key: cord-0818101-4ubnyyiz authors: Wang, Qiang; Yang, Xuan; Li, Rongrong title: The impact of the COVID-19 pandemic on the energy market – A comparative relationship between oil and coal date: 2021-12-01 journal: Energy Strategy Reviews DOI: 10.1016/j.esr.2021.100761 sha: bc9156e1ce6ee4d621862e75346190d06cfd5392 doc_id: 818101 cord_uid: 4ubnyyiz The COVID-19 epidemic has severely affected the world economy and energy markets. In order to alleviate the shock, stabilize the financial market, and promote economic recovery, the Fed announced an unlimited QE policy. In order to understand the impact of the policy on the energy market under the extreme events, the study selected WTI crude oil and coal prices from January 1, 2018 to May 7, 2021 as the research objects. Taking the two years before the epidemic, the epidemic stage was further divided into four small stages according to the three peaks of the epidemic in the US. The MF-DCCA model calculations show that coal and WTI crude oil have an interactive relationship. The risks between them are not just averaged and superimposed, but transmitted and interacted. MF-DFA calculation results show that due to the disorder of energy supply and demand under the epidemic, market efficiency in the first quarter of 2020 has dropped rapidly. However, market efficiency will be decoupled from the development of the epidemic in the second half of 2020. Especially after the announcement of the QE policy, market efficiency has improved significantly. However, under the excessive monetary policy, market efficiency will decline in the first half of 2021. This shows that the policy has a certain effect on alleviating the impact of the epidemic on the energy market. In the long term, this improvement is not sustainable. As prices rise, inflation continues. In the future, the volatility and risk of the energy futures market will increase. shows the daily closing price fluctuations of WTI crude oil and coal, from 11 two years before the COVID-19 to the fourth peak of the epidemic (2018.1.1-12 In order to ensure the time series of the sample is synchronized, eliminating the 14 non-trading and non-matching missing days data. So, it is a total of 775 data, 15 respectively. The data of WTI crude oil comes from the US Energy Information The Figure 2 can be seen that the volatility of the two markets before and after the 6 epidemic has changed greatly. The market price curve exhibits nonlinear and non-7 stationary characteristics. After the epidemic, the closing prices of WTI crude oil and 8 coal show unusual characteristics and have different trends in the short time. In 9 particular, WTI crude oil futures even became negative on April 23. This phenomenon 10 shows that market occurred extreme events "COVID-19" epidemic during this period, 11 which changed the original volatility of the market. This has increased the volatility and 12 risks of the entire market. In order to adjust the energy market, the Federal Reserve 13 introduced an unlimited QE policy to alleviate the "serious damage" caused by the 14 epidemic to the energy market economy. The intuitive change is, after that, the volatility 15 decreased and prices rose slowly, even higher than before. At the same time, it is worth 16 noting that the changes trend of WTI crude oil and coal price are very similar. There is 17 and analyze the changes in the energy market during the epidemic in the US, this paper 2 uses the market data of the two years before the epidemic as a control. In order to further 3 analyze the phased changes of the market under the development of the epidemic, this 4 paper combines the three peak periods (Figure 1 . April 6, 2020, July 17, 2020 and 5 January 8, 2021) of the epidemic in the US to divide the epidemic period into four 6 stages. The first stage is from January 30, 2020 to April 6, 2020. Energy market prices 7 are gradually falling. Since March 3, the Fed has used tools such as emergency 8 reduction and unlimited QE policy to deal with the impact of the epidemic on the US 9 energy market economy [35] .The second stage is the period of severe volatility from 10 April 7, 2020 to July 17, 2020. On April 20, 2020, oil prices fell for the first time to a 11 historical negative value of -38 U.S. dollars. It triggered a global financial turmoil, 12 leading to a sharp decline in the US economy. On April 9, the Fed increased its monetary 13 support and provided loans of up to $2.3 trillion to stabilize the market economy. After 14 experiencing the plunge, oil prices began to rise slowly. The volatility in the market has 15 subsided. But coal prices continue to fall. The third stage is from July 18, 2020 to the epidemic on market confidence [36] . In December, the United States passed a 900 19 billion U.S. dollar aid program. During this period, WTI crude oil and international 20 coal prices continue to rise. The final stage is from January 9, 2021 to May 7, 2021. In 21 1 package[37]. At present, the market is still in uncertainty. During this period, the 2 government also put forward a series of new energy viewpoints. 