Analysis of oil price fluctuation under the influence of crude oil stocks and US dollar index — Based on time series network model
Introduction
As a commodity with the biggest impact on the world, crude oil not only occupies a very important proportion in the global trade, but also is closely related to the politics and economy of all countries in the world. As a leading commodity in the market, the influencing factors of oil price are complicated [1], [2]. In recent years, the international crude oil price fluctuates frequently under the joint action of various factors, which increases the risk of the crude oil market and brings great uncertainty to the forecast of crude oil price. Therefore, how to analyze and predict the oil price as accurately as possible is particularly important for academic researchers, policy makers and investors in the oil market [3]. As the world’s largest developing country and a big consumer of population, China’s dependence on foreign crude oil consumption keeps rising, and the change of international oil price will affect the overall economic development. Therefore, it is an important and complicated problem to explore the oil price fluctuation and its driving factors.
In the past few decades, more and more literature has been devoted to studying the relationship between crude oil price and their influencing factors. The violent fluctuations should be driven by economic crisis, crude oil supply and demand, stocks, OPEC behavior, the US dollar exchange rate, military conflicts, extreme weather, speculative trade, psychological expectations and other factors [4], [5], [6], [7], [8], [9], [10]. For example, Kilian [5] identified potential demand and supply shocks for the global crude oil market, suggesting that these shocks are not only important for explaining the fluctuations in the actual oil price, but also the dynamic impact of these shocks on the actual oil price. Miao et al. [6] selected six categories of factors (including supply, demand, financial markets, commodity markets, speculation and geopolitics) and tested their importance in the context of estimating various prediction models. Zhang et al. [9] analyzed the impact of the US dollar exchange rate on international crude oil prices from the perspective of market transactions, and determined that there is a significant long-term equilibrium cointegration relationship between the two markets. It is found that the depreciation of the dollar is a key to pushing up international oil prices. Mobert [11] believed that speculators’ belief dispersion has a significant impact on the New York Mercantile Exchange’s crude oil prices and their volatility, and further confirmed the role of speculation as a precursor to price fluctuation. In general, international crude oil price trend of mid-long range is only affected by other benchmark trends, and still conforms to the law of price determination between supply and demand [12]. The volatility of short-term oil prices is usually higher than that of long-term oil prices [13], while the short-term is subject to major fluctuations due to factors such as financial fluctuations and unexpected events. Therefore, in the context of increasing international crude oil price volatility, the study of the short-term fluctuations in crude oil prices and the relationship between their influencing factors will play an important role, while the short-term forecast of oil prices will also face more challenges.
However, research on the dynamics of crude oil prices and their influencing factors is still focused on econometric methods such as vector autoregressive models (VAR), cointegration tests, univariate and/or multivariate GARCH regression frameworks. Sadorsky [14] used the VAR model to reveal the negative correlation between oil prices and stock returns. Bollerslev [15], [16] proposed ARCH and GARCH models to capture the fluctuation transmission between oil prices and foreign exchange rates. Li et al. [17] used the econometric method of cointegration and causality test to study the relationship between oil price and Chinese stock market from the industry level. Although the cointegration and VAR framework examines the long-term and short-term relationships, it ignores the time-varying fluctuations between variables. The traditional univariate GARCH model can only characterize the fluctuation of a single variable, while the multivariate GARCH needs to guarantee some key mathematical features. It is worth noting that the crude oil market shows various intrinsic interaction factors with different time scales, such as short-term fluctuations, significant mutations with medium-term effects, and long-term fluctuations [18], [19], [20], [21]. Therefore, Huang et al. [22] proposed an empirical mode decomposition method that can decompose any complex data set into a finite intrinsic mode function. Yu et al. [23] decomposed the raw data into a series of modalities at different time scales, analyzed the linkage relationship between the carbon market and the crude oil market, and find out the main internal factors and modes that make important contributions to the co-movement mechanism.
