The Fluctuation Characteristics and Periodic Patterns of Potato Prices in China

Abstract

The aim of this paper was to provide a more scientific and effective analysis of the fluctuation pattern of the Chinese potato market by extracting the characteristics of the price fluctuation cycle to effectively grasp the characteristics of price changes in the potato market, thus promoting the stable and healthy development of the Chinese potato industry, and to expand the application scenarios of the EEMD model to provide a reference for the study of price fluctuation patterns in other agricultural markets. This study used an ensemble empirical modal decomposition (EEMD) model to examine time-series data on Chinese wholesale potato market prices from January 2005 to December 2021. The results showed that (1) Chinese wholesale potato market prices are characterized by some rigidity, with sharp changes in growth rates; (2) Chinese wholesale potato market prices are dominated by short- and medium-term fluctuations, and the decomposed components can better reflect the characteristics of the original series fluctuations; (3) Chinese wholesale potato market monthly prices have long- and short-term fluctuations with a 6- and 19-month cycle, and are dominated by short-term high-frequency fluctuations; (4) monthly price fluctuations in the Chinese wholesale potato market are more intense in high-frequency than low-frequency fluctuations, and there is a strong correlation between high- and low-frequency fluctuations in precipitation, temperature and potato prices. Finally, suggestions were made for creating and improving a national potato price information platform and strengthening the information early warning mechanism; improving the potato production interest linkage mechanism and enhancing potato farmers’ ability to cope with market and natural risks; and improving the potato reserve system and potato storage facilities.

1. Introduction

High yields, nutritional value and adaptability to the environment [1] have made the potato an important global food source [2]. The potato industry plays an important role in increasing incomes and creating jobs for people in developing countries [3], and in 2022 global potato production totaled 350 million tons, of which China produced 18.516 million tons, accounting for approximately 5.3 per cent of the world total. Historically, potatoes have made a significant contribution to China’s population growth and social development as an important supplement to staple foods. In China, the potato industry now has a new mission to consolidate the gains of poverty eradication and promote rural revitalization [4]. Price is a market signal for agricultural products and plays an important role in regulating production and consumption [5], while changes in agricultural prices have been an important concern for the Chinese government as they affect farmers’ incomes and the stable supply of products [6,7,8]. Ensuring food security is an important concern for countries around the world, and large fluctuations in agricultural prices not only harm farmers’ immediate interests, but also threaten national food security. Evidence suggests that changes in food prices can lead to instability in farmers’ incomes and jeopardize food security in developing countries [9,10], which, in turn, can lead to a reduction in household investment in human capital and agricultural production, reduce individual productive capacity [11] and discourage the transfer of agricultural labor to manufacturing. This affects the stable development of developing countries’ economies [12,13]. Moreover, price shocks can lead to lower real incomes and higher costs in all sectors, and higher social unemployment increases political instability [14,15].

Agricultural products are the basis of human survival and development [16], and their prices have always been a focus of national and social attention. The earliest empirical studies of agricultural price volatility emerged in the early 20th century, when Henry L. Moore pioneered the study of “business cycles” and “cotton price volatility”, which directly contributed to the development of academic research on agricultural products. Studies on the cycles of international and domestic agricultural price volatility have gradually been conducted and have found that food price volatility is caused by imbalances in food supply and demand, and that government policies to regulate prices can mitigate large fluctuations in food prices [17,18]. However, international food prices are able to influence domestic food prices through the transmission mechanism of prices [19], and the pattern of agricultural price fluctuations should be grasped from both domestic and international markets. In addition, natural factors can also be influential in the price volatility of agricultural products. Climate change may lead to higher prices of maize, wheat and rice by affecting their production, but the impact of meteorological factors on the price volatility of agricultural products is more complex and cannot be predicted using linear models. The study of price volatility in agricultural products has also been the focus of research by Chinese scholars. From the perspective of research objects, the research results have mainly focused on the investigation of price volatility and patterns of bulk agricultural products such as aquatic products, livestock products, grain crops and vegetables [20,21,22]. From the perspective of research methods, most scholars have used different methods, such as statistical models and econometric models, to investigate price volatility patterns based on the market characteristics of bulk agricultural products. For example, the GARCH family model [23] and ARCH class [24] models have been used to investigate the clustering and asymmetry of agricultural prices, and the Census X12 seasonal adjustment method [25] and H–P filter method [26] have been used to investigate the seasonal changes and cyclical patterns of agricultural price fluctuations.

