The Asymmetric Tail Risk Spillover from the International Soybean Market to China’s Soybean Industry Chain

Abstract

China is the largest soybean importer and consumer in the world. Soybean oil is the most-consumed vegetable oil in China, while soybean meal is the most important protein feed raw material in China, which affects the costs of animal husbandry. Volatility in the international soybean market would generate risk spillovers to China’s soybean industrial chain. This paper analyzed the channel of risk spillover from the international soybean market to China’s soybean industry chain and the asymmetry of the risk spillover. The degree of risk spillover from the international soybean market to the Chinese soybean industry chain was measured by the Copula–CoVaR model. The moderating role of inventory and demand in asymmetric risk spillovers was analyzed by quantile regression. We draw the following conclusions: First, the international soybean market impacts China’s soybean industry chain through soybeans rather than soybean meal and oil. The price fluctuation of China soybean market is obviously lower than that of the international soybean market. Second, there are apparent asymmetric risk spillovers from the international soybean market to China’s soybean industry chain, especially the soybean meal market. Third, increasing the Chinese soybean inventory and growing demand could effectively prevent the downside risk spillover from international markets to China’s soybean market. This also explains the asymmetry of risk spillovers. The research enriches the research perspective on food security, and the analysis of risk spillover mechanisms provides a scientific basis for relevant companies to develop risk-management strategies.

1. Introduction

China is the world’s largest importer of soybeans, which are China’s most crucial grain imports, accounting for more than 60 percent of China’s total grain imports. The Chinese government has attached increasing importance to increasing domestic soybean production capacity in recent years, implementing various measures such as subsidies and incentives for local governments. However, with a large population and little arable land per capita, there is limited room for the Chinese soybean production capacity to increase. Thus, China is likely to rely on the international soybean market for a long time. Moderate soybean imports are also in line with China’s comparative advantage [1].

China’s high reliance on the international soybean market could lead to price risk exposure. Dramatic price fluctuations in international agricultural markets are directly transmitted to countries with high import reliance, impacting the agricultural markets of importing countries [2,3]. China is the world’s largest commodity trader but it does not have a discursive power in prices and is a passive recipient of prices [4]. Extreme international soybean price risks have been frequent in recent years due to geopolitical conflicts and international shipping risks [5].

The extreme price risk is disruptive to China’s soybean industry chain. Soybean oil is the most-consumed vegetable oil and soybean meal is the most important protein feed raw material in China. Extreme price risks in the international soybean market would impinge on the feed costs of China’s livestock industry and the price of edible soybean oil. In addition, the response time to extreme market risks is short, and it is difficult for enterprises to deal with [6]. China’s soybean processors, highly dependent on imported soybeans, would go bankrupt in the face of soaring and falling international soybean prices [7]. Therefore, it is of great significance to measure the tail risk spillover degree of China’s soybean industry chain from the international soybean market and analyze its spillover characteristics to ensure food security in China. However, existing research has not paid enough attention to the tail risks of China’s soybean industry chain.

The international soybean market has a risk spillover effect on the Chinese soybean industry chain, mainly through soybean rather than its by-product market. Soybean by-products are soybean meal and soybean oil. In the past decade, more than 85 percent of China’s soybean imports have been used for crushing to produce soybean oil and meal, and the amount of soybeans used for crushing in 2022 even exceeds the total amount of soybeans imported. China’s soybean meal imports accounted for less than 1% of its consumption, and soybean oil imports accounted for up to 7% of its consumption. Consequently, China’s soybean meal and oil are mainly produced from imported soybeans rather than being directly imported. The fluctuation of the international soybean price would generate a direct impact on China’s soybean industry chain. While there is an indirect price linkage between the international soybean by-product market and China’s soybean industry chain [8]. This article, therefore, selected an international soybean price to represent the international soybean market.

China’s soybean industry chain involves a wide range of upstream for soybean imports and production, midstream for soybean processing (mainly for soybean meal and soybean oil crushing), and downstream for soybean meal feed processing and edible oil production [9,10]. Soybean prices affect soybean meal and soybean oil prices, while soybean meal prices affect feed prices. The fluctuation of international soybean prices would affect the feed costs of China’s livestock husbandry and edible oil prices, affecting residents’ living costs. This article uses soybean, soybean meal, soybean oil and feed to represent the Chinese soybean industry chain.

From the industrial chain perspective, the soybean industry’s market risk management can effectively grasp the overall risk level of China’s soybean industry, identify vulnerable points, and highlight key risk-management sectors [11]. Fluctuations in international soybean prices cause fluctuations in China’s soybean prices, ultimately impacting its livestock husbandry and edible vegetable oil markets. Therefore, based on the industrial chain perspective, managing the tail risk of soybean prices is very significant. Tail price risk is a low-probability event, but it often results in the largest losses for businesses and is a significant cause of bankruptcy for many businesses [12]. Focusing on the tail risk spillover from the international soybean market to China’s soybean industry chain is an important initiative to enhance the resilience of China’s soybean industry chain and ensure food security. The study focused on the tail risk spillover degree, risk spillover path, and asymmetric mechanism from the international soybean market to China’s soybean industry chain. This research can provide a basis for tail price risk management for Chinese soybean processors and a targeted strategy for the Chinese government to support the development of the soybean industry.

