1 Introduction
Non-ferrous metals play important roles in the development of entity enterprises, the modernization of national science and technology. Nowadays, China's manufacturing industry has a large demand for important non-ferrous metal raw materials at the critical time of realizing the ten-year action program of "Made in China 2025". Therefore, preventing the potential systemic risk caused by the price fluctuation of non-ferrous metals has become an issue that China's financial market should pay attention to. The outbreak of COVID-19 in early 2020 has caused a huge impact on the production and sales of China's manufacturing industry. Many manufacturing enterprises have gone bankrupt due to capital chain disruption, and the price of non-ferrous metals in the futures markets experienced drastic fluctuations after the outbreak. For example, the continuous contract price of copper, an important non-ferrous metal, raised from 37, 200 yuan to 77, 720 yuan during the COVID-19, an increase of 108.92%.
With the development of Chinese financial derivatives market, non-ferrous metal futures varieties are constantly listed, which meet the hedging needs of different production enterprises. The development of the non-ferrous metal industry lays a solid foundation for the non-ferrous metal futures market, and the non-ferrous metal futures market serves the non-ferrous metal industry. For example, copper can meet the manufacturing needs of products including electricity, light industry, transportation, construction, electronic communication and so on. We choose the base metals, which are important to the national economy and social production, including iron, manganese, copper, aluminum, lead, zinc, tin and nickel. In addition, there are two special non-ferrous metals, gold and silver, which are viewed as haven assets, have been rising when stock markets have fallen sharply due to COVID-19 panic. As most financial asset markets tend to have a structural change when big events occur
[1], under the impact of the epidemic, different kinds of non-ferrous metals show different price changes. Thus, this paper uses the eight important non-ferrous metals to represent the non-ferrous metals derivative market, to study the risk spillover relationship between them, particularly paying attention to whether the risk spillover has structural changes after the COVID-19 epidemic.
As GARCH model has good ability to capture the volatility of time series
[2], most of the scholars also used GARCH family models to study risk spillover
[3]. Chen, et al.
[4] used DCC-GARCH model to analyze the nonlinear correlation of price fluctuation and time variation of cotton futures in China and the United States, and calculated the value of risk spillover effect between the cotton futures markets in China and the United States by using the parameters and the CoVaR model. Wen, et al.
[5] combined TGARCH and Copula models to quantitatively measure the extreme risk spillover effect and the asymmetric result between the stock market and bond market after COVID-19. However, it is difficult to estimate the parameters of the model if more financial markets are involved. In addition to GARCH model methods, there are also studies on risk spillovers based on Granger causality test, VAR model and Copula. Xie, et al.
[6] used GED based ARMA-(T) GARCH-VAR model and Granger causality test to investigate the downside and upside extreme risk spillover effects of nine submarkets in China's pan-financial market during different periods. Flavin and Lagoa-Varela
[7] found there was a significant increase in correlation between stock and bond markets during financial crises. Yu, et al.
[8] adopted the dynamic Granger causality method, and found that the outbreak of COVID-19 significantly increased the risk impact of infectious disease uncertainty on the overall volatility of crude oil, copper and gold futures, as well as its asymmetric impact on the volatility of non-livestock futures. Hou, et al.
[9] used CoVaR method to quantitatively measure the extreme risk spillover effect between the stock and bond markets and analyzed the characteristics of the extreme risk spillover effect. He, et al.
[10] revealed the relationship between the interdependence structure and extreme value risk spillover between global stock markets based on Vine Copula method. Zhang, et al.
[11] used the rolling window VAR method to calculate the time-varying volatility spillover index of the Stock markets in Shanghai, Hong Kong and New York after the 'Shanghai-Hong Kong Stock Connect', and then analyzed the size and change of the volatility spillover effect between the two markets. Gong, et al.
[12] used the variance decomposition spillover index based on TVP-VAR model to analyze the dynamic co-activity and risk transmission mechanism of volatility spillover in the financial system.