However, the stability of potato prices is a global issue [27]. China’s Central Rural Work Conference in 2022 stressed the need to make increasing farmers’ incomes the central task of the “Agriculture, Rural areas and Farmers”, and to do everything possible to broaden the channels for farmers to increase their incomes and become rich. China’s wholesale potato market prices have fluctuated dramatically since the turn of the century, with monthly price differences reaching around 1 yuan/kg in individual years, and large price fluctuations in the other years. The alternation of “low potato prices hurting farmers” and “high potato prices hurting people” has severely damaged the interests of Chinese potato farmers and consumers, limiting the industry’s ability to enrich people’s incomes and develop smoothly and healthily [28]. At present, research results around the pattern of potato price volatility in China have focused on the time-series fluctuations of potato prices, annual and quarterly fluctuations of field prices and the transmission relationship between production and marketing prices. In terms of research methods, most scholars have mainly used X-12 models, H–P filtering methods and VAR models to study the pattern of potato price volatility [29]. Some scholars have used spatial measures to explore the spatial patterns of price changes [30], while with the rapid development of artificial intelligence, methods such as BP neural network machine learning [31] have also been introduced into price forecasting. The above literature on potato and other agricultural commodities prices provides a reference for the study of agricultural market patterns, but most of the methods currently employed for the study of agricultural price fluctuation patterns still have some limitations.

It is well known that price volatility is a complex, multi-layered phenomenon caused by various elements in the agricultural supply chain and a number of external factors. According to Hayek’s theory, price volatility is the result of the compounding influence of multiple factors, often showing complex characteristics, such as non-linearity and non-stationarity, and the time-series changes in prices often contain numerous periodic patterns at different scales [32], while the decomposition of price volatility of potatoes and other agricultural products using the above methods often fails to fully extract the original series of multi-scale volatility patterns of the original series. The proposed ensemble empirical mode decomposition (EEMD) method provides an effective solution to this problem. It starts from the characteristics of the data and reveals the intrinsic fluctuation characteristics of the data by decomposing the fluctuation information of the original signal at different scales, which can be processed effectively for non-linear and non-stationary data. As an effective tool for extracting new fluctuation patterns of time-frequency signals, it has received wide attention and recognition in academic circles since its introduction, and has been introduced by foreign scholars into the study of fluctuation patterns in social sciences [33,34], while only some domestic scholars have adopted this method to study economic fluctuations, stock indices, option prices, etc. In contrast, only some scholars in China have used this method to study the volatility of economic fluctuations, stock indices, option prices and other more macroscopic issues [35,36], and few scholars in China have introduced it into the study of the volatility of a certain agricultural commodity price.

In this context, this paper aims to expand the application of the EEMD model to other agricultural commodity markets and to provide a reference for the study of price volatility, while at the same time providing a more scientific and effective analysis of the volatility of the Chinese potato market by extracting the cyclical characteristics of price fluctuations so as to effectively grasp the characteristics of price changes in the Chinese potato market and, thus, promote the stable and healthy development of the potato industry. To this end, an ensemble empirical modal decomposition model (EEMD) was used to study the volatility characteristics and cyclical patterns of the Chinese potato wholesale market’s monthly price series from January 2005 to December 2021 using Matlab software tools (2022b). The results of the H–P filter decomposition of potato prices were used to verify the scientific validity of the EEMD method results, combined with correlation coefficients to test the serial correlation.