3 The return rate series of the WTI crude oil and coal futures market is a non-normal 5 and non-linear time series structure with non-stationary characteristics. Traditional 6 statistical analysis methods cannot be directly used to analyze their fluctuation 7 characteristics and the relationship between them. The multifractal detrended 8 has great advantages in studying non-stationary time series. It can compare and analyze 10 the risk and volatility of the two markets through the q-order Hurst index. However, the 11 relationship between the two markets is usually more complex, interrelated and 12 mutually transmitted. Therefore, this paper introduces the multifractal detrended cross-13 correlation analysis (MF-DCCA) developed by Zhou et al. [39] to further analyze the 14 cross-correlation of the overall market and the risk conduction effect between the two 15 markets. 16 Before studying the cross-correlation relationship, we first used the MF-DFA 17 method to separately quantify the risk of WTI crude oil and coal futures markets, and 18 used the Hurst index of the return rate series to represent the persistence characteristics 19 and risk volatility of the two futures markets during the epidemic. Generally speaking, 20 (q=2)>0.5, the time series has a long-term continuous development trend, that is, the 1 market will maintain the same trend and continue to develop after this. When h 2 (q=2)<0.5, the time series has an anti-sustainable development trend, that is, the market 3 will show a different development trend from before. And, the greater the value of h 4 (q=2), the greater the volatility of the market during this period and the greater the 5 market risk. 6 Next, based on the overall market level, we use the MF-DCCA method to analyze 7 the risk transmission between the two markets. The process is as follows: 8 Step1: Construct the side sequence 9 First, set the time series of the returns of WTI crude oil and coal futures markets 10 as and . 11 (2) Where denotes the average over the whole time series. t=1, 2,⋯ , 12 Step 2: Divide the profile{ } =1 14 nonoverlapping segments of equal length s. Since ⁄ is not necessarily an integer, 16 in order not to discard the remaining data at the end of the time series. In order not to 17 disregard this part, the same procedure is repeated starting from the opposite end. So, 18 get 2 parts, that is { + } =1 and{ + } =1 . It can be expressed in detail as: 1 { ( −1) + } =1 , = 1,2, ⋯ , , = 1,2, ⋯ , s Step 3: Use the least squares method to fit the local trend of 2 subsequences to 2 obtain the V th fitting polynomials Step 4:Calculate the q fluctuation function 5 Step 5:If there is a certain power-rate cross-correlation relationship between the 6 two markets, the following relationship hold: 7 Change q to obtain ℎ (q) corresponding to different q, where, for each q value, 8 use the least-square (OLS) to linearize log ( ) log return, then draw a 9 scatter plot, where the slope of the regression line ℎ (q) is a power law exponent, 10 called the generalized-cross correlation exponent. 1 If ℎ ( ) has nothing to do with q, it means that there is no cross-correlation 2 relationship between the two market systems. On the contrary, it proves that there is a 3 correlation and shows a multifractal feature. Subdivided further, if ℎ ( )> 0.5, this 4 cross-correlation relationship is long-persistent. It implies that when the price of one 5 market rises (decreases), another market also will show an upward (decreasing) trend. 6 If ℎ ( ) <0.5, then this cross-correlation is anti-persistent. This means that when the 7 market changes, another market also shows an opposite trend. In addition, when q<0, 8 ℎ ( ) reflects the scaling behavior of small fluctuation factors in the market. When 9 q>0, ℎ ( ) reflects the scaling behavior of market fluctuation factors. 10 Step 6:Analyze multifractal spectrum 11 If the correlation between these two markets is non-linear and has multifractal 12 characteristics, then it can further get the Renyi mass exponent τ(q): 13 Using the Legendre transformation, the singular exponent α(q) and the In addition, the multifractal intensity is generally expressed by the multifractal 16 spectrum width ∆α and strength ∆h: trend. That is, these two energy futures markets have always had multifractal 2 characteristics. The deteriorating development of the epidemic has not affected the 3 multifractal characteristics of these two independent markets. 