key: cord-0990738-7npsx2bk
authors: Zhang, Yongmin; Ding, Shusheng; Shi, Haili
title: The impact of COVID‐19 on the interdependence between US and Chinese oil futures markets
date: 2022-04-08
journal: Journal of Futures Markets
DOI: 10.1002/fut.22326
sha: 642fc4cb2e553c2a358632fc536edddcbcf7f35a
doc_id: 990738
cord_uid: 7npsx2bk

The oil futures market plays a vital role in the global financial system, especially after the negative future oil price rose during the COVID‐19 pandemic. This paper investigates the COVID‐19 impact on the interdependence between the US and Chinese oil futures markets by extending the dynamic conditional correlation‐generalized autoregressive conditional heteroskedasticity (DCC‐GARCH) models with incorporating COVID‐19 variables and by applying vector autoregression (VAR) models. Our study reveals that the COVID‐19 pandemic enhanced the long‐run correlation between the two oil markets. In contrast, daily changes in pandemic severity had a negative effect on the short‐term transient correlation. Our results show that COVID‐19 changed the one‐direction causality from the US oil market to the Chinese market in the pre‐COVID period to a bidirectional causal relation between the two markets during the COVID period. It strengthened the volatility spillover effect from the Chinese to US markets. These findings are helpful to regulars' monitoring oil supply chain risk and investors' cross‐market hedging of spillover risks from a systematic risk perspective.

The above findings imply that the linkage between the US and Chinese oil futures markets became stronger after the outbreak of COVID-19 because the US oil futures traders followed information from the Chinese oil market more closely during the COVID period than during the pre-COVID period. Such enhanced information transmission from the Chinese market to the US market during the COVID period is consistent with the following facts. COVID-19 was initially found in China, a world manufacturing center and major oil importer, and supply chain disruptions due to the pandemic lockdown in China led to a huge negative impact on global commodity markets and the worldwide economy. Furthermore, China is one of a few countries that controlled the spread of COVID-19. The Chinese economy plays a vital role in stabilizing global supply chains and commodity markets, while the rest of the world is severely impacted by the COVID-19 pandemic.

The remainder of this paper is organized as follows: Section 2 presents the data and descriptive statistics. Section 3 scrutinizes the COVID-19 impact on the dynamic correlation between the US and Chinese oil markets by extending the DCC-GARCH models with the inclusion of COVID-19 variables. Section 4 develops VAR models to study the impact of COVID-19 on causality and spillover effects across two markets. Section 5 concludes our paper.

We consider the daily closing prices for NYMEX's oil in the US and oil futures in the Shanghai Futures Exchange (SFE) in China, which started to trade on March 26, 2018. Normally, futures contracts are combined using the nearest futures approach to become a continuous contract. In particular, the data of a futures contract with nearest maturity (usually in 1 month) has been used until its expiration and then the subsequent contract is used continuously until its expiration, and so on.

The outbreak of COVID-19 was announced on January 17, 2020. We select the period from March 26, 2018, to January 17, 2020, as the pre-COVID period and from January 18, 2020, to July 2, 2021, as the ongoing COVID-19 period. We use oilus and oilc to represent US and Chinese oil futures prices. We convert Chinese oil futures prices from RMB to US dollars to be consistent with US oil futures prices in currency. Table 1 displays the descriptive statistics for US and Chinese oil futures prices for both pre-COVID and ongoing-COVID periods. We observed that in both periods, the mean prices for US oil futures were lower than those for Chinese oil futures, while the standard deviations for US oil futures were higher than those for Chinese oil futures. More importantly, we find that both US and Chinese oil prices decreased by approximately 30%, while standard deviations increased by approximately 100% after the outbreak of COVID-19. This indicates that COVID-19 has significantly impacted the demand for oil in a negative direction, while it dramatically increases the price risk. Table 2 displays the descriptive statistics for US and Chinese oil futures returns for both periods.