2. Literature Review and Theoretical Analysis

2.1. Analysis of the Causality of Risk Spillovers

Since joining the World Trade Organization in 2001, the correlation between Chinese and international soybean prices has risen significantly [13]. International grain price volatility would have an enormous impact on grain markets that are notoriously dependent on grain imports [14]. The existing literature typically uses econometric methods such as correlation, cointegration analysis, Granger causality analysis, vector autoregressive models (VARs), and error correction models (ECMs) to explore the linkages between grain prices in developing countries and international grain prices [15,16,17]. The Copula approach and quantile regression are commonly used to measure tail risk in the financial sector and are less used in the agricultural product market [18,19]. In China’s grain market, the prices of corn, soybeans, wheat, and rice have a long-term cointegration relationship with international grain prices, in which the integration of corn, wheat, rice, and the international market is low. In contrast, the soybean market has the highest degree of integration with the international market [10]. The Chinese soybean market is subject to the relatively strong influence of fluctuation in the international market [20].

There are three main risk-transmission paths for the international soybean market to China’s soybean market: Firstly, through spot trading in international trade—if the two markets are fully integrated, then the price difference of soybeans will be equal to the transportation cost [2]. In addition, trade is the primary path linking China’s soybean market with the world’s soybean market [21]. The second is information transmission [22]. Soybeans are one of the mature international bulk agricultural products, and spot trading of soybeans is based on future market prices, which are also fed back into the future market. The third is the short-term impact of international speculative behavior. Speculative behavior will exacerbate short-term fluctuations in soybean futures prices. China chiefly imports soybeans, with smaller imports of soybean by-products. This is why this essay chose the international soybean market instead of its by-product market as the research object.

Price risk spillover refers to the transfer of price risk from one market to another and is commonly used in the risk analysis of financial and commodity markets [23]. It includes risk spillovers between markets in different sectors as well as between markets in the same category and chain [24,25]. Many researchers are analyzing price risk spillovers between agricultural markets from two perspectives: mean risk spillovers and volatility risk spillovers [26,27]. The results show a mean spillover effect and a volatility spillover effect between the Chinese and international soybean markets [28]; the structural change is related to geographical conflict [29]. In addition, the risk spillovers from international grain markets to developing country grain markets are exacerbated by extreme price volatility in international markets [6]. The tail price risk of bulk commodities would cause economic crisis and sovereign risk in emerging market countries [30]. However, there is little research on the tail risk of the international soybean market to the Chinese soybean market.

2.2. Risk Spillover Asymmetry Mechanisms

The asymmetry of risk spillovers across commodity markets, whereby prices rise significantly faster than they fall, is universal, but its specific causes are distinct and different [31]. Geopolitical conflicts could cause extreme fluctuations in oil prices, and developed countries are the net transmitters of risks [32]. The international crude oil market has had a stronger upward risk impact on China metal market [25]. Existing studies have suggested that the upward risk spillover of international soybean prices to Chinese soybean prices is significant, while its downward risk premium is not [28]. There are three crucial causes of the asymmetric risk spillover.

First, the Chinese government formulated policies to support the soybean industry and safeguard domestic soybean production [33]. When the purchase price of soybeans is lower than the target price, the government would purchase them at the target price. When the price of soybeans is too high, the government would reduce its inventory and sell soybeans. Storage policy, purchase at target price, and subsidies to soybean farmers can effectively support China’s domestic soybean prices, thereby reducing the impact of the decline in international soybean prices. Policies such as purchasing and storage could be reflected in the changes in soybean inventory. Second, China’s demand for soybeans is enormous and rapidly increasing. China’s soybean consumption grew from 65.43 million tonnes in 2010 to 115.53 million tonnes in 2022, with an average annual growth rate of 14.7 percent. Soybean is the most critical protein feed ingredient and an important source of vegetable oil in China. Rising living standards in China increased demand for meat, eggs, milk, and edible oil, leading to an upsurge in soybean consumption [34]. Third, compared to international grain traders, China’s grasp of the soybean industry chain is weak, and its power of discourse in the international soybean market is deficient [35]. The trade role of the country has a significant cross-influence on trade prices in all links of the industrial chain [36].

There is a large number of studies analyzing the asymmetry of risk spillovers between financial markets and crude oil markets from the perspectives of interest rate transmission, money supply, and economic expectations [37], but there are fewer studies focusing on spillovers from international tail risks in the soybean markets. Some studies identify the asymmetric risk spillover between soybean markets and analyze the theoretical reasons [20,28], but there is a lack of empirical testing of the formation mechanism of this phenomenon.

Existing research focuses on mean and volatility spillover effects among grain markets, while tail risk spillovers are understudied. In addition, research on price transmission between grain markets focuses on finished grains and needs more focus on China’s soybean industry chain. Based on these research gaps, this paper contributes to the existing literature in three respects. First, we analyzed the extent of tail risk spillovers from the international soybean market to the Chinese soybean market through the Copula–CoVaR (conditional value at risk) method. Second, the degree of impact of the international soybean market on China’s soybean, soybean oil, soybean meal, feed, and edible soybean oil market was analyzed. Research from the perspective of the industry chain could reveal the links in China’s soybean industry chain that are most vulnerable to the impact of the international soybean market. Third, the mechanism of asymmetric risk spillovers is empirically tested in terms of inventory and consumer demand through quantile regressions.

3. Methodology and Data

We first analyzed the tail risk of soybean markets through the value at risk (VaR). Then, we studied the tail risk spillover from the international soybean market to markets in the Chinese soybean industry chain through the conditional value at risk (CoVaR). Finally, we investigated the risk spillover asymmetry mechanism through quantile regression with moderating effects.