GARCH family models and VAR model focus on analyzing the relationship between market volatility and the risk spillover between markets, but it is difficult to measure the intensity of risk spillover. Diebold and Yilmaz
[13] adopted the generalized prediction error variance decomposition method, and proposed DY spillover index to calculate the risk spillovers of the financial system. DY spillover index method is not affected by the order of variables and can measure the direction and level of risk spillover between different markets, as well as the total size of risk spillover. Zhong, et al.
[14] used the model and rolling sample estimation method to investigate the tail risk spillover effect between China's stock market and macroeconomic system. Saishree and Padhi
[15] explored the linkage between the commodity and equity markets in India based on DY spillover index. Shahzad, et al.
[16] used the method to analyze the impact of COVID-19 outbreak on US equity sectors. Ziadat, et al.
[17] used the method to examine the inter-and intra-regional stock market relations for the GCC bloc. Guo, et al.
[18] used the DY spillover index to explore the relationship of Bitcoin and other assets. The relationship of other markets is also explored
[19–22].
However, the DY spillover index method ignores the possibility of different risk spillovers in different frequency domains
[23], and it is a static analysis which cannot capture the dynamic changes of market risk transmission. In the long run, non-ferrous metal futures prices more depend on supply and demand situation in its own industrial chain, and other non-ferrous metal prices, while in the short run their prices are more affected by the event impact, investor sentiment, etc. Therefore, in the different length of terms, there may be some difference of the market risk spillovers. To study the risk spillover among non-ferrous metal futures in different periodic terms, this paper adopts the method of ensemble empirical mode decomposition (EEMD) to decompose the prices of non-ferrous metal futures into low frequency term and high frequency term. The former mainly reflects the long-term trend of non-ferrous metal futures' price determined by its own supply and demand. The latter reflects more about short-term fluctuations of non-ferrous metal futures' prices which caused by investor sentiment and market speculation. Through the modeling analysis of decomposed prices, we can get the long-term and short-term risk spillovers among the important non-ferrous metals. Furthermore, the rolling window method can be used for dynamic analysis.
Empirical mode decomposition (EMD) and EEMD methods have been widely used in climate, engineering, finance and other fields, which can mainly help analyze the characteristics of different periodic terms of research objects from the perspective of frequency domain
[24, 25]. For example, Zhang, et al.
[26] proposed a new crude oil price prediction method based on EMD. He, et al.
[27] proposed the CSI 300 index timing strategy based on low-frequency component prediction by combining EEMD and
-insensitive support vector regression (SVR), which significantly outperformed the other strategy in terms of return and risk. Sun, et al.
[28] integrated EEMD, least squares support vector regression (LSSVR) and
-means clustering, and proposed a new foreign exchange rate prediction method, which is called decomposition - clustering - ensemble learning based on EEMD-LSSVR-K. Li, et al.
[29] identified the intrinsic quasi-periodic components of China's real estate market from multiple scales by combining EEMD. Tang, et al.
[30] proposed EEMD-based model to analyze the complexity of clean energy markets. Tang, et al.
[31] forecasted crude oil price based on EEMD. Therefore, EEMD method can help analyze the characteristics of different frequencies
[32–36].
The contribution of this paper is found in the following three aspects: First, we choose base non-ferrous metals, including copper, aluminum, lead, zinc, tin, nickel, and two special non-ferrous metals gold and silver. The former is important to manufacture, and the latter is mostly used as haven assets. Second, we propose a method to analyze the risk spillover of non-ferrous market from frequency domain. The prices of non-ferrous metals futures are decomposed into low-frequency term and high-frequency term by EEMD, which respectively represent the long-term trend and short-term fluctuation. Then the decomposed terms are dynamically analyzed by using dynamic DY spillover index. Third, we focus on the structural changes of risk spillover index during the outbreak of the COVID-19. This method makes up for the shortcoming that the traditional DY spillover index, which cannot analyze the risk spillover from the perspective of frequency domain. The decomposed dynamic DY spillover index provides a reference for preventing price risk contagion and systemic risk outbreak among non-ferrous metals in the future.