2. Theoretical Framework and Methods

2.1. Theoretical Framework

Spider web theory was first proposed by Schultz, Ricci and Tinbergen in 1930 and later named by Kaldor. According to classical economics, when the market equilibrium of a commodity is broken, the market will automatically restore equilibrium through market adjustment, but the spider web theory found that in some cases for some commodities with long production cycles, equilibrium is not necessarily restored automatically when it is broken. The limited rationality of farmers leads to market cognitive biases and herd behavior, and the low level of information access and organization leads to price changes that affect farmers’ supply more than demand, i.e., a dispersive spider web (Figure 1) in which the price elasticity of supply is greater than the price elasticity of demand for agricultural products. In a constant market environment, farmers make production decisions based on the last crop’s price and combine the expected price formed using experience and the expected profit of different crops to make production investments [37].

As a cash crop that is strongly influenced by natural weather conditions and market prices, potato production also conforms to the principles of the spider web model and is subject to constant changes in supply as a result of both price and unusual weather changes, leading to the formation of a new equilibrium price. Farmers’ production decisions are influenced by the current market price of potatoes, which is determined by the previous period’s production, and this inability to adjust production in time to align with price changes leads to problems of large fluctuations in potato prices. Figure 2 shows the transmission process that eventually leads to changing the potato price equilibrium in the short and long term due to large fluctuations in potato market prices and unusual changes in weather, which affect both the supply and demand, and expectations and behavior. When potato prices fluctuate significantly, in the short term this leads to consumers adjusting their consumption of potatoes and finding substitutes in response to price changes, and in the long term this affects consumer preferences, resulting in changes in total potato demand, with producers adjusting the scale of production of potatoes and growing substitute crops in the short term; in the long term this changes the income expectations of potato farmers and leads to long-term changes in production behavior based on the size of their own risk appetite, and ultimately affects the total potato supply. Furthermore, the two-way feedback mechanism between price fluctuations and expectations and behavior reinforces the path dependence of potato price fluctuations, making potato prices intensify on a frequently fluctuating path. The frequent occurrence of meteorological anomalies, such as droughts and floods, can affect supply by directly affecting potato production, and meteorological changes can increase expectations of potato production uncertainty, which in turn affects potato production and consumer behavior. An analysis of the mechanisms of price and weather variability in the potato market shows that large fluctuations in potato prices and unusual weather events can affect the welfare of producers and demanders in the short term and lead to changes in supply and demand, as well as changes in potato consumption preferences and producer market expectations, which will ultimately affect the level of potato market development and change the distribution of benefits among the various actors in the potato industry chain.

2.2. Research Methodology

EEMD is an innovation and development of the empirical mode decomposition (EMD) method, which effectively solves the problem of modal confounding in EMD decomposition. The EMD decomposition method was proposed by Chinese American scientist Huang [38] and it can decompose the fluctuation cycle patterns and long-term trends embedded in a signal and obtain the IMF components at different time scales. It does not need preset a priori basis functions like Fourier transform and wavelet decomposition methods, but performs signal decomposition based on the original change characteristics of the data, which has the outstanding advantages of being more complete, direct and adaptive. The EMD method is essentially a dynamic filter cluster decomposed according to the adaptive “basis” of the original signal, which can be used to extract the oscillation characteristics of the signal from the inherent potato price time-scale data to obtain the full IMF score of any potato price. The full IMF component of the composite original signal is then obtained.

The EMD decomposition of each component must satisfy two conditions at the same time: (i) the number of extreme value points and the number of zero points are equal or differ by at most one; and (ii) the upper and lower envelopes of the extreme minimum values are zero at any given time. The specific steps of the EMD decomposition were as follows.

(1)

The upper and lower envelopes X_up (t) and X_low (t) of the original series of potato prices were fitted using the cubic spline function.

(2)

The mean value of the upper and lower envelopes m(t) was calculated:

m(t) = [X_up (t) + X_low (t)]/2

(3)

The new series h₁ (t) was derived.

h₁ (t) = X(t) − m(t)

If h₁(t) satisfied the previous two conditions, h₁ (t) was recorded as imf₁ (t); if h₁ (t) was not satisfied, then h₁ (t) was used instead of X(t) and m₁ (t) was used instead of m(t), with the mean of h₁ (t) and the previous process carried out again until the imf₁ (t) met the previous conditions and the first potato price component was obtained.