4 Further analyze the multi-divisional characteristics and volatility of the energy 5 market during the four stages of the epidemic. As shown in Figure 5 and Figure 6 , the 6 volatility of the energy market at different stages is significantly different. In the third 7 stage, market volatility is significantly weakened. This stage is a period of relatively 8 stable market development so far. In the fourth stage, the market's volatility increased 9 significantly. On the whole, the second stage is the most volatile period in the WTI 10 market so far, while the coal market is the most volatile in the fourth stage. These can be found that the volatility of the WTI market was greater than that of the coal 14 market in the first two periods. In the later period, the volatility of the coal market was 15 more severe than that of the WTI market. Despite the number of new cases per day in 16 US hit a peak again, the above results show that market volatility is not completely 17 affected by the epidemic. Crude oil and coal because of their mutual substitution, their markets also exist 2 mutual interference and connection Therefore, during the epidemic, investors who enter 3 the energy futures market for trading and regulators who formulate relevant policies 4 should consider these two markets as a whole related to each other. If only consider one 5 of them, and ignore another market, it will lead to deviations in the overall 6 understanding of the market. But due to the differences between the two markets and 7 the interference of the external environment, this relationship cannot be directly 8 observed. Based on this, this article further uses the MF-DCCA method, using 9 multifractal spectrum and related parameters to further reveal the nonlinear dynamic 10 characteristics during the epidemic. 11 Figure 7 shows the multifractal spectrum between the two energy market return 12 series. It can be seen from the double logarithmic graph of the covariance fluctuation 13 function that for different q. When changing from -4 to 4, the corresponding double 14 logarithmic surface is uneven. It indicates that there is a power-law cross-correlation 15 between function and time scale. The mass exponent τ(q) and cross correlation 16 exponent h(q) both increase nonlinearly with the change of q. At the same time, the 17 multifractal spectrum f(α) chart presents a single-peak bell shape. In summary, it is 18 shown that there is a cross-correlation between the returns of the WTI crude oil and 19 coal market in the research interval. In Table 2 and Table 3 ,when q=2, the persistent characteristics of the independent 3 energy market in the research interval can be judged from the Hurst exponent (h(q=2)). 4 The cross-correlation exponent ( ) = 2 can determine the impact of the 5 correlation between the two energy markets. The value of △ h represents the 6 multifractal degree of the market structure and the market efficiency. The smaller the 7 value, the higher the efficiency of the market. J o u r n a l P r e -p r o o f represents the direct addition of the two energy futures markets. The interactive market represents that the two energy futures markets are mutually 2 influencing and restricting each other. Comparing the market efficiency of the four periods of the epidemic with that of 1 the market before, it is found that after the epidemic, △h increased rapidly, and the 2 market efficiency decreased significantly. In particular, the first stage is most obvious. 3 In Table 3 stage is slightly higher than that of the WTI crude oil market. However, the cross-20 correlation exponent of the two markets is 0.5385, which is greater than 0.5. This shows 21 that there is a trend of "synchronization" in the interactive market composed of crude 1 oil futures and coal futures. 2 After entering the second stage of the development of the epidemic, the number of 3 daily new cases has continued to increase. However, the efficiency of the two markets 4 did not decline further due to the severity of the epidemic. The△h of the WTI crude oil 5 market dropped to 0.6480, and the △h of the coal market dropped to 0.4918. The △ 6 h of the interactive market also dropped from 0.4868 to 0.4827, and market efficiency 7 has improved. When q<0, the hq of both markets is greater than 0.5. We can deduce 8 that small fluctuations in these two independent markets are long-persistence and have 9 long-term memory. The cross-correlation exponent is 0.4055, which is less than 0.5. 10 This also shows that WTI crude oil and coal markets have opposite development trends. 