China’s maize planting area has remained stable at over 600 million mu year-round, and production has been around 250 million tons year-round. These large production scales and yields give China a global leading position in the maize industry. China rivals world-renowned maize-producing regions such as the U.S. Maize Belt and the Ukrainian Maize Belt, which together constitute one of the largest maize-producing regions in the world. Maize is an important grain crop in China, which plays a vital role in guaranteeing China’s food security and serves as an essential feed source (He et al., 2020; Xin, 2022). After 2000, the sown area and total output of maize in China have risen (Figure 1). In 2000, the sown area of maize in China was only 23,056.1 thousand hectares, and by 2021, the sown area of maize in China will be 43,324.2 thousand hectares, with a growth rate of 87.91%. At the same time, China’s total maize production has risen from 106,000 kilotons in 2000 to 272,551 kilotons in 2021, a growth rate of 157.12%. It should be noted that the sown area and total production of maize began to decline after 2016 due to the MSSR in China.

Maize is mainly grown in 20 provinces in China, namely Hebei, Shanxi, Inner Mongolia, Liaoning, Jilin, Heilongjiang, Jiangsu, Anhui, Shandong, Henan, Hubei, Guangxi, Chongqing, Sichuan, Guizhou, Yunnan, Shaanxi, Gansu, Ningxia and Xinjiang. Inner Mongolia, Jilin, Liaoning, and Heilongjiang are the main maize-producing provinces with more favorable climatic conditions. These four provinces were selected by Chinese policy to pilot the reform of the maize storage system. In 2021, their planted area was 17,854.2 thousand hectares, accounting for 41.2% of the national total. Figure 2 shows the sown area and production of maize in these four provinces in 2021. Among them, Heilongjiang leads the way regarding maize sown area and total production.

Since 2004, China has conducted a series of explorations of the grain purchase system to increase farmers’ incentives to grow grain. Among them, the maize purchase policy is the core initiative of China’s grain price support. Affected by the financial crisis of 2008, maize prices in the international market dropped significantly, and domestic maize prices also faced downward pressure. The government initiated the maize purchase policy to smooth out the fluctuation of maize prices. Implementing the policy not only protects the income of maize farmers but also mobilizes farmers’ enthusiasm to grow grain to a large extent (Gale, 2013; Hejazi and Marchant, 2017). However, this market purchase has distorted the operation of the maize market, resulting in an imbalance between maize supply and demand (Gu et al., 2018). On the one hand, domestic maize is much higher than the international market price, and the import volume of maize surges yearly. On the other hand, the competitiveness of domestic maize has declined, resulting in a large amount of maize being converted into stockpiles, and the financial burden of national grain storage has increased (Gong et al., 2021).

The Chinese government decided to abolish the maize storage system in March 2016, fully liberalize the maize market, and adopt a new form of “market-based pricing and producer subsidies.” Maize prices are determined by market supply and demand, and all market players can trade freely according to market prices, and the state subsidizes producers. The policy reform shows that the government intervention is gradually weakened, the maize price mechanism is gradually formed, and the market will play a decisive role in the allocation of resources (Han et al., 2022). Based on guaranteeing that the income of maize farmers will not be affected too much, the government gradually adjusts the maize planting structure through the market mechanism and producer subsidies, thus improving the competitiveness of maize (Ye et al., 2022). In addition, unlike other staple grains such as wheat and rice, maize is used chiefly for deep processing such as feed and alcohol. The MSSR is mainly centered on “difficulty in selling grain” and “difficulty in exporting, “the state and the pilot district governments have introduced the support policy, credit guarantee fund for purchasing loans, and the new maize. The state and pilot zone governments have successively introduced support policies for purchasing new maize, credit guarantee funds for purchase loans, and free high-speed transportation of maize to support the MSSR.

The study of Adamopoulos et al. (2022) shows that low agricultural productivity in developing countries mainly comes from small planting scale, low productive investment, and inefficient allocation of agricultural resources. To further explore the impact of MSSR on farmers’ maize productivity, this paper combines the existing research to construct the analytical framework of “institutional reform–planting scale/productive investment/land allocation efficiency—maize productivity.” The analytical framework is shown in Figure 3.