(4)

The following calculation was performed:

r(t) = x(t) − imf₁ (t)

We wrote the difference sequence as r₁ (t) and repeated the previous steps to obtain imf₂ (t), repeating n times to obtain

r₁ (t) − imf₂ (t) = r₂ (t)

r_n−1 (t) − imf_n (t) = r_n (t)

(5)

When r_n (t) was very small or became a monotonic function, it could no longer filter out the pattern function to stop the above decomposition process.

Eventually, the original potato price signal x(t) was decomposed into n IMF components and a residual term r_n (t), i.e.,

$$\mathrm{X}\left(\mathrm{t}\right)=\frac{1}{N}{\displaystyle \sum }_{n=0}^{\infty }im{f}_i\left(t\right)+r_n\left(t\right)$$

EMD assumes that noise cannot affect the original series being decomposed, and that the eigenmodal functions can effectively reflect the changes in the original data. However, the reality is that, as Wu and Huang suggested, the original data are inevitably affected by noise, and the decomposed components cannot be completely extracted from the original data variation, which is the modal confusion problem [39]. To overcome this problem they proposed an EEMD decomposition model introducing white noise sequences, which was repeated N times after adding white noise to potato price sequences, so that the added white noise was eventually eliminated, effectively overcoming the problem of modal confusion and decomposing various types of signals effectively.

The decomposition steps are as follows.

(1)

The white noise sequence ω_i (t) with mean value 0 and SD c is added to the potato price sequence x(t), i.e.,

X_i (t) = x(t) + ω_i (t)

(2)

The newly composed potato price time series Xi (t) is decomposed using EMD to obtain the individual IMFs, and the jth IMF obtained after the ith addition of white noise is denoted as C_ij (t), with the residual equation noted as r_i (t).

(3)

The potato price IMF obtained from the above equation is averaged overall to obtain the final IMF after integrating the empirical modal decomposition, i.e.,

$$C_j\left(\mathrm{t}\right)=\frac{1}{N}{\displaystyle \sum }_{n=1}^N{C}_{ij}\left(t\right)$$

In the above equation, N is the number of times white noise is added and C_j (t) is the jth potato price eigenmodal function (IMF) obtained using the ensemble empirical modal decomposition.

2.3. IMF Component Eigenmodes Reconstruction Method

The eigenmode reorganization method is used to reorganize different components into high- and low-frequency signals [40] in order to study the high-frequency short-term fluctuations and long-term low-frequency fluctuation patterns of the original series, respectively. Therefore, this paper used the fine-to-coarse high- and low-frequency reconstruction method to perform high- and low-frequency reconstruction on the IMF sub-series decomposed from potato prices. The basic idea was that short-term market volatility should oscillate up and down around the price mean, while major events could have some degree of positive or negative economic impact on prices. The mean of the reconstructed high frequency term should then deviate significantly from 0. The steps of component reconstruction are as follows:

(1)

Calculate the imf₁ (t) − imf_n (t) mean;

(2)

Determine the IMF series that significantly deviates from zero using a t-test at significance level a;

(3)

If imf_n(t) deviates significantly from the zero mean, the first (n − 1) IMFs are summed to obtain the high-frequency series, the nth and subsequent IMF components are summed to the low-frequency series and the remaining term is the trend component.

2.4. Sequence Correlation Test, Period Measurement

This study decomposed potato price time-series data into several IMF components of different time scales through the EEMD model, on the basis of which the numerical characteristics of the different components were evaluated through Pearson coefficients, analysis of variance and mean periods.