11 It is observed from the Hurst index (h(q=2)) of the WTI crude oil futures market is 12 0.5308, which has long-persistence. The later development will continue the trend of 13 this stage. The Hurst index of the coal market is 0.3720, showing anti-persistence 14 characteristics. This means that the follow-up development trend is different from this 15 In the stage III, as the development of the epidemic enters a new peak period, the 17 38.64%. And the market's Hurst exponent (h(q=2)) and cross-correlation exponent 2 ( = 2) are both greater than 0.5. The entire market shows long-persistence 3 characteristics and "synchronous" trends. However, after entering the stage IV, market 4 efficiency declined. The ( = 2) of the interactive market is 0.4953, that is, the 5 development of the two markets is not completely synchronized. Among that, the h(q=2) 6 of the WTI crude oil market is 0.4751, and the market is anti-persistent. The h(q=2) of 7 the coal market is 0.5028, and the market exhibits long-persistent. 8 In summary, it can be seen from Table 3 that the development status of WTI and 9 coal markets in the four periods is different. However, whether it is a single market or 10 an interactive market, market efficiency has rapidly declined after the epidemic. 11 However, with the continuous development of the epidemic in US, market efficiency 12 in the second and third stages has not declined, but has improved. Especially in the 13 stage III, the market efficiency of WTI crude oil market increased by 34.86%. The 14 market efficiency of coal market increased by 9.54%. The efficiency of the interactive 15 market has increased by 38.64%. After entering stage IV, market efficiency declined 16 rapidly. They dropped by 33.81%, 37.43% and 53.49% respectively. Moreover, the 17 persistence characteristics h (q=2) of the market in the first three stages are consistent 18 with the actual development trend, which shows that the value of h (q=2) can describe 19 the development trend of the energy market in the short term. In addition, it is worth 20 noting that when the WTI crude oil and coal markets are analyzed as independent 21 individuals and as a whole, the risk of the interactive market is greater than the average 1 market and less than the integrated market. This also fully shows that there is transitivity 2 and contagion between the two markets. 3 In this study, compared with the energy market efficiency two years before the 5 outbreak of the epidemic, we analyzed the changes in market efficiency during the three 6 peak periods of the epidemic. On the whole, after the outbreak of the epidemic, market 7 efficiency declined rapidly and risks increased. However, it is different because of the 8 Fed's QE policy, global energy supply and demand relationship and energy views of 9 US. Next, this paper discusses the causes of the interactive market after the outbreak, 10 and focuses on analyzing the phased changes in energy market efficiency. 11 (1) The previous calculation results show that the WTI crude oil and coal markets 12 have always had an interactive relationship before and after the epidemic. Specifically, 13 in phase I and phase II, the two markets have a long-persistent correlation, showing a 14 lagging trend of "same rise and fall". This shows that the changes in the WTI crude oil 15 market may lead to similar developments in the coal market. In phase III and phase IV, 16 the two markets showed opposite development trends. Moreover, when q> 0, the ℎ 17 value of the interactive market return sequence is lower than 0.5, and the two markets 18 are inversely correlated. However, the response of the two markets to internal factors is 19 still a long-lasting cross-correlation. During the epidemic, external factors dominate. 20 That is, the epidemic, the economy (unlimited quantitative easing policy) and energy 21 policies have a greater impact on the WTI crude oil and coal market. 1 Moreover, there is risk transmission in the interactive markets. As shown in Table 2 3, the risk of a single market is less than the risk of an interactive market at each stage. 3 To a certain extent, the risks of low-efficiency markets are passed on to high-efficiency 4 markets, and high-efficiency markets also weaken the risks of low-efficiency markets. 