3.1. The Copula–CoVaR Approach

Considering the advantages of the ARMA-GARCH model in dealing with autocorrelation and heteroskedasticity [38], this article employed the model to construct marginal distributions to capture the actual interdependencies among markets. Based on AIC, we selected the appropriate types and order of the ARMA-GARCH models. This model is not important, so this article will not introduce it in detail.

The VaR is one of the generally applied tail risk measures. It represents the maximum possible loss (or gain) an asset could suffer within a given time horizon and confidence level. The downside value-at-risk (VaR-D) and the upside value-at-risk (VaR-U) are defined, respectively, as:

$$\mathrm{Pr}(r_t\le Va{R}_{\alpha ,t})=\alpha$$

(1)

$$\mathrm{Pr}(r_t\ge Va{R}_{\alpha ,t})=\alpha$$

(2)

where Pr denotes the probability function and r_t represents the rate of return; α is 0.05 and 0.95, respectively. The extent of risk spillover from the international soybean market to each market in the Chinese soybean industry chain is quantified by CoVaR, which can reflect the extreme risk spillover situation well [39]. The CoVaR in this paper indicates the extreme price risk of China’s soybean industry chain under the extreme situation of the international soybean market. The downside and upside CoVaRs are defined, respectively, as:

$$\mathrm{Pr}(r_t^i\le CoVa{R}_{\beta \cdot t}|r_t^j\le Va{R}_{\alpha ,t})=\beta$$

(3)

$$\mathrm{Pr}(r_t^i\ge CoVa{R}_{\beta \cdot t}|r_t^j\ge Va{R}_{\alpha ,t})=\beta$$

(4)

where i denotes the international soybean market and j represents the markets in the Chinese soybean industry chain. In addition, α and β of the downside CoVaR are both 0.05; and α and β of the upside CoVaR are both 0.95. Based on the above formula, the calculation of CoVaR requires the joint distribution function of the two markets, and Sklar’s theorem provides an excellent solution to this requirement [40].

$$F(x_1,\dots x_n)=C(F(x_1),\dots ,F(x_n)),\forall (x_1,\dots x_n)\in R^n$$

(5)

$$f(x_1,\cdots ,x_n)={\displaystyle \frac{\partial C(F(x_1),\cdots ,F(x_n))}{\partial F(x_1)\cdots \partial F(x_n)}}f(x_1)\cdots f(x_n)=c(F(x_1),\cdots ,F(x_n))$$

(6)

If F is an n-dimensional joint distribution function with continuous marginal distribution functions F₁, F₂,..., F_n, then there is a unique n-dimensional Copula function C that connects the marginal distributions to their joint distributions, which can characterize the nonlinear relationship. The specific Copula model would be chosen based on the AIC values. Equation (6) is obtained by taking the derivative of Equation (5). With reference to existing studies [19], the specific steps to solve for CoVaR are as follows:

$$C(F_{r_t^i}(CoVa{R}_{\beta ,t}^i),F_{r_t^j}(Va{R}_{\alpha ,t}^j))=\alpha \beta$$

(7)

$$1-F_{r_t^i}(CoVa{R}_{\beta ,t}^i)-F_{r_t^j}(Va{R}_{\alpha ,t}^j)+C(F_{r_t^i}(CoVa{R}_{\beta ,t}^i),F_{r_t^j}(Va{R}_{\alpha ,t}^j))=\alpha \beta$$

(8)

$$CoVa{R}_{\beta ,t}^i=F_{r_t^i}^{-1}(F_{r_t^i}(CoVa{R}_{\beta ,t}^i))$$

(9)

To further identify the direction and magnitude of the risk spillover, this article calculates ∆CoVaR and %∆CoVaR. The ∆CoVaR is the difference between the VaR of the market returns in the Chinese soybean industry chain conditional on the extreme state of the international soybean market and the VaR of the market returns in the Chinese soybean industry chain conditional on the normal state of the international soybean market.

$$\Delta CoVaR=CoVa{R}_{\beta ,t}^i(Va{R}_{\alpha ,t}^j)-CoVa{R}_{\beta ,t}^i(Va{R}_{0.5,t}^j)$$

(10)

$$\%\Delta CoVaR={\displaystyle \frac{\Delta CoVaR}{CoVa{R}_{\beta ,t}^i(Va{R}_{0.5,t}^j)}}$$

(11)

3.2. Quantile Regression

Based on the above analysis, the international soybean market mainly affects China’s soybean industry chain through soybean spot goods rather than by-products. Additionally, inventory behavior and consumer demand are important factors regulating the degree of risk spillover from the international soybean market to the Chinese soybean market. From inventory changes and consumption demand, this study used quantile regression to analyze the asymmetry of risk spillovers from the international soybean market to the Chinese soybean market.

$$Q^{\tau }(Soy|Insoy)={\alpha }_0^{\tau }+{\alpha }_1^{\tau }Insoy+{\alpha }_2^{\tau }Insoy\times Soysur+{\alpha }_3^{\tau }Soysur+{\alpha }_4^{\tau }CL+{\delta }_i^{\tau }Ci$$

(12)

$$Q^{\tau }(Soy|Insoy)={\beta }_0^{\tau }+{\beta }_1^{\tau }Insoy+{\beta }_2^{\tau }Insoy\times CL+{\beta }_3^{\tau }Soysur+{\beta }_4^{\tau }CL+{\delta }_i^{\tau }Ci$$

(13)

We focused on the moderating effect at the extremes and selected five tertiles of 0.95, 0.9, 0.5, 0.1, and 0.05. The article also considers the impact of financial markets, other large grain markets, extreme events, and US–China trade frictions. The variable description is shown in

Table 1. Variable descriptions.