The Pearson coefficient is used to test the linear correlation of two series and is calculated as follows:

$${\rho }_{x,y}=\frac{cov\left(X,Y\right)}{{\sigma }_x{\sigma }_y}=\frac{E\left(XY\right)-E\left(X\right)E\left(Y\right)}{\sqrt{E\left(X^2\right)-E^2\left(X\right)}\sqrt{E\left(Y^2\right)-E^2\left(Y\right)}}$$

Referring to the method of Wang [4], the average period was measured for different time scales of the IMF component:

$$average cycle=\frac{2\times length of the time series}{number of extrema in the time series}$$

3. Data Sources

In order to maximize the study of the longer-term fluctuation pattern characteristics of potato prices, according to the availability and consistency of data, the study used a total of 204 monthly prices from January 2005 to December 2021 of the potato wholesale market in China. The data were calculated based on the daily prices of more than 100 wholesale markets designated by the Ministry of Agriculture and Rural Affairs, collected by the Ministry of Commerce’s National Agricultural Products Business Information Public Service Platform. In order to exclude the effect of inflation on potato price changes, the monthly prices of wholesale potato markets were deflated using the consumer price index with 2005 used as the base period. The rainfall and temperature data used for the study were obtained from the data element set of ground-based meteorological monitoring stations in various regions of China from the National Meteorological Information Centre. The currency of the prices used in the study is RMB, and at the exchange rate of 4 May 2023 1 RMB ≈ 0.1310 EUR.

4. Results and Analysis

4.1. Potato Price Time Series Change Analysis in China

As can be seen from Figure 3 and Figure 4, from January 2005 to December 2021 there were 204 wholesale potato market monthly prices, which overall seemed to fluctuate frequently; before 2010, 1.35 yuan/kg was the approximate center of fluctuations, while after 2010, prices changed overall to a new level, with 1.8 yuan/kg at the center of fluctuations. The growth rate change basically showed a one month increase, one month decrease change trend. Comparing the monthly price sequence changes and price growth rate changes, it was found that the monthly price growth rate fluctuations in the wholesale potato market were more intense than the price sequence fluctuations, which had a certain degree of inertia, usually with 6–10 months of continuous increase after about half a year of continuous decline. The wholesale potato market monthly price’s rigidity characteristics were more obvious, and the speed of the price’s change was not smooth.

For the monthly wholesale potato market monthly price, chronological changes could be seen, with wholesale potato market monthly price fluctuations in China the largest in 2005, 2010, 2011, 2016 and 2020. In 2010, the highest monthly price wholesale potato market in China was 2.67 yuan/kg and the lowest price was 1.88 yuan/kg, with a price difference of 0.79 yuan/kg, and the largest monthly price drop occurred in June 2010, reaching 17%. In 2011, the highest monthly price of potatoes in the wholesale market was 2.35 yuan/kg, the lowest price was 1.41 yuan/kg and the price difference of 0.94 yuan/kg showed that the overall trend in the monthly price of potatoes in 2011 continued to fall (from January to December it fell by 38.9%). In 2016 and 2020, monthly prices in the wholesale market also fluctuated more sharply, with the maximum spread reaching 1.15 yuan/kg and 0.84 yuan/kg, respectively. In addition to the above years, other years’ monthly prices basically remained in the range of 1.0–2.5 yuan/kg, the monthly price difference was basically within 0.4 yuan/kg and there were no more significant fluctuations. The monthly price changes in the wholesale potato market in China were more dramatic and had a certain volatility pattern, as shown by the monthly price changes in the time series and growth rate of the wholesale potato market. However, it was not possible to derive more detailed and objective intrinsic volatility patterns based on the raw series and growth rate analysis alone, so it was necessary to further analyze the monthly price changes in the wholesale potato market in China with the help of an ensemble empirical modal decomposition model.

4.2. Price Volatility Decomposition

According to the principle of the EEMD model, it was necessary to set the standard deviation of adding white noise and integration number before EEMD decomposition. The signal decomposition effect is better when the integration number N is set to 100 or 200 and the standard deviation of white noise ε is set to 0.1 or 0.2 [39]. Therefore, in this paper, we set the ensemble number N to 100, added the white noise series with a standard deviation ε = 0.1 and used Matlab software to program the EEMD of the monthly time-frequency price signals of the wholesale potato market in China from January 2005 to December 2021, obtaining a total of six IMF components with different fluctuation periods and frequencies, as shown in Figure 5.

As can be seen from

Table 1. Cyclical characteristics of each component of the monthly price decomposition of the Chinese wholesale potato market from 2005 to 2021 and their correlation coefficients with the original series.