5 This shows that the risks within the two markets with interactive correlations are not 6 only averaged and superimposed, but transmitted and interacted. coal and crude oil, as 7 part of the replacement relationship of two important fossil energy sources, determine 8 the cross-correlation and risk transmission of the two futures markets. That is, the shift 9 in demand for crude oil and coal by related industries and products will cause relative 10 changes in the prices of the two markets. For example, when crude oil prices continue 11 to remain high, large energy consuming countries usually look for alternative energy 12 sources. For countries with abundant coal reserves or more convenient to import coal, 13 energy consumption has shifted from crude oil to relatively cheap and stable supply of 14 1 commitments to reduce production [42, 43] . Worries about oversupply have increased. Relative to oil, thermal coal prices remain elastic, albeit at a low level. Since the 12 21st century, as countries continue to pay attention to global climate change and 13 accelerate the adjustment of energy structure, many developed countries have gradually 14 adopted clean energy to replace coal energy consumption. Therefore, coal supply is 15 gradually decreasing. Under the first round of the epidemic, the coal market was less 16 affected than the crude oil market. The crude oil market is less efficient than the coal 17 futures market. 18 Starting in the second half of 2020, the efficiency of the coal market is lower than 19 that of the WTI crude oil market. Affected by the epidemic and mismatch of supply and 20 demand, coal prices have continued to decline. Although coal prices in the fourth stage 21 began to rise, this is essentially a cumulative outbreak of contradictions between supply 1 and demand in the thermal coal market after the outbreak. 2 On the supply side, coal prices fell during the epidemic. Many companies have 3 closed some mining areas and reduced production. The production capacity has not 4 been fully released. In 2019, Australia's coal production was 13.21 Ejoules. In 2020, 5 coal production fell by 6.2%. Global coal production has also fallen by 5.1% [40] . 6 Although coal prices continue to rise in third stage. However, in the fourth stage, coal 7 demand shifted to the off-season, and coal price adjustments fell. The epidemic in 8 Europe spread again. It casts a shadow on the global economy. Most markets have 9 adjusted significantly. Industrial chain companies such as steel and coking are facing 10 severe and complex market risks. Stable supply and demand in the steel and coal 11 markets are facing challenges. In particular, coal market prices fluctuate sharply, and 12 corporate market demand for hedging has increased. As a result, the risk of the coal 13 market in the fourth stage has increased. On the whole, from the second half of 20 to 14 the first half of 21, the coal market is less efficient than the crude oil market. And this 15 is precisely the interaction between the markets, leading to changes in the coal market 16 risk later than the crude oil market. 17 (3) Whether it is a single energy market or an interactive market, since the global 18 outbreak of COVID-19, market efficiency has dropped significantly. The epidemic has 19 had a significant impact on the world economy and energy market, and the supply and 20 demand of energy are seriously imbalanced. After the epidemic in January 2020, the efficiency of the energy market has 1 dropped significantly, and the prices of WTI crude oil and coal have fallen: In the short 2 term, COVID-19 has had a significant impact on the operation of the global energy 3 market. On the one hand, energy demand has been drastically reduced. Under the 4 epidemic, the global economy is weak, and many countries have stopped work and 5 production to curb the spread of the epidemic. On the other hand, coal supply has been 6 delayed, but crude oil production has not decreased. Some countries are generally 7 implementing the "limited production" policy. However, the OPEC crude oil producing 8 countries have not made a commitment to reduce production for a long time, and US 9 crude oil inventories are backlogged [45] .Disorders in the supply chain have 10 exacerbated the imbalance of world commercial supply and demand, leading to 11 skyrocketing inventories, plummeting prices, declining investment, and declining 12 market efficiency. 13 Since then, the US has experienced two peaks, but the efficiency of the energy 14 market has not continued to decline as speculated. Instead, it has improved in the second 15 and third phases. Especially in the third stage, the improvement is quite obvious. The 16 overall energy market price showed a "V"-shaped trend. In the first half of 2020, the 17 global economic downturn has affected energy prices sharply. However, as the Fed's 18 continues its unlimited QE policy, energy prices continue to rise. The Federal Reserve 19 maintains near-zero interest rates and strongly pushes up the prices of energy 20 commodities. Market efficiency has been greatly improved. At the same time, measures 21 such as employment subsidies, interest rate cuts and loans have stimulated the market, 1 which has also stabilized the investment sentiment of market traders. To a certain extent, 2 it offset the impact of the epidemic on the energy market economy. The market 3 efficiency improved. 