	Variable	Definition	Measurement	Unit
Dependent variable	Soy	China’s soybean return	Logarithmic return on monthly soybean prices in China	%
Independent variable	Insoy	International soybean return	Logarithmic return on the monthly price of international soybean	%
Mechanism variables	Soysur	China’s soybean stock-to-consumption ratio	Ratio of closing stocks to consumption	%
	CL	Consumer demand	Consumption expenditure per capita	RMB/person
Control Variables (Ci)	FM	Financial market	Logarithmic return on the monthly China Securities Index 300(CSI 300)	%
	Corn	Major agricultural commodities	Logarithmic return on the monthly price of Chinese corn	%
	EE	Extreme events	During the severe period of the epidemic in the United States and Brazil, the value is 1(2020.03–2021.08), while during other periods it was 0	-
	GC	Geopolitical conflicts	During the period when China imposed tariffs on U.S. soybeans during the U.S.-China trade friction, the value is 1(2018.04–2020.03), and the rest of the time it is 0	-

. The risk spillover from the international soybean market to the Chinese soybean market is again verified by the coefficient of the Insoy. If it is significantly greater than zero, this indicates that a rise (fall) in international soybean prices causes an increase (fall) in Chinese soybean prices. In addition, we analyze the moderating role of inventory movements and consumer demand in the risk spillover from the international soybean market to the Chinese soybean market through $Insoy\times Soysur$ and $Insoy\times CL$.

3.3. Data and Variables

In the risk premium measurement, this article selected weekly price data for the international soybean market (Chicago soybean spot), the Chinese soybean market and its by-products market (soybean, soybean oil, soybean meal spot), the Chinese feed market (compound feeds of swine, eggfowl, and meat bird), and Chinese edible soybean oil for the period from January 2014 to February 2023. The above data are derived from agdata (http://www.agdata.cn/, accessed on 23 December 2023). To exclude the effect of exchange rate fluctuations, we have standardized international soybean prices in RMB at the corresponding exchange rates. About 85 percent of soybean meal in China is used for poultry and swine. Thus, this essay uses swine, eggfowl, and meat bird compound feeds to represent the Chinese feed markets.

In analyzing risk spillover asymmetry, we used data from January 2013 to December 2022 on the international soybean prices, Chinese soybean prices, stock-to-consumption ratios, and per capita consumption expenditures. The international and Chinese soybean prices and the inventory consumption ratio come from agdata. Consumption expenditure per capita is derived from the China National Statistical Office. If the current stock-to-consumption ratio increases relative to the previous period, then the current stock has increased, and the stock-to-consumption ratio can be a good indicator of relative stock changes. Consumption expenditure per capita responds to macroeconomic growth and is a proxy variable for soybean consumption demand [41]. Considering the strong seasonality of soybeans, we removed the seasonal factor from soybean prices with the moving-average method.

4. Empirical Results

The risk spillover from the international soybean market to China’s soybean industry chain is especially realized through imported soybeans. China’s soybean meal and soybean oil are mainly crushed from imported soybeans, and direct imports of both are small. China’s soybean meal consumption grew from 49 million tonnes in 2013 to 74 million tonnes in 2022, an increase of 51.0%, with imports comprising less than 1% of its consumption. China’s total consumption of soybean oil grew from 13.09 million tonnes in 2013 to 17.7 million tonnes in 2022, an increase of 35.2 percent, with imports accounting for less than 8 percent of its total consumption.

In the long term, international soybean prices and China’s soybean and soybean meal prices have a consistent trend; in the past three years, the three prices were at historically high levels (Figure 1). International soybean prices have been lower and more competitive than Chinese soybean prices. Because China is a major importer of soybeans and the United States is a major exporter of soybeans, the price of Chinese soybeans would include transportation, storage, and other costs.

Table 2. Descriptive statistics of the weekly returns.

	Mean	Max	Min	S. D	Skewness	Kurtosis	J.B.	LB-Q	ARCH-LM	ADF
Insoy	0.054	25.911	−24.926	3.569	0.014	13.592	0.000	0.089	0.000	0.01
Soy	0.051	8.097	−6.557	1.251	0.087	9.145	0.000	0.000	0.000	0.01
Soym	0.023	12.438	−9.104	2.169	0.618	2.717	0.000	0.000	0.008	0.01
Soyo	0.065	8.582	−9.494	1.955	−0.447	3.355	0.000	0.000	0.000	0.01
Swinef	0.033	4.725	−2.797	0.685	1.694	13.685	0.000	0.007	0.127	0.01
Eggf	0.037	10.301	−10.595	1.068	−1.387	67.451	0.000	0.000	0.000	0.01
Meatbf	0.034	10.345	−10.345	0.724	−0.066	171.653	0.000	0.000	0.000	0.01
Esoyo	0.031	1.190	−0.579	0.205	0.979	4.108	0.000	0.000	0.000	0.01

Note: Soym and Soyo represent China’s soybean meal and soybean oil markets, respectively. Swinef, Eggf, and Meatbf represent Chinese swine, eggfowl, and meat bird compound feeds, respectively. Esoyo represents Chinese edible soybean oil. J.B. is the p-value of Jarque–Bera statistics; LB-Q is the p-value of Ljung–Box test with lagged order 12; ARCH-LM is the p-value of ARCH test with lagged order 12.

summarizes the statistical characteristics of weekly returns across markets. The maximum and minimum values of the weekly returns of international soybeans were 25.91 and −24.93, respectively, with a standard deviation of 3.57, demonstrating extremely high volatility and risk. Highly correlated with international soybean, China’s soybean meal and soybean oil are relatively highly volatile and risky. The volatility and risk of edible soybean oil is negligible. Additionally, the residuals of ARMA-GARCH had no autocorrelation or heteroscedasticity according to the results (

Table 3. The results of the ARMA-GARCH models.