Component	Average Period (Number of Months)	Variance	Variance Ratio/%	Variance Contribution Rate/%	Pearson Correlation Coefficient
IMF1	1.94	0.004	4.03	5.02	0.27 **
IMF2	5.51	0.019	17.33	21.61	0.47 **
IMF3	12.75	0.018	17.02	21.23	0.55 **
IMF4	18.55	0.007	6.86	8.56	0.26 **
IMF5	51.00	0.013	11.81	14.73	0.52 **
IMF6	68.00	0.001	0.51	0.63	0.26 **
R		0.025	22.63	28.22	0.55 **

Note: “**” indicates significant correlation at 0.01 level.

, the average period of IMF1–IMF6 increased sequentially, reflecting the volatility characteristics of different time-scale components of the monthly price of the wholesale potato market in China. The Pearson correlation coefficient was used to obtain the consistency of the IMF components and the original series changes. It can be seen that the components IMF1–IMF6 and the original series were significantly correlated at the level of 99%, indicating that the components obtained after the EEMD had a more obvious correlation with the original series of monthly prices of the wholesale potato market. The fluctuation characteristics of the original series were more realistically represented by each component. The Pearson correlation coefficient of IMF3 was the largest among the components, followed by IMF5 and IMF2, indicating that the short- and medium-term high-frequency volatility components better reflected the characteristics of monthly price fluctuations. The variance contribution rate reflected the respective contribution shares of different components in the original series fluctuations, which was more consistent with the size of the Pearson correlation coefficient of each component. The larger variance contribution of IMF2 and IMF3 for short- and medium-term fluctuations reflected the larger influence of short- and medium-term factors affecting monthly prices in the wholesale potato market on potato price fluctuations.

4.3. Comparison of the Reconstruction of High and Low Frequency Components of Wave Motion and H–P Filtering Results

The EEMD was used to filter out the intrinsic fluctuation characteristics of the original signal from high frequencies to low frequencies sequentially, with high-frequency signals being more random and low-frequency signals having obvious periodic characteristics. In order to integrate the different components of the signal into a high- and low-frequency component that better reflects the economic meaning, the mean value of the IMF components was tested for significant differences from zero using a t-test [40], and the test results are shown in Figure 6.

It can be seen that there was a significant jump in the T-value at the 4th component, so the IMF1–IMF3 components could be grouped and determined as high-frequency components, while IMF4–IMF6 could be considered low-frequency components to obtain the time-series data of high- and low-frequency fluctuations of monthly price components in the potato wholesale market, as shown in Figure 7. At the same time, the H–P filtering method was used to decompose the monthly price data of the wholesale potato market and extract the cyclic series of price fluctuations (Figure 7), and it was found that the fluctuation components obtained using the H–P filtering method and the high-frequency series obtained using EEMD had a strong consistency, which further confirmed the effectiveness of the EEMD model. While the H–P filter method could only obtain a cyclic fluctuation series, it could not be derived from the low-frequency periodic pattern of price fluctuations, so the EEMD model could effectively complement the H–P filter method’s research deficiencies, improving the study of potato price fluctuations.

4.4. High- and Low-Frequency Component Fluctuation Characteristics Analysis

As can be seen from

Table 2. Characteristics of monthly price fluctuations of high- and low-frequency components of the Chinese wholesale potato market from 2005 to 2021 and their correlation coefficients with the original series.

Component	Average Period (Number of Months)	Variance	Variance Ratio/%	Variance Contribution Rate/%	Raw Series Pearson Correlation Coefficient	Precipitation Pearson Correlation Coefficient	Temperature Pearson Correlation Coefficient
High Frequency	5.83	0.05	44.66	52.02	0.71 **	−0.17	0.19
Low Frequency	18.55	0.02	18.57	21.63	0.62 **	0.31	0.06
R		0.02	22.63	26.36	0.55 **	0.02	0.02

Note: “**” indicates significant correlation at 0.01 level.