We estimate the dynamic correlations between US and Chinese oil futures risks by employing the DCC-GARCH model proposed by Engle (2002) . Let D be a dummy variable that takes the value of 0 for the pre-COVID period and the value of 1 for the ongoing COVID period. We also denote w t as the new COVID-19 cases that were reported to the WHO on Day t. Then, we propose the following model to investigate the COVID-19 impact on the covariance dynamics between two oil futures returns:

(1) where Cov t is the covariance of US and Chinese oil futures returns on Day t, ε t l − is the residual from US oil return based on the DCC-GARCH model, εt l − is the residual from Chinese oil return based on the DCC-GARCH model, and α 0 , α 1 , μ k , λ l , θ are coefficients.

In the above model, α 1 indicates the impact of COVID-19 on long-run covariance, while θ presents the COVID-19 impact on short-term covariance.

We further define Corr = t Cov σ σ̃t t t as the dynamic correlation between US and Chinese oil futures return. Now we propose the correlation model as following:

where σ t is the standard deviation of the US oil futures returns at Day t and σ̃t is the standard deviation of the Chinese oil futures returns at Day t. In the above model, β 1 indicates the impact of COVID-19 on long-run correlation, while θ′ presents the COVID-19 impact on short-term transient correlation. Figures 1 and 2 deliver the covariance and correlation evolutions between US and Chinese oil futures returns from March 26, 2018, to July 2, 2021. From Figure 1 and Figure 2 , we find that there is a huge covariance spike at the beginning of March 2020 when the WHO announced the beginning of the pandemic, and the correlation became much more volatile during the COVID-19 period. The Model (1) covariance fitting results in Table 3 show that the coefficient for the dummy variable D is positively significant, and the coefficient for DW t−1 is insignificant. Table 4 presents the correlation fittings results for Model (2) which show the coefficient for the dummy variable D is also positively significant, but for the coefficient for DW t−1 is negatively significant. This finding reveals that the COVID-19 pandemic strengthened the linkage between the two oil futures markets from a long-run perspective. In contrast, the daily new COVID-19 cases had little impact on covariance or has negatively significant impact on short term (transient) correlation. Since US and China had very different COVID severity and prevention measures, the negative (or insignificant) impact of COVID-19 on the transient correlations (or covariances) could be the reflection of difference in short term oi demand under the different pandemic effect on daily business activities in the two countries. 

In this section, we develop VAR models to analyze the impact of COVID-19 on causality relations and return and volatility spillover effects across US and Chinese futures markets.

Before we built the VAR models, we conducted ADF tests for the stationarity of oil futures returns for US and Chinese markets. Table A1 shows the ADF test results, which indicate that both futures returns were stationary time series in both the pre-COVID and ongoing-COVID periods. Here the daily returns are calculated as: Abbreviations: cov_oil, covariance between US and Chinese markets; dummy, dummy variable; L., lagged term; resi, residual term.

***p < 0.01, **p < 0.05, and *p < 0.1.

where P t and P t−1 are the oil prices on Days t and t − 1, respectively. We denote R t u and R t c for US and Chinese oil futures returns at Day t. Next, we used LR statistics and FPE, AIC, HQIC, and SBIC information measures to determine the optimal number of VAR lags. Table A2 presents the above five measures for different lags for the pre-COVID and ongoing-COVID periods. From Table A2 , we find that optimal lags were 3 for the pre-COVID period and 4 for the ongoing-COVID period. Then, we establish VAR models as follows:

pre-COVID period VAR model: and ongoing-COVID period VAR model: Roots of the VAR comparison matrix were all within the unit circle for the pre-COVID and ongoing-COVID periods, respectively. Therefore, our VAR systems were stable in both periods. Table 5 lists four Granger causality tests for the US and Chinese oil futures returns in two periods. We discover the fact that US oil futures return moves are a Granger cause for Chinese oil futures return moves but not in the reverse direction before the outbreak of COVID-19. However, the two markets become a mutual Granger cause for each other regarding the return move during the COVID period. Abbreviations: corr_oil, correlation between US and Chinese markets; dummy, dummy variable; L., lagged term; resi, residual term.

