the impact of the COVID-19 is greater on the oil prices (OPs) as compared to the natural gas (NGP) and the heating oil price (HOP). Moreover, the finding reveals that COVID-19 impact on EPs are consistently negative across all the quantile. The degree of the impact increases when the relationship changes from short to long run. The pandemic has affected the energy price in the short run, which needs prudent policies to fully grasp the magnitude of the COVID-19 impact on energy prices. The main purpose of this paper is to evaluate the impact of coronavirus on the different energy prices (EPs) such as West Intermediate Texas price (WOP) and Brent oil price (BOP), natural gas (NGP) and heating oil price (HOP). The international energy market and uncertainty have a strong relationship (Khan et al., 2020b) , which is manifested in the current pandemic. The world is passing through a difficult phase due to the COVID-19 pandemic, which has a far-reaching impact on every segment of the economy (Nyga-Łukaszewska and Su et al., 2021a) . The international Monetary Fund (IMF) predicts that the global economy will shrink by 4.4% in 2020 and shock is more severe than the global financial crisis in 2008 (Mensi et al., 2020; Smith et al., 2021) . It shows that stagnant economic growth has a negative impact on energy consumption and demand (Gozgor et al., 2018; Wang et al., 2021b; Tao et al., 2021a) . The crisis has battered the global economy, causing the worst recession since the great depression of the 1930s. The economic shock caused by the imposing lockdowns proves costlier than the pandemic itself. Moreover, the pandemic is strangling the global economy, leading to layoffs, a tourism crisis and plunging oil prices (OPs) (Khurshid et al., 2020) . It has restricted transportation, air traveling and factories are closed, which results in less oil demand. Moreover, energy consumption declines because of the pandemic. Figure 2 shows the global energy consumption by source and the share of energy source in total. It explores that global energy consumption has increased rapidly in the last three decades, driven by robust economic growth (Azadeh and Tarverdian, 2007; Aydin, 2015a) . The consumption of coal, oil and gas has been rising over the years of the total energy. This rising demand has increased energy security and diversifying energy sources has become the important priority of the energy policy (Aydin, 2012; Aydin, 2015b) . The energy consumption will increase for sustainable economic J o u r n a l P r e -p r o o f growth, which can be vulnerable to political and economic events (Kön and Büke, 2010; Feng et al., 2012) . Note: OC denotes oil consumption; GC indicates gas consumption; CC represents coal consumption; HC expresses hydro consumption and NC denotes nuclear consumption. The uncertainty caused by the pandemic has led to global lockdowns, travel restrictions, stagnant economic growth and the declaration of a state of emergency by the World Health Organization (WHO), which is extremely detrimental to EPs (Navon et al., 2021; Su et al., 2021e; ). There is a high level of uncertainty in wake of the COVID-19, which has disturbed EPs dynamics and investors sentiments (Mensi et al., 2020; Mzoughi et al., 2020; Su et al., 2021b) . The mobility constraints lead to demand drop and a huge decline in EPs because the aviation and transport sectors account for 60% of oil demand (Nyga-Łukaszewska and . The international energy market is in the doldrums in the pre-pandemic period as a result of the feud between the major oil producers. This price has resulted in the oversupply in the market, which reflects in the low OPs (Wang et al., 2021b) . The leading oil producer countries supply more oil than ever, leading to a collapse in price. However, the situation becomes worse with the breakout of the pandemic and leads 50% fall in EPs in March 2020. OPs have witnessed a historical fall and turn into negative and demand declines dramatically. EPs experience a rapid decline in April 2020 and first time in history oil futures falls negative (IEA, 2020a Gao et al., 2021; Tao et al., 2021c) . It is normal for prices to rise after the recession, and NGP returns to prepandemic levels. Section 2 of this paper reviews the previous literature, which follows the methodology in section 3. Whereas the data is explained in section 4, which is followed by an empirical analysis in section 5. The conclusion of this study is highlighted in section 6. Nyga-Łukaszewska and Aruga (2020) analyze the COVID-19 impact on oil and gas market. They find pandemic has severely