, the monthly price fluctuations of the wholesale potato market exhibited a variance contribution and Pearson correlation coefficient of the high-frequency components reflecting the short period of 52.02% and 0.71, respectively, while the low frequency components were 21.63% and 0.62, indicating that the long and short cycles of monthly price fluctuations of the wholesale potato market in China lasted 19 months and 6 months, respectively. The high-frequency component was more correlated with the original series’ fluctuations and had a higher contribution to the variance of the original series, indicating that the monthly price of the wholesale potato market in China was dominated by short-term high-frequency fluctuations with a 6-month cycle. The low-frequency long-period fluctuations affected monthly prices in the wholesale potato market to a lesser extent, but their variance contribution still reached 22% and they were also significantly correlated with the original series. The impact of low-frequency fluctuations on potato price volatility, with a 19-month cycle, also cannot be ignored.

According to the coefficient of variation, we could obtain the year-to-year fluctuation size of the high- and low-frequency components. From Figure 8, it can be seen that the coefficient of variation of the high-frequency component was larger than that of the low-frequency component in the study period, and the monthly price fluctuation of the high-frequency component changed more obviously than that of the low-frequency component in each year, while the intensity of the fluctuation of the high-frequency component showed a regular change in the cycle of approximately 3 years in general. This indicated that wholesale potato market monthly price fluctuations of the high-frequency short-term fluctuations in China were large, and there were cyclical changes in the intensity of fluctuations, while low-frequency long-term fluctuations were generally more stable and less volatile. The potato price high-frequency component fluctuated most strongly in 2011 and 2016, with coefficients of variation of 33% and 35%, respectively, while the next largest fluctuation size of 29.1% was in 2020, and in 2018 the fluctuation intensity was lowest and the price fluctuation was smoothest. The cyclical variation of the fluctuation intensity of the low-frequency component was about 4 years, and the coefficient of variation of the low-frequency component was largest in 2009 with 8.1%, and the second largest in 2005, showing that the wholesale market price of potatoes under the long-period scale had the most dramatic changes in these two years, while the coefficient of variation of the low-frequency component was smallest in 2006. Overall, the high-frequency short-cycle fluctuations in the monthly price of potatoes in the wholesale market showed more dramatic changes in intensity than the low-frequency long-cycle fluctuations in China.

As a crop that is susceptible to changes in temperature and precipitation, potato yields are often affected when weather conditions change abnormally, and this ultimately leads to abnormal fluctuations in potato prices. Figure 8 shows that the high- and low-frequency components of potato prices and the coefficients of variation of precipitation and temperature from 2005 to 2021 were correlated, with peaks and troughs overlapping in individual years. The correlation coefficients between precipitation, temperature and price fluctuations in Table 2 show that there was a strong correlation between high- and low-frequency fluctuations in precipitation, temperature and the potato prices. The correlation between precipitation, temperature change and high-frequency fluctuations in the potato price was similar, but one correlation was negative and one was positive. Figure 8 shows that high-frequency fluctuations in potato prices and temperature change were positively correlated, with peaks and troughs more consistent, while high-frequency fluctuations in price and precipitation change were negatively correlated, indicating that there was some synergy between temperature change and price. There was a lag between low-frequency fluctuations and precipitation. The correlation between low-frequency fluctuations and precipitation was greater than for temperature, suggesting that changes in precipitation in recent years, especially relating to flooding, have had a greater impact on potato production and ultimately on price fluctuations. It was clear from the research that abnormal changes in temperature and precipitation could have an impact on potato production and ultimately on potato price fluctuations, and that the impact of meteorological conditions on potato production cannot be ignored.

5. Conclusions and Policy Recommendations

5.1. Conclusions

This study used an ensemble empirical modal decomposition model (EEMD) to examine the characteristics and cyclical patterns of price volatility in the Chinese wholesale potato market from January 2005 to December 2021, drawing the following key conclusions.

The price rigidity of the wholesale potato market was characterized by more pronounced and more dramatic changes in growth rates in China. With 2010 as the cut-off point, the monthly price changes in the wholesale potato market broadly showed different fluctuations, with prices generally moving to a new level after 2010. Monthly price growth rate fluctuations in the wholesale potato market were more dramatic than price time series fluctuations, with the rigidity of monthly prices being more pronounced, while the rate of price change was not smooth. Monthly price fluctuations in the wholesale potato market in China were greatest in 2010, 2011, 2016 and 2020.