First, MSSR can increase farmers’ maize productivity by expanding the planting area. The scale of operation is an essential determinant of agricultural productivity. With increasing scale, farmers can adopt more advanced technology for production, thus increasing maize productivity (Coelli et al., 2003; Giang et al., 2019). Due to the specialized character of agricultural assets, MSSR can increase farmers’ maize planting area by expanding the planting area. Asset specialization refers to the difficulty of changing the use of certain assets that are locked in for a specific purpose during the production process. If they are used for other purposes, they may become less valuable or even worthless (Klein et al., 1978). In the pre-MSSR period, farmers’ average maize yield will be reduced, and some large-scale farmers are unable to withdraw from production due to the strong specialization of the input assets, so in order to maintain the income, they can only choose the production method of “many kinds of multi-profit” to carry out production. Therefore, the MSSR can increase maize productivity by adjusting the planting scale of farmers.

Secondly, MSSR can increase farmers’ maize productivity by increasing their productive investment. Increasing productive investment in agriculture can improve agricultural production conditions and thus increase agricultural productivity (Hu et al., 2023). On the one hand, MSSR can enhance the short-term investment of farm households. After the reform, farmers’ returns consisted of maize selling prices and producer subsidies. The market determines the price of maize and presents a significant price difference according to the quality of maize. Farmers will use organic fertilizers instead of chemical fertilizers, selected seeds, and other initiatives to improve the quality of maize to obtain more income. On the other hand, MSSR can increase farmers’ long-term investment. Farmers will increase long-term investment to improve soil fertility to obtain long-term profits. In addition, maize producer subsidies can increase maize productivity through human capital accumulation. After receiving subsidies, farmers can participate in technical training to accelerate human capital accumulation, thus increasing maize productivity (Li and Lin, 2022). Therefore, MSSR can increase maize productivity by increasing the productive investment of farmers.

Thirdly, MSSR can improve farmers’ maize productivity by optimizing land resource allocation. Improved resource allocation efficiency is important to enhance agricultural productivity (Yang et al., 2022). Land transfer is an essential means to improve agricultural resource allocation. Land rent is an important influence factor on land transfer rate, and when land rent is too high, land transfer will be hindered. It can be seen that the rent of land transfer will be affected by the fluctuation of maize price, and the MSSR can directly affect the maize price and, thus, the rent of land transfer. During the implementation of the maize purchase policy, the “government purchase” makes the land rent rise, and land transfer is hindered (Liu and Qin, 2019). After the MSSR, the price of maize began to fall, and the expected income of farmers decreased. At this time, the rent of land transfer also fell sharply, and the land transfer rate increased (Xu et al., 2021). Therefore, MSSR can improve the degree of land mismatch to increase the maize productivity of farmers. In summary, we found that MSSR can increase maize productivity by expanding the area of maize cultivation, increasing productive investment in maize, and optimizing the allocation of land resources.

The data used in this paper come from the National Rural Fixed Observation Points (NRFOP) of the Ministry of Agriculture and Rural Affairs of China, which is an official micro-research data of the Chinese government (Chari et al., 2021; Ge et al., 2023). This survey began in 1986 and is China’s largest farm household-level tracking research data (Chen, 2012). This data has several advantages: its large and wide sample size and survey scope, making it highly representative. Currently, the annual sample size of the survey is about 21,000 farm households, involving 360 villages and 31 provinces (cities and districts) across China. Then, the content of the survey is rich and exhaustive. The data includes many aspects such as land, household management, inputs of production materials, household consumption, and so on. In particular, the data are very detailed on the inputs and outputs of households in the production of different crops, including specific data on the inputs and outputs of food production of farm households in each region. Moreover, the quality of the data is high. The survey team used a bookkeeping method to collect data from farmers. Since the data is tracking data, almost all interviewed farmers are familiar with the research process, which is undoubtedly significant for improving data quality.

In the existing literature, Shen and Yao (2008) explored the sampling of the data and described the initial sampling rules for the data. At the beginning of the survey initiation for the NRFOP in 1986, the researcher adopted a combination of typical and random sampling to conduct the survey. In the first step, typical sampling was adopted to determine the research villages. The province was divided into three regions based on geographic stratification: plains, hills, and mountains. The counties in the province were divided into three types of counties: high, medium, and low levels based on income stratification. The representative counties were selected as the research regions. In the second step, villages were randomly selected in the above research area to conduct farm household interviews, and 50 to 120 households were randomly selected in each village for the research. Benjamin et al. (2005) and Wang et al. (2020) compared and analyzed the fixed observation point data with the census data from the National Bureau of Statistics and found that the two data were relatively consistent regarding agricultural production, household income, etc. Adopting rural fixed observation point data for this study is more reasonable. We processed samples with entry errors, coding errors, and missing data, excluding farmers not engaged in maize production. Due to different research purposes, we selected a sample of maize farmers from 2003 to 2019 when measuring maize productivity. In analyzing the impact of MSSR on farmers’ maize productivity, we used the sample of maize growers from 2014 to 2019 because the MSSR was piloted in 2016.