	Insoy	Soy	Soym	Soyo	Swinef	Eggf	Meatbf	Esoyo
Mean model
ϕ0	0.075 (0.205)	0.017 (0.040)	−0.053 (0.148)	−0.029 (0.095)	0.000 (0.000)	−0.013 (0.030)	0.001 (0.026)	−0.032 (0.022)
ϕ1	0.320 ** (0.156)	0.181 ** (0.073)	0.458 *** (0.042)	0.308 *** (0.044)	0.054 *** (0.001)	0.776 *** (0.04)	0.810 *** (0.038)	0.662 *** (0.036)
ϕ2	-	-	-	-	-	-	-	0.340 *** (0.036)
φ1	−0.317 ** (0.155)	−0.234 *** (0.064)	-	-	−0.055 *** (0.001)	−0.462 *** (0.062)	−0.544 *** (0.054)	−0.475 *** (0.003)
φ2	-	−0.352 *** (0.003)	-	-	-	-	-	−0.367 *** (0.012)
φ3	-	-	-	-	-	-	-	−0.133 *** (0.034)
Variance mode
α0	3.494 *** (0.609)	0.530 *** (0.145)	0.567 ** (0.250)	0.019 (0.021)	0.000 (0.000)	0.048 *** (0.014)	0.031 *** (0.007)	0.001 (0.001)
α1	0.110 (0.080)	0.999 *** (0.243)	0.165 *** (0.058)	0.066 *** (0.021)	0.020 *** (0.002)	0.421 ** (0.194)	0.152 ** (0.070)	0.150 *** (0.049)
β1	−0.363 ** (0.162)	-	0.686 *** (0.097)	0.933 *** (0.025)	0.001 *** (0.000)	0.254 *** (0.096)	0.363 *** (0.107)	0.234 (0.220)
β2	-	-	-	-	-	-	-	0.616 *** (0.208)
leverage term
σ1	0.497 *** (0.147)	-	-	-	-	-	-	-
Shape(p)	0.000	0.000	0.001	0.000	0.000	0.000	0.014	0.001
LB-Q(p)	0.258	0.821	0.228	0.657	0.993	1.000	0.682	0.310
ARCH-LM(p)	0.845	0.960	0.763	0.916	0.569		1.000	0.180

Note: The Insoy was fitted with EGARCH model; and the other were fitted with standard GARCH model. Soym and Soyo represent China’s soybean meal and soybean oil markets, respectively. Swinef, Eggf, and Meatbf represent Chinese swine, eggfowl, and meat bird compound feeds, respectively. Esoyo represents Chinese edible soybean oil. *** and ** indicate significance at the 1%, 5%, and levels, respectively.

), and the Copula model can therefore be used for further research (

Table 4. The results of Copula.

	Type	AIC	par1	par2	Kendall
Insoy-Soy	Survival Clayton	−13.24	0.20 *** (0.08)	-	0.09
Insoy-Soym	Survival Clayton	−13.87	0.28 *** (0.08)	-	0.12
Insoy–Soyo	Student t Copula	−2.87	0.11 *** (0.05)	10.00 *** (3.92)	0.07
Insoy-Esoyo	Gumbel	2.00	1.00 *** (0.03)	-	0.00
Insoy-Swinef	Gumbel	2.08	1.00 *** (0.05)	-	0.00
Insoy-Eggf	Clayton	−3.90	0.16 *** (0.07)	-	0.07
Insoy-Meatbf	Gaussian	−13.99	0.19 *** (0.05)	-	0.12

Note: The t-value is in parentheses. ***, indicate significance at the 1%, levels.

).

4.1. Risk Spillover from International Soybean Market to China’s Soybean Industry Chain

Figure 2 displays the upside VaR at the 95% quantile and the downside VaR at the 5% quantile for each return series. The international soybean market has the largest upside VaR and downside VaR of 4.908 and −4.656, respectively, indicating high risk. Compared to the international soybean market, the risk of the Chinese soybean market is significantly smaller, with an upside VaR and downside VaR of 1.203 and −1.095, respectively. China’s soybean by-products market is relatively risky. The upside VaR and downside VaR for soybean meal are 3.233 and −2.793, respectively, while the upside VaR and downside VaR for soybean oil are 2.756 and −2.882, respectively. In the Chinese feed market, the VaR of swine compound feed is small, while the VaR of eggfowl and meat bird compound feed is relatively large. Generally, the price volatility and tail risk of Chinese soybean and their by-products market are significantly smaller than that in the international soybean market. The probable reason for this is that a large proportion of Chinese grain merchants are state-owned enterprises, which tend to smooth out soybean price volatility through storage behavior.