The rest structures are as follows: The second part is the data and research methods used in this paper, the third part is the decomposition results of EEMD, the fourth part is the empirical analysis, and the fifth part is the conclusion.
2 Data and Research Methodology
2.1 Data
This article intends to study the changes of risk spillovers among eight non-ferrous metal futures (aluminum, copper, zinc, nickel, tin, lead, gold and silver) in China during the COVID-19 pandemic. This paper selects the closing prices of the main consecutive contracts of non-ferrous metals futures from January 2, 2018 to December 31, 2021, totally 971 trading days. All data is collected from WIND. The outbreak date of COVID-19 is set by World Health Organization (WHO) on 30 January 2020. The chart below shows the price movements of the futures over the sample time range.
Generally, the prices of the futures fell rapidly in the early stage of the outbreak and rose slowly with the further development of the COVID-19. Among them, the prices of aluminum, copper and tin were basically stable before the outbreak of the epidemic (January 2020) with small fluctuation. However, after the outbreak of the epidemic, the prices of aluminum, copper and tin fell sharply at first, and gradually climbed to a new high in recent years from April, with a more significant fluctuation than before. Before the outbreak of the COVID-19, the prices of zinc and lead fluctuated greatly and showed a trend of sharp decline, and their prices also fell greatly when the epidemic broke out, but gradually recovered or remained stable after the epidemic was under control. Gold and silver, as two important investment tools, also behaved differently under the impact of the epidemic. Although the price trends of the two metals were basically consistent, the price of silver had a bigger shock in the early stage of the epidemic and rebounded from the point of 2857 to 6877 within half a year. Comparatively, gold had less fluctuation as it has stronger monetary attribute.
Through the observation of the price trend after the outbreak of the epidemic, it can be found that the structures of futures prices had changed significantly. In this paper, the ensemble empirical mode decomposition method is used to decompose the prices of different non-ferrous metals, and then the decomposed signals are synthesized into low frequency term and high frequency term according to Wilcoxon signed ranks test. The low-frequency term more reflects the price determined by the upstream and downstream industry chain, while the high-frequency term reflects more about the influence of events and investor sentiment. The study of risk spillover from long-term and short-term perspectives is helpful to further analyze the structural changes of non-ferrous metal price risk spillover.
2.2 Ensemble Empirical Mode Decomposition
The price of financial derivatives always shows non-linear and non-stationary characteristic, which does not meet the requirements of linearity and stationarity in general time series analysis. Huang, et al.
[37] proposed Empirical Mode Decomposition (EMD), and compared with traditional data analysis methods such as Fourier transform and wavelet analysis, EMD is self-adaptive and local, independent of any external functions and parameters, and can be useful to analyze the nonlinear and non-stationary data.
EMD can extract intrinsic modes from the original time series, based on the local characteristic scale of data itself, and represent each intrinsic mode as an intrinsic mode function (IMF), which meets the following two conditions:
1) The functions have the same numbers of extrema and zero-crossings or differ at the most by one;
2) The functions are symmetric with respect to local zero mean.
The two conditions ensure that an IMF is a nearly periodic function and the mean is set to zero. IMF is a harmonic-like function, but with variable amplitude and frequency at different times.
The specific steps of EMD method are as follows:
1) Identify all the maxima and minima of time series .
2) Generate its upper and lower envelopes, with cubic spline interpolation, and calculate the point-by-point mean () from upper and lower envelopes.
3) Extract the mean from the time series and define the difference of and as :
4) Check the properties of :
If it is an IMF, denote as the ith IMF and replace with the residual . The ith IMF is often denoted as and the i is called its index;
If it is not, replace with ;
5) Repeat steps 1)4) until the residual satisfies some stopping criterion.