hit the global energy market. It shows that COVID-19 has a negative effect on OP in the U.S. and Japan, while a positive impact on gas prices. examine the COVID-19 impact on energy consumption in India. The outcomes reveal that energy consumption declines dramatically during the lockdown. However, it recovers as the lockdown is relaxed. Mzoughi et al. (2020) find that the COVID-19 pandemic leads to a reduction in the oil demand and price collapse to negative. However, the negative response of the oil market is brief and recovers in the short run. Kingsly and Henri (2020) show that OP plummets as the pandemic intensifies in Europe and North America. The oil demand will decline in the major importers countries as the industrial production may slow down. Novan et al. (2021) report that energy sectors have reacted quickly to the COVID-19. The strict lockdown has slammed the airline and transport industry which is reflected in the low demand for energy and low prices. Gao et al. (2021) The application of wavelet analysis in economics is common (Su et al., 2019; Su et al., 2020) . It is like a wave oscillation starting at zero and altering and reverse back to zero (Yahya et al., 2019) . We employ a wavelet with different frequencies to fit the time series in both the time and frequency domain (Graps, 1995; Torrence and Webster, 1999; Crowley, 2007) . We can construct dyadically the wavelet and converted a pair of particularly constructed function and such that: where and denote the father wavelet and mother wavelet, respectively. The former detects the smooth and low-frequency components of the variable and later detect the comprehensive and high-frequency components of the variable. Thus, the achieved wavelet is illustrated as follow based on the above equations. The number of observations restricts the maximum number of scales that the scrutiny can measure ( ≥2 ). A special attribute of the wavelet expansion is the coefficient of the positioning attribute , ( )which signifies the information content of the function at the estimated position 2 − and frequency 2 . Thus, the 2 (ℝ) can be expanded underlying wavelet at the arbitrary level 0 ∈ ℕ over different scales. where. of ( ). The wavelet transform is performed of the discrete sampled by Discrete wavelet transform (DWT). It is based on the scaling filter ( , =0,……. −.1 2 ) and the wavelet filter ( , = 0, … . , − 1 3 ). ∈ℕ denotes the length of the filter (Percival and Walden, 1997) . The wavelet filter fulfils these three characteristics. The low-and high-pass filters are defined as quadrature mirror filters, which Similarly, the scaling filter fulfils the conditions. J o u r n a l P r e -p r o o f The wavelet and scaling coefficients of DWT at the pth level for p∈{1,…, p} are defined as: The maximal overlap discrete wavelet transform (MODWT) suggested by Percival and Walden (1997) The discrete wavelet transform (DWT) has drawbacks of the dyadic length restriction for the series to be transformed and that it is non-shift invariant. Thus, the MODWT is introduced to overcome the weaknesses of DWT by giving up its orthogonality and gaining the ability to handle any sample size regardless of whether the series is dyadic or not (Haniff and Masih, 2018) . The MODWT has the advantage that it is capable of managing a multivariate analysis (Crowley and Lee, 2005) . Moreover, the disadvantage of the quantile regression is the lack of power capture dependence in the entirety (Gupta et al., 2018) . It overlooks the possibility of the nature (i.e., low, normal, or high levels) of COVID-19 also could influence the way EPs is predicted. The quantile regression does not consider the nature of large and small changes that impact the association (Gao et al., 2021) . The asymmetric relationship, such as the positive shock of one variable may have a different influence on the other as compared to the negative shock which is not assessed. Thus, the QQ method is applied to detect the dependence in its entirety so that correlation between variables could vary at each point of their respective distributions. Similarly, the QQ method offers a complete picture of dependence (Gupta et al., 2018) . It can reveal the effects of shocks at varying degrees and heterogonous tail dependence structure. The QQ approach has the advantage of flexibility, which normally detects the functional form of the dependency nexus between variables. We briefly explain the characteristic of the QQ method proposed by Sim and Zhou (2015) and specify the model to investigate the COVID-19 impact on the energy price. The QQ method