The international soybean market has an asymmetric tail risk spillover effect to China’s soybean industry chain. In terms of the degree of upside risk spillover, the international soybean market has a high upside risk spillover to China’s soybean and its by-products market, with 56.90%, 78.10%, and 28.58% for China’s soybeans, soybean meal, and soybean oil, respectively (

Table 5. Risk spillover from the international soybean market to the Chinese soybean chain.

	CoVaR-D	ΔCoVaR-D	%ΔCoVaR-D	CoVaR-U	ΔCoVaR-U	%ΔCoVaR-U
Insoy-Soy	−1.043	0.000	0.00	2.385	0.865	56.90
Insoy-Soym	−1.073	0.000	0.00	2.801	1.228	78.10
Insoy–Soyo	−3.712	0.000	0.00	4.186	0.930	28.58
Insoy-Swinef	0.080	0.000	0.00	0.082	0.000	0.00
Insoy-Eggf	0.073	−0.227	75.61	1.399	0.011	0.79
Insoy-Meatbf	0.240	0.000	0.00	1.381	0.096	7.47
Insoy-Esoyo	0.358	0.000	0.00	0.958	0.01	0.10

Note: Soym and Soyo represent China’s soybean meal and soybean oil markets, respectively. Swinef, Eggf, and Meatbf represent Chinese swine, eggfowl, and meat bird compound feeds, respectively. Esoyo represents Chinese edible soybean oil.

). The international soybean market has the highest upside risk spillover to China’s soybean meal market. There could be three main reasons for this. First, China’s soybean meal is highly dependent on imported soybeans and is greatly influenced by the international soybean market. Second, China has a huge demand for soybean meal. The improving standard of living of the Chinese people has boosted the demand for feed soybean meal in the livestock industry. Third, soybean meal is the most important protein feed ingredient in China, with weak substitutability, which exacerbates the rigidity of demand for fed soybean meal [42]. The degree of risk spillover from the international soybean market to the Chinese soybean oil market is also high. Soybean is the largest oilseed imported into China. China’s soybean oil also relies on imported soybeans. However, soybean oil has a substitution relationship with rapeseed oil and palm oil, which can alleviate the price impact of the international soybean market on the Chinese soybean oil market [43].

The degree of risk spillover from the international soybean market to the Chinese feed market is relatively small. The degrees of upside risk spillover from the international soybean market to the compound feed markets of swine, eggfowl, and meat bird are 0.00%, 0.79%, and 7.47%, respectively. The international soybean market significantly impacts the compound feed market of meat bird. Among the various types of compound feeds in China, corn generally accounts for the highest proportion, over 60 percent, while soybean meal accounts for the second highest proportion, but this is often less than 30 percent. Therefore, fluctuations in the price of soybean meal have a limited impact on feed prices. In addition, the upside risk spillover from the international soybean market to the Chinese edible soybean oil market is also small, at 0.00%.

Overall, the international soybean market has a significant risk spillover effect on the Chinese soybean industry chain, and the upside risk spillover is significant, while the downside risk spillover is not obvious. As the soybean chain deepens, the level of risk spillover decreases due to time lags. In China’s soybean industry chain, the soybean meal market is the most affected by the international soybean market.

4.2. Analysis of Asymmetric Risk Spillover

Based on the above analysis, the international soybean market affects China’s soybean industry chain through imported soybeans rather than its by-product market. Therefore, in the asymmetry risk spillover analysis, we choose the soybean stock and demand. Based on Equations (12) and (13), we empirically verified the mechanism of asymmetric risk spillover from the international soybean market to the Chinese soybean market from the perspectives of changes in China’s soybean inventories and demand. Most of the coefficients of the Insoy are significantly positive in the regression results (

Table 6. Moderating effect of soybean inventory.

Quartile	Insoy	$\mathit{Insoy}\times \mathit{S}\mathit{o}$ysur	Control Variables	Constant	R2
0.95	0.860 *** (0.14)	0.008 (0.007)	Yes	−0.506 (0.387)	0.820
0.90	0.854 *** (0.14)	0.008 (0.01)	Yes	−0.397 (0.39)	0.805
0.50	1.041 *** (0.11)	−0.004 (0.01)	Yes	0.064 (0.30)	0.818
0.10	1.079 *** (−0.22)	−0.007 (0.01)	Yes	−0.021 (0.64)	0.794
0.05	1.279 *** (0.13)	−0.024 *** (0.01)	Yes	0.193 (0.36)	0.823

Note: The t-value is in parentheses. *** indicate significance at the 1%, levels.

and

Table 7. Moderating effect of soybean demand.

Quartile	Insoy	$\mathit{Insoy}\times$CL	Control Variables	Constant	R2
0.95	0.497 (1.08)	0.051 (0.11)	Yes	−0.495 (0.36)	0.819
0.90	0.173 (1.25)	0.085 (0.13)	Yes	−0.285 (0.42)	0.803
0.50	1.731 ** (0.87)	−0.079 (0.09)	Yes	−0.008 (0.29)	0.817
0.10	3.375 * (1.96)	−0.253 * (0.18)	Yes	0.412 (0.65)	0.796
0.05	2.800 *** (0.98)	−0.194 ** (0.10)	Yes	0.299 (0.33)	0.820

Note: The t-value is in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

), suggesting that a rise in international soybean prices would cause a surge in China’s soybean prices, and a fall would cause a fall in China’s soybean prices. This further validates that there is a risk spillover from the international soybean market to the Chinese soybean market.