In recent years, the research of complex networks is infiltrating into many different fields such as mathematics, life sciences and engineering disciplines, especially in the field of time series analysis. Map time series to complex networks, explore the dynamics of time series from network organization, and study the topological properties of networks from a statistical perspective [24]. Zhang et al. [25] constructed a complex network using pseudo-periodic time series, and found that time series with different dynamic characteristics have different topologies and standard metrics of complex network structures can be used to distinguish different time series dynamic states. An et al. [26] introduced the coarse granulation method to map time series into complex networks, explored the fluctuation law of crude oil price autocorrelation, and helped to understand the mechanism of crude oil price fluctuations, using autocorrelation coefficients, symbolization and coarse grains. The process defines the autocorrelation volatility model and provides an idea for studying the fluctuation of univariate autocorrelation.
Because we need to overcome the shortcomings of econometrics and also want to refine the fluctuation characteristics of short-term crude oil prices and their influencing factors, meanwhile it has not been common to study the oil price after decomposition in the previous literature. Therefore, in this paper, we analyze the time-varying characteristics of high-frequency oil price and its main influencing factors(crude oil stocks and US dollar index)from the perspective of complex networks. Combined with the actual events, we attempt to study the inherent law of co-movements of high-frequency oil price, stocks and the US dollar index in periods with different volatility characteristics.
Main novel contributions in our studies are as follows: First, from the perspective of complex networks, the fluctuating intensity, coarse granulation and symbolization process are used to define the co-movement modes model and analyze the time-varying relationship of co-movement in different periods, revealing some features that cannot be captured by econometric methods. Secondly, grasping the characteristics of oil price fluctuations in inverse relationship with stocks and US dollar index, the conversion of the co-movement modes during different fluctuation periods was determined, and the difference of each mode conversion is discussed. Third, by analyzing the nature of the network, the key information hidden in time changes is extracted. At the same time, the important influence of the US dollar index in the period of sharp fluctuations in oil prices and the long-term role of stocks in short-term fluctuations in oil prices are also highlighted. The overall methodology flowchart is shown in Fig. 1.
The structure of this paper is as follows: Section 2 briefly introduces the selection and processing of data. Section 3 introduces the research method, establishes the model of co-movement threshold networks of high-frequency oil price, stocks and the US dollar index, and analyzes the nature of the network. Section 4 summarizes the empirical results. The final section gives the conclusions of this paper.
Section snippets
International crude oil price
Fig. 2 shows the overall trend of the Brent crude oil spot price from 1987 to 2017. Before the 21st century, the international crude oil market was relatively calm, and the overall situation was at a low level. In the 21st century, the international crude oil price fluctuated drastically, especially with the September 11 attacks as a dividing line. After 1998, China’s prices for crude oil products are basically linked to international oil prices, which makes it more practical to study the
Decomposition and reconstruction
Ensemble Empirical Mode Decomposition(EEMD)is an improved algorithm of Empirical Mode Decomposition(EMD) [37]. EMD is an adaptive space–time analysis method suitable for dealing with non-stationary nonlinear sequences. It can decompose complex signals into a finite number of Intrinsic Mode Functions (IMFs), and extract the features associated with various natural time scales [38]. This method can decompose the Brent crude oil price series into independent IMFs and analyze the crude oil price
Visualization of networks in various periods
Through the steps of constructing the network above, we obtain the high-frequency oil price, stocks and the US dollar index co-movement sub-networks corresponding to the six periods (Fig. 10). We find that the relationship between these nodes of six networks is closer, especially in the period of high-frequency oil price drastic fluctuations. In addition, we can also see that the co-movement modes during the entire research period are also quite numerous, of which the EEE mode occupies the main
Conclusions
This paper selects the Brent high-frequency oil spot price from 2000/1/7–2017/12/29 and the factors (crude oil stocks and the US dollar index) affecting the short-term fluctuation of crude oil price as sample data and determines the appropriate threshold. According to the pairwise relationship among the three sets of data, the fluctuation mode is taken as the time node and the transformation between nodes is taken as the edge, the co-movement threshold networks of high-frequency oil price,
CRediT authorship contribution statement
Jie Zhou: Conceptualization, Methodology, Software, Investigation. Mei Sun: Supervision. Dun Han: Writing - Review & Editing. Cuixia Gao: Resources.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgment
This research was supported by the National Nature Science Foundation of China (Grant No. 71774070).