is used to detect the impact of the pandemic on EPs. Moreover, it is used to assess the dependency pattern of EPs on COVID-19. The approach is the aggregate of quantile regression and nonparametric estimation of local linear regression (Li and Yuan, 2021) . Therefore, we use the QQ approach to examine the impact of the quantiles of COVID-19 on the quantiles of the various energy prices such as WOP, BOP, NGP and HOP. The procedure begins with the nonparametric quantile regression model. We examine the association between the ϑth quantile in the background of COVID τ by using local linear regression. As (. ) is unidentified, this function can be estimated by a first-order Taylor expansion about the quantile COVID τ . where show the partial derivative of ( −1 ) in the context of COVID. The outcome is termed the marginal effect which can be construed like the slope coefficient in the linear regression model. The parameters α ϑ (COVID t−1 ) and α ϑ (COVID τ ) are the conspicuous aspect of Equation (15) which are doubly indexed in ϑ and τ. Assumed that α ϑ (COVID t−1 ) and α ϑ (COVID τ ) are both functions of ϑ and COVID τ , and that COVID τ is a function of , it is clear that both α ϑ (COVID t−1 ) and α θ (COVID τ ) are functions of θ and τ. Furthermore, α ϑ (COVID t−1 ) and α θ (COVID τ )can be retitled as α 0 (ϑ, τ) and α 1 (ϑ, τ) respectively. Therefore, Equation (15) can be rephrased as: By replacing equation (14) into (16) we get (17) where EP represents the energy price of WOP, BOP, NGP and HOP. The part ( * ) of Equation (17) is the th conditional quantile of COVID. However, unlike the function of the standard conditional quantile, this manifestation replicates the association between the th quantile of and the th quantile of EP because the parameters 0 and 1 , are doubly indexed in ϑ and . Likewise, a linear relation is not supposed at any time between the quantiles of the studied variables. Estimating Equation (17) We decompose the regression series into short, medium and long-term trends. The short-run reveals COVID-19 impact on EPs in the short-term horizon (between 1 to 16 days). The medium-term horizon measures the variation between 32 to 64 days. Similarly, the COVID-19 impact on EPs in the long-term horizon from 128 to 256 days. We explain the correlation coefficients between COVID-19 and EPs in Table 2 . The outcome illustrates that COVID-19 and EPs are highly positively correlated for all the energy sources. Furthermore, the highest correlation coefficient is recorded for the NGP, followed by BOP and WOP. The correlations are highly significant, evident from the p-values which are statistically significant at 1% level. The energy market is facing challenges because of the price war between Russia and Saudi Arabia, which maybe further serious by the pandemic. From February to March 2020, the energy market has witnessed increasing oil supply mainly driven by the price rivalry between Russia and Saudi Arabia. Similarly, the U.S. oil supply hit a record level per day due to heavy borrowing by the U.S. companies. However, the oversupply is overshadowed by the COVID-19 and OPs have fallen by more than half in late March 2020. Moreover, the U.S. oil production falls in April 2020. The oil supply is rising, refineries and storage are brimming because the pandemic coincides with surging oil production. Crude oil is traded on futures contracts and fears of no storage may lead to a severe drop in WOP from $18 a barrel to around -$37 a barrel. Similarly, oil witnesses a severe drop to its lowest in 18 years at the global level and the demand has declined by nearly more than half since mid-March 2020 because of lockdown and traveling restrictions to contain the virus spread. The major producing countries such as Saudi Arabia and Russia keep pumping throughout March and have cut down in April by 10 million barrels per day. The production cut is not matched with the decrease in demand and EPs continue to decline. The gradual reopening of the various economies sustains the market sentiment, which causes EPs to recover to a moderate level. The prices recover as countries emerge from the lockdown and OPEC agree to a production cut. However, a fresh wave of the COVID-19 case could cause prolonged low crude OP as demand for fuel is plunging from longer movement restrictions globally. BOP declines in August 2020 because of a fresh wave of COVID-19 and new lockdowns may restrain demand. Meanwhile, the total global cases have crossed 20 million and record deaths in the U.S. which shows a weaker energy demand. The price shows a declining