Potato wholesale market prices were dominated by short- and medium-term fluctuations, and the decomposition of each component could better reflect the volatility characteristics of the original series in China. The components obtained after EEMD all had a more obvious correlation with the original monthly price series of the wholesale potato market, and the short- and medium-term high-frequency fluctuation components better reflected the characteristics of monthly price fluctuations, while the short- and medium-term factors affecting the monthly price fluctuations of the wholesale potato market had a greater impact on potato price fluctuations.

There were short- and long-term fluctuations in monthly potato wholesale market prices of 19 months and 6 months, with short-term high-frequency fluctuations dominating in China. Short-term high-frequency fluctuations with a 6-month cycle exist in the monthly wholesale potato market in China, and contributed more to the overall price change. The contribution of variance and correlation coefficients indicated that long-period fluctuations also had an impact on monthly price fluctuations in the wholesale potato market.

Monthly price fluctuations in China’s wholesale potato market were more volatile at high frequencies than at low frequencies, and there was a strong correlation between high- and low-frequency fluctuations in precipitation, temperature and potato prices. High-frequency short-term fluctuations in the monthly wholesale potato market in China were larger in magnitude and had a cyclical intensity of around three years, while low-frequency long-term fluctuations were generally smoother and less volatile. Abnormal changes in temperature and precipitation can have an impact on potato production, and this is ultimately reflected in potato price fluctuations.

5.2. Policy Recommendations

It is recommended to create and improve a national potato price information platform and strengthen the information early warning mechanism. Potato wholesale market prices are influenced by the production side, sales side and consumption side of the multi-link integrated impact, which should be established to cover potato production, wholesale, consumption and other aspects of the potato industry chain to provide a comprehensive information data platform, achieve industry dynamics, market conditions, natural disaster forecasts and other related comprehensive information. In particular, stronger countermeasures should be taken against potato wholesale prices, which fluctuate in 6 and 19-month cycles, and early warnings on potato prices should be made. Full use of new media tools should be made to release timely information on supply and demand, market conditions, epidemic dynamics and other information to scientifically guide potato farmers to enter or exit the market rationally. A sound basic database on potato prices should be established, with timely data collected regarding production and field and market wholesale prices, and improvements to emergency warning mechanisms should be made for random emergencies, such as natural disasters and epidemics, to minimize the adverse impact of price fluctuations on the development of the potato industry.

The potato production interest linkage mechanism should be improved to enhance the potato farmers’ ability to cope with market and natural risks. An interest linkage mechanism with positive interactions between potato production, processing and marketing should be built to improve the overall level of the potato industry chain and form a synergy for industrial development. Moderate-scale operations of potato production should be encouraged to improve potato farmers’ ability to cope with market risks, modernize potato production and promote the organic connection between small potato farmers and the big market to the greatest extent possible. By fostering potato cooperatives, family farms and specialized companies, the level of specialization in potato production could be further enhanced and the level of organization and management at the potato production end could be improved to effectively enhance the ability of potato producers to cope with market risks and weather disasters, stabilize producer confidence and promote the sustainable development of potato production.

The potato reserve system and potato storage facilities should be improved. An effective potato storage mechanism can play a role in staggering the market, entering the cellar at low prices and leaving the cellar at high prices, effectively responding to changes in potato prices and adjusting the amount and period of listing in a timely manner. Combined with the important role played by the epidemic reserve mechanism, the potato reserve mechanism should be further enhanced to enable the timely transfer of storage in the event of an epidemic outbreak and effective support for potato consumption in disaster areas. In addition to increasing the construction of storage facilities, the efficiency of the use of existing facilities, such as cold storage, human protection projects and traditional cellars, should be enhanced. Market-based operations should also be improved to strengthen daily monitoring and the regular evaluation of storage enterprises, and on this basis, the establishment of a public system of potato reserve enterprise credit and qualification classification and grading system should be investigated.