One stopping criterion for extracting an IMF is: Iterating predefined times after the residue satisfies the restriction that the number of zero-crossings and extrema do not differ by more than one and the whole sifting process can be stopped by any of the following predetermined criteria: either when the component or the residue becomes so small that it is less than the predetermined value of a substantial consequence, or when the residue becomes a monotonic function from which no more IMFs can be extracted. The total number of IMFs is limited to , where is the length of data series. The original time series can be expressed as the sum of some IMFs and a residue:
However, some signals are intermittent, modal mixing often occurs during decomposition, that means a single IMF component contains signals of different scales or signals of one scale are distributed in different IMF components, resulting in the lack of physical significance of decomposed IMF components. To overcome this problem, Wu and Huang
[38] developed Ensemble Empirical Mode Decomposition (EEMD). They add different white noise time series to the target data series to construct a set of "enhanced noise signal" time series. Then, each member sequence in the set is decomposed by EMD to obtain the set of each IMF component. Finally, the set average value of each IMF component is calculated as the final truth value of each IMF component. The steps of EEMD method are as follows:
1) Add a white noise sequence to the target sequence.
2) Use EMD to decompose the target sequence with white noise sequence into IMFs.
3) Repeat 1) and 2) iteratively, but with different white noise series each time; and obtain the (ensemble) means of corresponding IMFs of the decompositions as the final result.
2.3 DY Risk Spillover Index
To study the risk spillover between different nonferrous metal futures, this article chose the widely used DY risk spillover index. This method is proposed by Diebold and Yilmaz
[13], by using the vector autoregressive model (VAR) of the generalized variance decomposition principle to calculate the spillover index between different variables. The idea of this method is as follows: Firstly, construct
-order VAR model of m variables:
where is the sequence of variable, is the coefficient matrix, is the independent identically distributed disturbance vector, , is the covariance matrix of the disturbance term vector.
Based on the variance decomposition of generalized VAR, in the variance of -step prediction error of variable , the impact of the fluctuating impact of variable is:
where is the standard deviation of the error term of the th equation in VAR model, is the selection vector, the th element is 1 and the rest are 0. According to the standardization of the matrix elements, the generalized error variance matrix is obtained, and the elements are:
where , .
The directional risk spillovers from all other markets to market as:
And directional risk spillovers transmitted from market to all other markets as:
Then the net spillover is as the follow, which is used to reflect the net risk spillover degree of an asset to all other assets in the system:
3 EEMD Decomposition Results
We decomposed the prices of the non-ferrous metal futures using EEMD method. For example, we can see the decomposition result of copper from the Figure 2. There are five set price signals (imf1, imf2, imf3, imf4 and res) depending on the frequency, its cycle increased from top to bottom, and the signal amplitude increased significantly after the outbreak of COVID-19. The res represents the long-term trend of non-ferrous metal futures prices, which had a stable growth after the outbreak of the epidemic.
Figure 1 Prices of non-ferrous metal futures |
Full size|PPT slide
Figure 2 EEMD decomposition IMFs of copper futures price |
Full size|PPT slide
We then do reconstruction for the IMFs according to the Wilcoxon signed rank test, and compose the IMFs into low frequency term and high frequency term as shown in Figure 3. The high-frequency term basically revolves around the zero, reflecting the fluctuations caused by market sentiment and speculative trading. As it can be seen from Figure 3, the fluctuation of high-frequency term has significantly increased after the outbreak of epidemic. The low-frequency term reflects more about the long-term trend of future prices. After the outbreak of the epidemic, the long-term trend is basically stable and rising.