References (47)
- et al.
The crude oil market and the gold market: Evidence for cointegration, causality and price discovery
Resour. Policy
(2010) - et al.
Influential factors in crude oil price forecasting
Energy Econ.
(2017) - et al.
Modeling and forecasting energy consumption in China: Implications for Chinese energy demand and imports in 2020
Energy Econ.
(2008) - et al.
Spillover effect of US dollar exchange rate on oil prices
J. Policy Model.
(2008) - et al.
Oil prices, speculation, and fundamentals: Interpreting causal relations among spot and futures prices
Energy Econ.
(2009) - et al.
Component structure for nonstationary time series: Application to benchmark oil prices
Int. Rev. Financ. Anal.
(2008) Oil price shocks and stock market activity
Energy Econ.
(1999)Generalized autoregressive conditional heteroskedasticity
J. Econometrics
(1986)- et al.
Oil prices and stock market in China: A sector analysis using panel cointegration with multiple breaks
Energy Econ.
(2012) Oil shock and economic growth in Japan: A nonlinear approach
Energy Econ.
(2008)
Crude oil price analysis and forecasting using wavelet decomposed ensemble model
Energy
A compressed sensing based AI learning paradigm for crude oil price forecasting
Energy Econ.
Linear and nonlinear granger causality investigation between carbon market and crude oil market: A multi-scale approach
Energy Econ.
From time series to complex networks: The phase space coarse graining
Physica A
The role of fluctuating modes of autocorrelation in crude oil prices
Physica A
Analysis of the impact of crude oil price fluctuations on China’s stock market in different periods—Based on time series network model
Physica A
The empirical role of the exchange rate on the crude-oil price formation
Energy Econ.
Quantifying the speculative component in the real price of oil: The role of global oil inventories
J. Int. Money Finance
Interpolatory rational cubic spline with biased, point and interval tension
Comput. Graph.
Estimating restricted structural change models
J. Econometrics
Estimating the impact of extreme events on crude oil price: An EMD-based event analysis method
Energy Econ.
Features and evolution of international crude oil trade relationships: A trading-based network analysis
Energy
Complex network-based time series analysis
Physica A
Cited by (14)
The impact of consumer confidence on oil prices
2023, Energy EconomicsThe main transmission paths of price fluctuations for tungsten products along the industry chain
2023, Resources PolicyCitation Excerpt :The traditional single econometric method cannot solve this complex problem of price fluctuation transmission, and it is necessary to combine multiple methods innovatively. Price fluctuation states have internal dynamics, such as networks, modals, chaos, etc. (Xu et al., 2018; Zhou et al., 2021). The econometric models also cannot describe the network characteristics of the price fluctuation transmission.
Analysing and forecasting co-movement between innovative and traditional financial assets based on complex network and machine learning
2023, Research in International Business and FinanceCitation Excerpt :Most studies mentioned above primarily focus on the risk spillover between innovative and traditional financial assets using typical econometric models, such as vector autoregressive (VAR), autoregressive conditional heteroskedasticity (ARCH), cointegration tests, and causality tests. These models are based on linear regression and are usually used to study the long-term relationships between variables (Zhou et al., 2021). However, the co-movement among financial assets varies over time, forming a nonlinear and nonstationary complex system.
The research on modeling and application of dynamic grey forecasting model based on energy price-energy consumption-economic growth
2022, EnergyCitation Excerpt :For example, Barunik et al. [8] used the dynamic Nelson-Siegel model to explain the term structure of crude oil price and used the generalized regression framework of the neural network to forecast the future crude oil price by taking 24-year crude oil futures price as an experiment. Zhou et al. [9] provided a new idea for studying oil price fluctuation and its influencing factors in the future by putting forward the time series network model and using the high-frequency oil price decomposed by aggregate empirical mode and combining crude oil stocks and US dollar index. Zhao et al. [10] put forward a vector trend prediction method, which predicts the future trend of crude oil prices based on the vector trend sequence of crude oil historical prices, and provides suggestions for oil market investors to understand the trend of oil prices and investment decisions.