The increase in China’s soybean stocks and the growth of China’s soybean demand could effectively mitigate the extent of downside risk spillover from the international soybean market to the Chinese soybean market. The coefficient of $Insoy\times Soysur$ is significantly negative at the 0.05 and 0.1 quantile levels, indicating that the extent of downside risk spillovers from the international soybean market to the Chinese soybean market could be effectively reduced by increasing the soybean inventory in the event of a sharp fall in the international market price. When soybean prices fell, Chinese state-owned enterprises bought large quantities of soybeans, effectively increasing market demand and mitigating the decline in soybean prices. When market prices are low, if soybean prices continue to fall, companies expect consumer demand to rise and prices to increase in the future, thus purchasing stored soybeans at a low price, resulting in a reduction in the supply of soybeans on the market and preventing the price of soybeans from falling. If the price of soybeans rises, storage and other costs rise due to higher inventories. Enterprises would prioritize the sale of low-priced purchased stocks to ease financial pressure, increasing the supply of soybeans on the market and easing the sharp rise in soybean prices. When market prices are at a high level, if the price of soybeans falls, because the price level is still at a high level, enterprises selling their inventory of soybeans are still profitable and would continue to sell soybeans. Increased supply in the soybean market would exacerbate the decline in soybean prices. If the price of soybeans rises, companies would sell soybeans from their stockpiles for a larger profit. The increase in market supply would mitigate the price increase. In summary, when the price is high, the effect of inventory behavior in suppressing soybean price volatility in China is poor, which can alleviate the price rise, but cannot prevent the price from falling. While when the price level is low, the inventory behavior could effectively suppress the price volatility of domestic soybean.

The coefficient of $Insoy\times CL$ is significantly negative at the 0.05 and 0.1 quantile levels, demonstrating that growth in consumer demand could also effectively reduce the extent of downside risk spillovers from the international soybean market to the Chinese soybean market in the event of a sharp drop in international market price. During the study period, Chinese people’s living standards continued to improve and demand for meat, eggs, and milk increased. Increased demand for soybeans from China’s livestock sector could prevent declines in soybean prices. Should the price of soybeans decline, soybean processors would increase their soybean purchases in anticipation of increased future demand for soybeans in order to reduce the cost of raw materials, which could effectively increase the demand for soybeans in the market.

The coefficients of $Insoy\times Soysur$ and $Insoy\times CL$ are both greater than zero at the 0.95 and 0.9 quantile levels, indicating that the increase in stocks and consumer demand may exacerbate the degree of upside risk spillover from the international soybean market to the Chinese soybean market, which is consistent with theoretical analyses. However, the coefficients of both terms are not significant at the 0.95 and 0.9 quantile levels. The likely explanation is that China’s state-owned grain merchants are not just aiming to make a profit but also have the responsibility of calming food prices. China’s food policy aims to mitigate the extent of downside risk spillovers from the international soybean market to the Chinese soybean market to protect farmers’ incentives to grow soybeans and maintain domestic soybean production. There is a lack of management of the upside risk. In addition, extreme price increases tend to exceed the rate of improvement in living standards, and such price increases can inhibit consumption.

4.3. Robustness Analysis and Discussion

This article verified the robustness of the above results from three aspects. First, in measuring the degree of risk spillover, the suboptimal Copula function was selected for the fit, and the conclusions were generally consistent with those above (

Table 8. Fitting results of suboptimal Copula function.

	Type	AIC	par1	par2	Kendall
Insoy-Soy	Gaussian	−6.75	0.14 *** (0.05)	-	0.09
Insoy-Soym	Gaussian	−12.61	0.19 *** (0.05)	-	0.12
Insoy-Soyo	Gaussian	−3.69	0.11 ** (0.05)	-	0.07
Insoy-Esoyo	Student-t	12.56	−0.03 (0.05)	10.00 *** (2.67)	−0.02
Insoy-Swinef	Student-t	206.47	0.05 (0.06)	10.00 *** (0.60)	0.03
Insoy-Eggf	Gaussian	−2.48	0.11 ** (0.05)	-	0.07
Insoy-Meatbf	Gumbel	−11.89	1.14 *** (0.04)	-	0.12

Note: The t-value is in parentheses. ***, ** indicate significance at the 1%, 5% levels, respectively.

and

Table 9. Risk spillover based on suboptimal Copula function.

	CoVaR-D	ΔCoVaR-D	%ΔCoVaR-D	CoVaR-U	ΔCoVaR-U	%ΔCoVaR-U
Insoy-Soy	−1.291	0.000	0.00	1.693	0.231	15.80
Insoy-Soym	−3.483	0.000	0.00	4.181	0.493	13.37
Insoy-Soyo	−3.061	0.000	0.00	3.575	0.308	9.43
Insoy-Swinef	0.027	0.000	0.00	0.030	0.001	1.83
Insoy-Eggf	0.056	0.000	0.00	1.484	0.055	5.54
Insoy-Meatbf	0.303	0.000	0.00	1.616	0.324	25.06
Insoy-Esoyo	0.299	0.000	0.00	1.012	0.059	6.22

Note: Same as Table 5.

). Second, in the quantile regression analysis, the Engel–Granger two-step cointegration test was used to rule out fallacious regressions, and the test results proved that there were no pseudo-regressions (

Table 10. The stationarity test of residuals in quantile regression.

	0.95	0.90	0.5	0.10	0.05
$Insoy\times Soysur$	0.01	0.01	0.01	0.01	0.01
$Insoy\times CL$	0.01	0.01	0.01	0.01	0.01

). Third, significant difference tests were conducted between CoVaR and VaR and between CoVaR-D and CoVaR-U, and the results showed that there were significant differences (

Table 11. The test results for the significance and asymmetry of risk spillovers.