Figure 3 The components of copper future price |
Full size|PPT slide
4 Empirical Analysis
4.1 Full-Sample Static Risk Spillover Index
Before we do dynamic analysis based on our method, we do full-sample static analysis to obtain an initial risk spillover of non-ferrous metal futures market. Table 1 shows the risk spillover index of full-sample. The big directional risk spillovers index ("From" and "To") indicates that the risk spillover of non-ferrous metal futures market was significant in the sample. The net spillover means that aluminum, nickel, gold and silver tended to transmit risk to others, as well as copper, zinc, tin and lead tended to receive risk from others. Specially, as hedging tools, gold and silver were less affected by others, and the relationship of them was strong. The links among important industrial metals (Aluminum, copper, zinc, nickel, tin, lead) were strong, and they were also affected by gold and silver. Moreover, based on the risk spillover index, we use the Gephi to generate contagion networks in Figure 4, which visually presents the above conclusions.
Table 1 Full-Sample static risk spillover index |
| | Al | Cu | Zn | Ni | Sn | Pb | Au | Ag | From |
| Al | 55.65 | 12.45 | 7.46 | 6.32 | 10.32 | 5.15 | 1.16 | 1.49 | 44.35 |
| Cu | 14.64 | 23.61 | 7.43 | 18.91 | 11.68 | 9.00 | 8.17 | 6.56 | 76.39 |
| Zn | 19.49 | 10.90 | 32.94 | 6.69 | 8.77 | 9.69 | 3.26 | 8.26 | 67.06 |
| Ni | 7.80 | 4.62 | 3.55 | 68.79 | 10.23 | 1.89 | 0.85 | 2.27 | 31.21 |
| Sn | 24.20 | 6.91 | 5.30 | 18.54 | 40.00 | 2.08 | 1.54 | 1.43 | 60.00 |
| Pb | 1.33 | 4.25 | 6.58 | 10.27 | 1.75 | 42.33 | 18.32 | 15.16 | 57.67 |
| Au | 0.04 | 0.41 | 0.35 | 1.03 | 0.09 | 0.72 | 56.08 | 41.27 | 43.92 |
| Ag | 1.79 | 0.78 | 0.54 | 3.63 | 0.62 | 1.42 | 41.44 | 49.77 | 50.23 |
| To | 69.29 | 40.33 | 31.22 | 65.39 | 43.46 | 29.95 | 74.75 | 76.44 | 53.85 |
| Net | 24.94 | −36.06 | −35.84 | 34.18 | −16.54 | −27.72 | 30.82 | 26.21 | 0.00 |
Figure 4 Net Contagion Paths |
Full size|PPT slide
The static risk spillover of full-sample gives us a preliminary analysis. However, it is unable to show the dynamic changes and risk spillovers in different cycles, which is not conducive to our analysis of the impact of the epidemic on market risk spillovers. Thus, we then do the dynamic analysis based on EEMD and dynamic risk spillover index.
4.2 Different Term Dynamic Risk Spillover Index
Based on the components of EEMD of non-ferrous metal futures prices, this paper uses the generalized error variance decomposition of the VAR model to calculate the dynamic DY risk spillover index among non-ferrous metal market from different frequency domain. According to the AIC criterion, the VAR model selects the second-order; the first-order difference sequence of the low-frequency term is stable, and the original sequence of the high-frequency term is stable, so the first-order difference sequence of the low-frequency term and the original sequence of the high-frequency term are modeled respectively. When calculating the generalized impulse response function, the variance decomposition result is gradually stable as the number of forward prediction steps H increases. In this paper, the number of forward prediction steps H is selected as 10. In the calculation of the dynamic risk spillover index, the rolling window is selected as 150 days. We call the dynamic directional risk spillovers from others "inflow index", the dynamic directional risk spillovers to others "outflow index", and net directional spillover index "net index".