***p < 0.01, **p < 0.05, and *p < 0.1.

This interesting finding could be explained as follows. China is one of a few countries that contained COVID-19, while the rest of the world is under the active spread of the virus. Therefore, China plays a more significant role as the world manufactory center and the critical player in the world supply chain during the COVID period than during the pre-COVID period due to the pandemic lockdown in most countries. Thus, global investors are more sensitive to Chinese oil demand during the COVID period than during the pre-COVID period. Consequently, the Chinese oil futures market is also a Granger cause for the US oil futures market and vice versa after the outbreak of COVID-19.

Impulse and response analyses were undertaken for the US and Chinese oil futures returns VAR systems to study the response of one oil market return when the other market return was under a one standard deviation shock, before and after the COVID-19 outbreak in China. Figures 3 and 4 show impulse and response functions for US and Chinese oil futures returns for pre-COVID and ongoing-COVID periods. From Figure 3 , we observe the following. In the pre-COVID period, Chinese oil futures returns jump to 0.01 in response to the US return shock of 0.02. In contrast, US oil futures returns jump to approximately 0.005 in response to a positive 0.01 shock in Chinese futures returns. Moreover, the Chinese market has a daily delay in responding to US impulse shocks, and this response lasts approximately 6 days. In comparison, the US response to Chinese impulse shocks lasts only 3 days. This asymmetric impulse and response between the two markets is in consistent with our finding in Section 4.2 that the US oil market was a Granger cause for the Chinese market before the outbreak of COVID-19. The slower response of the Chinese market to the US shock than the US response to the shock from the Chinese market also reveals that the Chinese oil futures market is not as efficient as the US oil futures market.

During the COVID period in Figure 4 , one-standard deviations of impulsive shocks are 0.06 for the US market and 0.02 for Chinese oil futures, which are much more than their one-standard deviations of 0.02 for the US market and 0.01 for the Chinese market in the pre-COVID period. This indicates that COVID-19 has brought more uncertainty to both oil futures markets, such as the negative oil futures price at the beginning of the pandemic. Furthermore, each market's response to the impulsive shock from the other market lasts 8 days, which is more than the lasting days in the pre-COVID period. The above results show that the outbreak of COVID-19 caused a dramatic shock move to people's perceptions of the world economy. Both the US and Chinese oil futures markets responded with their volatilities in a strong and persistent way.

Finally, the similar impulse and response patterns from the two markets during the COVID period are in consistent with our results in Section 4.2. Both US and Chinese oil futures returns are a Granger cause to each other after the outbreak of COVID-19.

Variance decomposition of standard errors for the two oil futures VAR systems will give a detailed decomposition of the standard error of each oil return into percentage contributions from the variations of both futures returns. These variance decompositions give a quantitative comparison of how strongly the US market affects the Chinese market and vice versa. Tables 6 and 7 display the variance decomposition results for each oil return's standard error in the pre-COVID and ongoing-COVID periods, respectively, from which we can make the following observations. The standard errors of Chinese oil market returns can be attributed to US market variations with contributions of 50% and 25% in the pre-COVID and ongoing COVID periods, respectively. On the other hand, the standard errors of US oil returns have 5% and 10% contributions from Chinese markets before and after the outbreak of COVID 19. Therefore, the increased influence of the Chinese market on the US market confirms the findings from the Granger causality test in Section 4.2. The results demonstrate that the increase Chinese impact on the US market after the COVID-19 outbreak is due to the change in one-direction causality from the US market to the Chinese market in the pre-COVID period to the bidirectional causal relation between the two oil futures markets during the COVID-19 period. Finally, from the results of Grange tests, impulse and response analysis and variance decomposition in Section 4, we can infer that COVID-19 has made the interdependence between the US and Chinese oil markets significantly strong, which is also indicated by the correlation analysis in Section 3.