trend in September 2020 due to concern over the resurgence of pandemics and the expectation of oversupply. However, BOP and WOP remain at an average of $41.80 and $40.05, respectively. Likewise, the energy market has posted the largest decline in November 2020 since March as new tougher lockdown measures to control pandemic which can threaten demand recovery. Meanwhile, the improvement in China and India, booming freight market is supportive of the energy demand revival. Similarly, the U.S. presidential election and the OPEC+ meeting in November have a significant impact on the price and future J o u r n a l P r e -p r o o f outlook. The OPEC face challenges to keep the supply in control as Libya and Iraq are expected to return output. However, Iraq affirms to support the OPEC+ production cuts, Norway will supply to the pre-COVID-19 level. The vaccine trial in November 2020, which is very successful and the market looks for a brighter future with a victory over the pandemic. The dramatic improvement in sentiment as a result of the oil market fundamentals and supported by the financial market. Furthermore, the WHO data shows that COVID-19 is declining, which is a positive indication for the oil market and the expectation of demand is rising. The oil demand has surged by 50% since October but recovery still facing obstacles in global fuel demand. BOP surges since January and WOP reach $60 a barrel because of rising demand and reopening of economies. During the period, NGP is suppressed by the COVID-19 as demand declines. The impact is more evident in the short run, mainly caused by the pandemic and the mild winter seasons. Moreover, the gas industry in the crisis in the pre-COVID-19 period which is furthered by the emergence of the pandemic (Gao et al., 2021; Su et al., 2021d) . A price hike is witnessed in the gas in February 2021 because the world is opening up and the economy is again on the rise and prices are rising along with it. The rising prices are normal after the recession and the gas prices are going back to the level before the pandemic. We exhibit the QQ results of the original and decomposed series in which reveals that pandemic at the upper quantile leads to higher declines in WOP. As WOP are the U.S. oil price benchmark, that has witnessed excessive supply from the producer in the pre-pandemic period and the energy market is further damaged by the COVID-19. The results are similar to Gao et al. (2021) which explains the pandemic influence the two major countries, China and the U.S. and conclude that COVID-19 has a severe impact on the energy market in these countries. We highlight the decomposed data outcomes in Figure 4 (B-D) . It shows that the COVID-19 influence the WOP negatively in the short run in the medium to upper quantile (0.55-0.75). The highest impact of the pandemic on WOP (-0.008) is witnessed, which reveals that uncertainty caused by the COVID-19 leads to a rapid decline in WOP in the short run. WOP declines rapidly when the pandemic breakout. Note: graph demonstrates quantiles of COVID-19 in the x-axis and the quantiles of EPs in the y-axis The QQ technique decomposes the estimates of quantile regression which allowing particular estimates to be achieved for various quantiles of the explanatory variable. In this study, the quantile regression is based on the θth quantile of CVID-19 on EPs. The parameters of COVID-19 and EPs are indexed by θ and τ. Furthermore, the QQ process comprises more disaggregated material about the COVID-19 and EPs nexus than the quantile regression. The QQ approach is heterogeneous across different quantiles. In this regard, the QQ approach is used because of the decomposition characteristics to recover the estimates from the standard quantile regression. Thus, the parameters of quantile regression are indexed by , which is estimated by averaging the QQ parameters along τ. The slope coefficient measures the impact of COVID-19 on EPs and is represented by 1 ϑ and estimated as follows: where s is quantile numbers τ = [0.10,0.15,…..0.90]. regression. The outcome reveals that the effect of COVID-19 on WOP is consistently negative across all quantiles. However, the effect of COVID-19 on WOP is detected in the upper quantile, which suggests that higher uncertainty caused by the pandemic results in rapid decline. It concludes that WOP is less responsive to COVID-19 at the start, while the lockdown and traveling restriction leads to a rapid decline in oil demand, which is readily reflected in WOP. The average QQ regression outcomes express the impact of different measurements