4.2.1 Long-Term Dynamic Risk Spillover Index
As shown in Figure 5, the inflow index and outflow index of eight metals remained basically stable, and the net index fluctuated around zero. In the early stage of the outbreak, the outflow index of aluminum metal decreased while the inflow index increased, and there was a small net inflow risk come from other metals. The risk spillover of copper and aluminum is similar, and they both showed a net outflow risk spillover to other metals in the early stages of the outbreak. On the contrary, the spillover index of zinc metal in the early stage of the epidemic increased for a short time and returned to the pre-epidemic level in May 2020. The net index of nickel, tin and lead fluctuated around zero after the outbreak and did not change significantly. Gold remained at a low level of risk outflow effect before the outbreak, and after the outbreak showed a net risk inflow effect, which did not resume as a risk outflow until the second half of 2021. The net index of silver during the epidemic fluctuated periodically around the value of zero, and there was no significant change before and after the epidemic. In general, the epidemic has a little effect on the long-term risk spillover. This may because long-term component of non-ferrous metal futures is mostly determined by the supply and demand relationship of its own industrial chain. The futures prices of non-ferrous metals showed an overall upward trend in the second half of 2020. This is due to the economic recovery of various countries after the inflection point of the epidemic, the increasing demand for industrial raw materials by real enterprises led to the continuous rise of typical copper, aluminum, zinc, and tin metal futures prices. On the contrary, precious metals such as gold and silver are not important industrial metals required for the development of real enterprises. They are mainly used as hedging tools during the economic downturn. Driven by the continuous quantitative easing during the US epidemic, the dollar has depreciated, and the price of gold has continued to rise. In January 2021, Biden came to power and began to encourage vaccination against the COVID-19. The US dollar index bottomed out and began to rise continuously, and the US released a signal of interest rate hikes at the end of 2021. Driven by the rising dollar, the price of gold did not fall sharply in a short period of time, which to a certain extent reflects that despite the appreciation of the dollar, inflationary pressures still exist. So, the price of gold has been controlled and stabilized. Overall, the impact of the epidemic on the demand for non-ferrous metals is relatively stable in the long run. Especially after the normalization of the epidemic, most futures prices have rebounded, which reflects the large demand for non-ferrous metals in the market. Therefore, the long-term risk spillover relationship of various non-ferrous metals is relatively stable, and the changes during the epidemic are small.
Figure 5 Long-term risk spillover index for non-ferrous metal futures |
Full size|PPT slide
4.2.2 Short-Term Dynamic Risk Spillover Index
As shown in Figure 6, the structure of short-term spillover index is different from that of the long-term. After the outbreak of the COVID-19, the risk spillover situation of aluminum metal has not changed significantly. However, the inflow index of zinc increased rapidly, while the outflow index fell rapidly in a short period. The net index rapidly decreased from around 0 to , showing obvious characteristics of risk spillovers from others. Nickel showed a feature of risk spillover to others until the first half of 2019, and the outflow index increased significantly to above 200 starting in the second half of 2019. In general, before the epidemic, nickel showed strong characteristics of spillovers to others when the inflow index remained basically stable. In the early stage of the outbreak, the outflow index gradually recovered to its original low level, and the inflow index remained basically stable, showing a net risk inflow from other assets overall. In the later period, the nickel spillover index gradually increased, but failed to reach the high level before the epidemic, showing a low level of risk spillover to other assets. As safe-haven assets, gold and silver are very similar in short-term risk spillover performance. The inflow index of the two has remained stable around 50 during the epidemic, while the outflow index increased rapidly in the early stage of the outbreak, resulting in the positive net index. In the late stage of the epidemic, the net index of silver basically returned to the level before the epidemic, while the outflow index of gold rose sharply in a short period of time from September 2021 and exceeded the highest level in the recent two years.
Figure 6 Short-term risk spillover index for non-ferrous metal futures |
Full size|PPT slide
In general, the non-ferrous metal future market is more likely to be affected by investor sentiment, speculative transactions, political and economic events in the short term. The short-term risk spillovers of important non-ferrous metals have undergone structural changes to a certain extent after the break of COVID-19.