This paper has investigated the impact of COVID-19 on the interdependence between the US and Chinese oil futures markets. The study of the pandemic impact on dynamic cross-market correlations is carried out by extending DCC-GARCH models to incorporate COVID-19 influencing variables. The impact of the pandemic on causality and spillover effects between two oil futures markets has been analyzed by the Granger test, impulse and response functions, and variance decomposition method under the VAR framework. Our research shows that the COVID-19 pandemic enhanced the long-run correlation between US and Chinese oil futures returns, while the daily new cases of COVID-19 had little or negative effect on short-term transient correlation. Furthermore, we demonstrated that COVID-19 has also enhanced the causal relation from one-directional causality (from the US to China) to bidirectional causality between the two futures markets. In addition, the Chinese oil futures market's response to US market impulsive shocks lasted longer than the US response to Chinese shocks in the pre-COVID period. On the other hand, however, both markets' responses to the other market's impulse lasted a similar amount of time during the COVID-19 period. Finally, COVID-19 has also significantly increased the volatility spillover effect from the Chinese oil market to the US oil market.

This paper is supported by Zhejiang Philosophy and Social Sciences grant (21NDJC058YB), Ningbo-CASS collaborative grant (NZKT201701) and International Research Center for Sustainable Finance.

These data were derived from the following commercial data provider WIND available in the public domain: https:// www.wind.com.cn/newsite/data.html ORCID Yongmin Zhang http://orcid.org/0000-0002-4221-5603 

The effect of oil price shocks on asset markets: Evidence from oil inventory news

The relationship between US and Canadian wheat futures

Price overractions in the commodity futures market: An intraday analysis of the Covid-19 pandemic impact

Primary commodity prices: Co-movements, common factors and fundamentals

Risk spillovers and time-varying links between international oil and China's commodity futures markets: Fresh evidence from the higher-order moments

The spillover effects of commodity financialization: Evidence from commodity market volatility

Cross market predictions for commodity prices

Dynamic conditional correlation: A simple class of multivariate generalized autoregressive conditional heteroskedasticity models

Co-movement of international crude oil price and Indian stock market: Evidences from nonlinear cointegration tests

Testing stock market linkages for Poland and Hungary: A multivariate GARCH approach

Price linkages between Chinese and world copper futures markets

Oil stocks, risk factors, and tail behavior

Capturing the dynamics of the China crude oil futures: Markov switching, co-movement, and volatility forecasting

South African political unrest, oil prices, and the time varying risk premium in the gold futures market

The excess co-movement of commodity prices

Volatility dynamics under an endogenous Markov-switching framework: A cross-market approach

Connectedness between oil and agricultural commodity prices during tranquil and volatile period. Is crude oil a victim indeed

Index investment and the financialization of commodities

Dynamic integration and transmission channels among interest rates and oil price shocks. The Quarterly Review of Economics and Finance

How important are common factors in driving non-fuel commodity prices? A dynamic factor analysis (European Central Bank Working Paper 1072)

Asymmetric volatility spillovers between economic policy uncertainty and stock markets: Evidence from China

The price impact of futures trades and their intraday seasonality

Modelling asymmetric conditional dependence between Shanghai and Hong Kong stock markets

Return and volatility transmission between China's and international crude oil futures markets: A first look

The return and volatility co-movement in commodity futures markets: The effects of liquidity risk

Liquidity effects on price and return co-movements in commodity futures markets

Policy impact on volatility dynamics in commodity futures markets: Evidence from China

A key determinant of commodity price co-movement: The role of daily market liquidity

COVID-19's impact on the spillover effect across the Chinese and U.S. stock markets

COVID-19 impact on commodity futures volatilities