	H0: CoVaR-D = VaR-D	H0: CoVaR-U = VaR-U	H0: CoVaR-U = CoVaR-D
Insoy-Soy	0.000	0.000	0.000
Insoy-Soym	0.000	0.000	0.000
Insoy-Soyo	0.000	0.000	0.000
Insoy-Swinef	0.000	0.000	0.000
Insoy-Eggf	0.029	0.000	0.000
Insoy-Meatbf	0.000	0.000	0.000
Insoy-Esoyo	0.000	0.000	0.000

Note: Same as Table 5.

). This confirms the existence of asymmetric risk spillovers from the international soybean market to the Chinese soybean industry chain. All of the tests proved that the previous results are reliable.

China has a huge demand for soybeans and is the world’s main importer of soybeans. Changes in its domestic demand will affect the balance between supply and demand in the world soybean trade market. Therefore, changes in domestic soybean prices would also be transmitted to the international market. In addition, the international soybean market and the Chinese soybean market would be affected by some common factors. In order to better portray the linkage between the Chinese soybean market and the international soybean market, we measured the degree of tail risk spillover from the Chinese soybean market to the international soybean market (Figure 3). China’s soybean industry chain also has a significant upside risk spillover to the international soybean market, and China’s soybean meal market has the highest degree of risk spillover (24.02%). This corroborates the previous conclusion that China’s soybean imports are mainly used for crushing to produce soybean meal; therefore, the Chinese soybean meal market is most closely linked to the international soybean market.

5. Conclusions and Policy Implications

This article first theoretically analyzed the mechanism of risk spillover from the international soybean market to China’s soybean industry chain and the mechanism of asymmetric risk spillover. Then, based on the Copula–CoVaR model, it measured the VaR of each market and the degree of risk spillover from the international soybean market to China’s soybean industry chain. Finally, the mechanism of asymmetric risk spillover is verified by the quantile regression method. The main conclusions are as follows: Firstly, the price volatility and risk of China’s soybean and its by-products market is smaller than that of the international soybean market. The Chinese soybean meal market, which is highly dependent on imported soybeans, also has high volatility and risk. The Chinese feed and edible soybean oil markets are considerably less risky than other markets. Secondly, the international soybean market has apparent risk spillovers to China’s soybean industry chain, and the degree of upward risk spillover is greater than the degree of downward risk spillover. The rise in international soybean prices would significantly drive up the prices of Chinese soybeans and their by-products and would have the greatest impact on the Chinese soybean meal market. Finally, the increase in soybean stocks and consumer demand can effectively mitigate the downside risk spillover from the international soybean market to the Chinese soybean market. China’s huge consumer demand makes it easy for soybean prices to rise but makes it difficult for them to fall. In addition, China’s soybean imports are concentrated, and its international pricing power is weak, so it is often a passive recipient when prices rise.

Based on the above findings, the following policy implications are proposed. First, China needs to further improve its international soybean price-monitoring and risk-management system. China is the largest importer of soybeans and is highly dependent on the international market. It is therefore necessary to monitor the international soybean market and collect important information on extreme weather, geopolitical conflicts, and international shipping conditions in the main source countries for soybean imports, so as to provide industry-wide data for Chinese enterprises to avoid risks. Second, it is necessary to promote the reduction and replacement of soybean meal in China. China’s soybean meal is extremely dependent on imported soybeans. The international soybean market has the greatest upside risk spillover to China’s soybean meal market. On the premise of ensuring the security of staple grains, China should appropriately adjust its planting structure and increase the proportion of rapeseed and soybean planting. Furthermore, the Chinese government should increase investment in scientific research on meal-based feeds, support enterprises in exploring diversification of protein feed sources, and increase the applicability of rapeseed meal and other meal-based feeds. Third, Chinese state-owned enterprises could further play an important role in stabilizing domestic soybean prices. Chinese state-owned enterprises could increase their pricing power in international trade by investing in international grain enterprises and overseas farmland, and further increase their soybean imports from the countries along the Belt and Road to reduce the concentration of soybean imports. In a short period of time, the current situation that China’s soybeans are highly dependent on the international market cannot be changed. China should increase its soybean storage capacity appropriately and regulate domestic soybean prices through storage behavior. China’s state-owned grain enterprises should buy soybeans when prices are low to increase stocks, while setting minimum purchase prices to protect the incentive to grow soybeans domestically. When the price of soybeans is high, state-owned enterprises should sell their stocks of soybeans to avoid a rapid rise in the cost of soybean processing enterprises, which would be fatal to the soybean industry chain. In addition, China’s soybean processing enterprises attach importance to the extreme rise in the price of imported soybeans and pay timely attention to the soybean production situation in the main producing countries. Enterprises can use futures hedging instruments to avoid the risk of price rises.

This article still has some limitations. The mechanism test of asymmetric risk spillover from the international large market to the Chinese soybean market is not thorough. The influence of extreme price risk on soybean growers needs further exploration. The study period was a decade of growth from low to high levels of income in China, and the demand for soybeans increased every year. However, soybean demand has now peaked and will not grow in the future. Therefore, the risk spillover characteristics of international soybeans to Chinese soybeans may change. In addition, other mechanisms of asymmetric risk spillover can be explored. These shortcomings can be further analyzed in future research.