4 Further combine the analysis of the supply and demand side. From the fourth 5 quarter of 2020, under the stimulus of continued fiscal policy, the US economy has 6 recovered. At the same time, the weather is getting colder and energy demand is 7 gradually returning to normal. Production capacity has gradually recovered, and crude 8 oil inventories have continued to fall. The global energy market gradually balances 9 supply and demand. The Australian Thermal Coal Price Index and WTI crude oil prices 10 rebounded rapidly in the fourth quarter of 2020, gradually returning to their pre-11 epidemic levels. 12 However, as the Fed continues to increase its quantitative easing policy, the scale 13 of US debt continues to expand. Inflation pressure continued, energy prices continued 14 to rise, and the energy crisis began. The excessive fiscal policy has weakened the 15 economic market's stimulus, and market efficiency has declined. In addition, the Biden 16 government has implemented a broader clean energy plan. On the first day of his tenure, 17 Biden signed an administrative order to cancel the construction of the Keystone XL oil 18 pipeline. In April, Biden's government announced the abolition of subsidies to fossil 19 fuel companies. This move will promote a certain increase in US oil exports in the short 20 term. At the same time, after the US rejoined the "Paris Climate Agreement" and set a 21 carbon neutrality goal, the country's demand for fossil energy decreased and 1 consumption fell. It also brought heavy losses to the crude oil and coal markets. 2 On the whole, the Fed's unlimited QE policy has stabilized the US economy under 3 the epidemic. In the short term, it does play an important role in saving the financial 4 crisis and alleviating the economic recession. However, excessive reliance on monetary 5 stimulus, interest rate cuts, subsidies and other policies cannot actually promote the 6 continued growth of the market. Excessive fiscal stimulus policies have caused 7 commodity prices to rise and inflation to last longer. In the future, the volatility and risk 8 of the energy futures market will increase. 9 10 The COVID-19 epidemic has had a major impact on the US energy market and 12 caused severe market turbulence. After the epidemic, the US quickly adopted unlimited 13 QE policies to stabilize and restore the energy market economy. This paper selected 14 coal and WTI crude oil prices from January 1, 2018 to May 7, 2021 as the research 15 objects. Taking the two years before the COVID-19 as a control, the epidemic was 16 further divided into four stages based on the three peaks of the epidemic in the US. The 17 paper used the MF-DFA and MF-DCCA methods to compare the market efficiency and 18 risk transmission effects of these two markets at different stages, and further analyzes 19 the impact of factors such as the COVID-19 epidemic and QE policies. The main 20 conclusions are as follows: 21 two outbreak peaks in the US, the efficiency of the energy market began to improve. 1 Especially in the fourth quarter of 2020, the improvement is quite obvious. This policy 2 has a certain effect on alleviating the impact of the epidemic on the energy market and 3 stabilizing the market economy. However, market risks increased again in 2021 and 4 market efficiency declined. Therefore, in the long time, the stimulus of excessive 5 monetary policy to the economy gradually weakens. It will even cause commodity 6 prices to rise and inflation. In the future, the volatility and risk of the energy futures 7 market will increase. [45] IEA, OPEC shift to maintain market share will cause global inventory 4 increases and lower prices., 2020. 5 Risk perception and 2 oil and gasoline markets under COVID-19 Implications of COVID-19 pandemic on the global 5 trade networks, Emerging Markets Finance and Trade Coronavirus (COVID-19) -An epidemic or 7 pandemic for financial markets Financial markets under the global pandemic of 10 COVID-19 COVID-19 tests the market 12 stability reserve The bubble contagion effect of 14 The impact of extreme events on energy price 2 COVID-19 and oil market crash: 2 to CDC, by State/Territory EIA, NYMEX Futures Prices Coal prices Coronavirus fears and macroeconomic expectations Inflation thresholds and policy-rule inertia: 7 some simulation results COVID-19, Seignorage, Quantitative Easing and the Fiscal-9 Multifractal detrended fluctuation analysis of nonstationary time series Multifractal detrended cross-correlation analysis for two 14 nonstationary signals Supply as a factor in the destabilization of the oil market