4.3 Dynamics Total Risk Spillover Index
Figure 7 shows that both the dynamics total risk spillover index ("total index") of long-term and the short-term are basically stable between 40-80, and there is a certain time lag in the spillover response of the COVID-19. We compared the changes in the total index at the beginning of the outbreak. The long-term total index had gradually increased from 39.74 to 61.99 since January 13. The short-term total index has risen rapidly since March 2020, which also corresponds to the rapid growth of the number of infected people in China. At the same time, the stock market panicked and China's stock market fell sharply. The Shanghai Composite Index fell to 2,646.80 on March 19. When this short-term panic and emergencies occurred, the short-term total index rose vertically, reaching the highest point in the recent two years.
Figure 7 Total risk spillover index of non-ferrous metal futures |
Full size|PPT slide
In the financial market, investors divide assets into risk assets and safe-haven assets according to the risk level of the assets they invest in. Gold and silver are scarce resources and have dual attributes of currency and commodities. They are more considered as safe-haven asset in the long run. If the total amount of the investment funds is roughly constant, funds will flow between risk assets and safe-haven assets under different market conditions. The emergence of the COVID-19 will undoubtedly have a huge impact on the real economy, and investors are transferring their funds out of concerns about the future economic situation. In the non-ferrous metal futures market, many investors may transfer their funds from other non-ferrous metal varieties to gold and silver futures after the outbreak, thereby intensifying risk transmission within the non-ferrous metal futures market in the short term. The increase in the total index of the non-ferrous metal future market provided some evidence. When people's expectations for the future economy recover, funds flow back to risky asset markets such as the stock market, and the total index gradually falls.
In general, the long-term risk spillover effect determined by the expected changes in the supply and demand is earlier, but the impact is smaller than that in the short-term; while the short-term non-ferrous metal future prices depend on the overall sentiment of the market. The rapid increase of the risk spillover index in a short period of time basically corresponds to the change of sentiment in China's stock market.
5 Conclusion
From the perspective of different frequency domains, this paper studies the impact of COVID-19 on the risk spillover effect of non-ferrous metal futures in China. In the empirical part, this article selects the main continuous future contract prices of China's main non-ferrous metals (aluminum, copper, zinc, nickel, tin, lead, gold, silver), and uses the method of ensemble empirical mode decomposition to decompose them into low-frequency and high-frequency terms, which represent the long-term trends and short-term fluctuations respectively. Then the dynamic DY risk spillover index model is established based on the decomposed series. Based on the method, this paper analyzed the difference of risk spillover performance of the non-ferrous metals during the COVID-19 from different frequency domain. In the long run, non-ferrous metal futures are more affected by the supply and demand of its own industrial chain. And the long-term risk spillover of each non-ferrous metal futures changed little after the epidemic. In the short run, the risk spillover situation of various non-ferrous metals prices has changed to a large extent with the outbreak of the epidemic. This change more reflects investors' future expectations for the market, the flow of funds from risky assets to haven assets and mood swings intervened by short-term political events.
To sum up, this paper proposes a method to analyze the risk spillover effect between non-ferrous metal futures from both long-term and short-term perspectives and provides a reference for investors or related entities to hedge and avoid risks. For important non-ferrous metals, government should ensure the stability of supply and demand as much as possible to prevent long-term risk spillovers caused by industrial cha, for the long-term trend price fluctuations caused by the supply and demand side will further cause market panic. This panic is reflected in high-frequency price fluctuations through short-term trading, causing structural changes in short-term risk spillovers. And we found that the prices of many important metal futures such as copper, aluminum, and silver fluctuated dramatically when COVID-19 occurred. Therefore, it is vital to stabilize appropriate industrial chain serving the real economy.
In addition, we should list more non-ferrous metal commodity futures and options, develop the derivatives market from both depth and breadth, provide investors with more varieties of investment hedging tools, and prevent the occurrence of systemic risks. Although this article analyzed the change process of the dynamic risk spillover index of non-ferrous metals during the COVID-19, there are certain limitations on the factors that need to be considered for risk spillovers, which can be improved by combining more other factors in the follow-up research.
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
PDF(1844 KB)
Figure 1 Prices of non-ferrous metal futures
Table 1 Full-Sample static risk spillover index