and are described in Figure 7 (e-h). It demonstrates that the impact of COVID-19 on BOP is consistently negative across all the quantiles. The finding endorses that the extent of the impact is more causative in the short and long run. It is explained that when the pandemic has started; BOP falls faster, and in the medium term, because of the partial recovery of the economy, oil demand increased and prices stabilized. However, in the long run, BOP declines again as a result of the second wave of the pandemic, which increases the uncertainty and has a negative effect on BOP. (e) (f) The validity of the results are examined by comparing QQ and QQ regression, which is stated in Figure 9 (i-l). It demonstrates that the impact of COVID-19 on NGP is negative except in the long run. The size of the impact is strong in short, which is evident that sudden break out of the pandemic results in the rapid decline in NGP, which follows a recovery in the long run. Quantile of HOP Impact of COVID-19 on HOP Quantile of COVID-19 -50 -40 -30 -20 -10 0 10 0 0.5 1 0 0.5 1 -10 -5 0 5 10 Quantile of HOP Impact of COVID-19 on HOP Quantile of COVID-19 -4 -2 0 2 4 6 J o u r n a l P r e -p r o o f (r) (s) Figure 11 . Quantile Regression and QQ estimates. We examine the COVID-19 impact on EPs by using the QQ method. The finding shows that EPs are extremely vulnerable to the uncertainty produced by the pandemic in the short run. The COVID-19 has a negative effect on EPs in the medium to upper quantile, which suggests that higher uncertainty caused by the pandemic results in rapid decline. The energy demand collapses because of the traveling restriction, lockdown and contraction of the economic activity. However, the impact of the COVID-19 is greater on OPs as compared to NGP and HOP. Moreover, the finding reveals that COVID-19 impact on EPs is consistently negative across all the quantile. The degree of the impact increase when the relationship changes from short to long run. The study offers some useful policy implications for the relevant. First, the results indicate that pandemic has greatly impacted the energy price in the short run. Thus, prudent policies are required to fully grasp the magnitude of the COVID-19 impact on energy prices. Our finding may help to describe the extent of pandemic impact and devising future policies. Similarly, the supply side can be managing by the concerned governments to support the national oil companies. Second, the result suggests that the energy market experience a crisis before the COVD-19 emergence which is aggravated by the pandemic. Therefore, to mitigate the expected similar future crisis and its magnitude, the concerned policymakers should solve pre-COVID-19 problems. The issues consist of the price war, geopolitical tensions and structural transformation which can minimize the magnitude of the impact and may help in speedy recovery. By solving the pre-COVID-19 issues may help because the market condition is more resilient to change. Moreover, the successful diversification of the energy system other than the conventional fossil fuels may help in minimizing the extent of pandemic shock. This is more important for the oil-exporting countries because of their heavy dependence on oil revenues and oil collapse may have negative repercussions. Last, the finding is useful for devising the energy management policies in response to the expected crisis in the future. The proactive measure can reduce the magnitude of the impact and risk of price collapse. We can expand this study to evaluate the explosive behaviour of EPs, as the pandemic has put pressure on global commodities. The disruption of the supply chain has affected commodities, which indicates an upward trend. The energy market is no exception and has been greatly impacted by the uncertainty of the pandemic. Similarly, the cointegration of these EPs can be examined in the context of pandemic uncertainty. It will be a useful discussion to investigate the relationship among different energy prices in the presence of COVID-19. Moreover, the pandemic has resulted in a rising trend in commodity prices which has severe consequences for the global economy. The EPs shows an explosive behaviour in the wake of the COVID-19 which is closely related with the other commodity market. Therefore, EPs bubble detection in different energy prices can be a useful area of future research that will offer useful information for the stakeholders. J o u r n a l P r e -p r o o f COVID 19's impact on crude oil and natural gas S&P GS Indexes COVID-19 Deaths Cases Impact on Oil Prices: Probable Scenarios on Saudi Arabia Economy Effects of Covid-19 on Crude Oil Price and Future Forecast Using a Model Application and Machine Learning Effects of COVID-19 on Indian energy consumption. Sustainability (2020) Analysis and mitigation opportunities of methane emissions from the energy sector The modeling and projection of primary energy consumption by the sources The Application of trend analysis for coal demand modeling Integration of genetic algorithm, computer simulation and design of experiments for forecasting electrical energy consumption Decomposing the co-movement of the business cycle: a timefrequency analysis of growth cycles in the euro area. Bank of Finland Research Discussion Paper Hourly Oil Price Volatility: The Role of COVID-19 Scenario analysis of urban energy saving and carbon abatement policies: a case study of Beijing city To what extent does COVID-19 drive stock market volatility? A comparison between the U.S. and China Energy consumption and economic growth: New evidence from the OECD countries An introduction to wavelets Does partisan conflict predict a reduction in US stock market (realized) volatility? Evidence from a quantile-onquantile regression model☆ Do Islamic stock returns hedge against inflation? A wavelet approach International Energy Agency International Energy Agency Impacts of COVID-19 on energy demand and consumption: Challenges, lessons and emerging opportunities Does Oil Prices Cause Financial Liquidity Crunch? Perspective from Geopolitical Risk How COVID-19 shock will drive the economy and climate? A data-driven approach to model and forecast COVID-19 and Oil Prices Forecasting of CO2 emissions from fuel combustion using trend analysis The nexus between industrial growth and electricity consumption in China-New evidence from a quantile-on-quantile approach Coronavirus outbreak and the global energy demand: A case of people's republic of china A theory for multiresolution signal decomposition: the wavelet representation Impact of COVID-19 outbreak on asymmetric multifractality of gold and oil prices The effects of COVID-19 pandemic on oil prices, CO2 emissions and the stock market: Evidence from a VAR model. CO2 Emissions and the Stock Market: Evidence from a VAR Model The impact of COVID-19 on the electricity sector in Spain: An econometric approach based on prices Effects of the COVID-19 Pandemic on Energy Systems and Electric Power Grids-A Review of the Challenges Ahead Energy Prices and COVID-Immunity: The Case of Crude Oil and Natural Gas Prices in the Analysis of subtidal coastal sea level fluctuations using wavelets The Essential Role of Pandemics-A Fresh Insight into the Oil Market BitCoin: A new basket for eggs The energy consumption and economic growth nexus in top ten energy-consuming countries: Fresh evidence from using the quantile-on-quantile approach COVID-19 pandemic, oil prices, stock market, geopolitical risk and policy uncertainty nexus in the U.S. economy: Fresh evidence from the wavelet-based approach Impact of COVID-19 pandemic on the energy markets. Economic Change and Restructuring Oil prices, U.S. stock return, and the dependence between their quantiles Assessing the impact of COVID-19 on global fossil fuel consumption and CO2 emissions Does geopolitical risk strengthen or depress oil prices and financial liquidity? Evidence from Saudi Arabia A Review of Resource Curse Burden on Inflation in Venezuela Does renewable energy redefine geopolitical risks Can new energy vehicles help to achieve carbon neutrality targets? Financial aspects of marine economic growth: From the perspective of coastal provinces and regions in China Can bank credit withstand falling house price in China Effects of the 2008 Financial Crisis and COVID-19 Pandemic on the Dynamic Relationship between the Chinese and International Fossil Fuel Markets The dynamic effect of eco-innovation and environmental taxes on carbon neutrality target in emerging seven (E7) economies Robo advisors, algorithmic trading and investment management: Wonders of fourth industrial revolution in financial markets Do financial and non-financial stocks hedge against lockdown in Covid-19? An event study analysis Interdecadal changes in the ENSO-monsoon system Temporal and spectral dependence between crude oil and agricultural commodities: A wavelet-based copula approach Coronavirus disease 2019 and the global economy Available at SSRN 3554381 Whether crude oil dependence and CO2 emissions influence military expenditure in net oil importing countries Geopolitical risk and crude oil security: A Chinese perspective