The MSSR started in 2016 and was piloted in Liaoning, Jilin, Heilongjiang, and Inner Mongolia, which provided a quasi-natural experiment for this study. However, development economists almost universally recognize the endogeneity of policy change (Acemoglu et al., 2001). The traditional DID model presupposes a parallel trend between the experimental and control groups before the policy shock; otherwise, it would not be able to exclude the interference of other factors on the policy effect. However, the pilot of the MSSR may not be a random event, which would not satisfy the parallel trend assumption and thus lead to biased estimates of policy effects. This paper draws on the PSM-DID model proposed by Heckman et al. (1998) to study the effect of MSSR on farmers’ maize productivity.

The estimation of the PSM-DID model is divided into two stages. The first is the PSM estimation. We follow the year-by-year matching method to find the control and experimental groups. We explicitly use the Logit model to estimate the propensity score and K-nearest neighbor matching to select the matching samples. The matching model is Equation (4):

Where P denotes the probability that a farmer participates in the MSSR, treat denotes whether the farmer participates in the MSSR. Participation is assigned a value of 1, and non-participation is assigned a value of 0. L denotes the Logit regression model. X denotes the covariate used for PSM. After matching, 41,841 samples are obtained in this paper, including 10,737 experimental and 31,104 control group samples. The second step is DID estimation. In this paper, the following model expressed as Equation (5) is constructed to analyze the impact of MSSR on farmers’ maize productivity:

Where i represents the region and t represents the year. MTFP represents the total factor productivity of maize. DID represent the policy variables of MSSR, X represents the control variables, μi and θt represent year-fixed effects and individual fixed effects, respectively, εit represents the classical randomized disturbance term, and α and β represent the parameters to be estimated. The control variables in this paper contain two main categories: one for household head characteristics that affect maize productivity and the other for household characteristics that affect maize productivity.

In this paper, we use the NRFOP farm household panel data to estimate the MTFP of households. Table 2 shows the estimation results of the Cobb–Douglas production function by applying the growth accounting method. The elasticities of physical capital, land, and labor in the maize production function are 0.377, 0.496, and 0.026, respectively, which are highly consistent with the characteristics of resource endowment for maize production in China. Land is the most rare element in maize production, capital plays an equally important role, and only labor is affluent (Li et al., 2010). The sum of the three coefficients is 0.899, indicating that maize production is diminishing returns to scale, which is consistent with the findings of Chari et al. (2021).

After measuring the elasticity of each input factor, we calculate the MTFP of farm households in China from 2003 to 2019 using Equation (3). This paper adopts two methods, the simple average and the median, to calculate the MTFP of farm households in all the years. The measurement results are shown in Figure 4. We can get the following several conclusions. First, MTFP as a whole shows a steady upward trend, which is consistent with the study of Lou (2022). The difference is that this paper measures MTFP from the macro level while we calculate MTFP from the micro farmer level. From the mean calculation, MTFP increased from 3.998 in 2003 to 4.281 in 2019, with a growth rate of 7.079%. Secondly, in the short term, there was a decrease in MTFP in some specific years, such as 2004, 2008, and 2017.

This part mainly analyzes the average treatment effect of MSSR on farmers’ maize productivity. The regression results are shown in Table 3. Model 1 represents the results without controlling any variables, Model 2 represents the results after controlling for farm household characteristics, and Model 3 represents the results after controlling for all control variables.

The results of Model 3 show that MSSR increases maize productivity by 6.0% on average. The impact of MSSR on farmers’ maize productivity may come from two aspects: first, MSSR reduces the area of maize planted in non-dominant zones, improves the degree of mismatch of resources, and thus improves maize productivity. Most of the reduction of maize in these areas was in non-dominant areas. Second, after the MSSR, maize subsidies with noticeable differences will drive the concentration of maize production to the advantageous areas, thus promoting the scale of maize production and giving play to comparative advantages (Liu and Qin, 2019; Di et al., 2023).

The effect of control variables on the maize productivity of farmers can be obtained from Model 3 in Table 3. First, Education shows a negative relationship with maize productivity, which is consistent with the findings of Gao et al. (2017). Education is an important means of human capital investment, so education can increase agricultural productivity. However, as the level of rural education increases, the probability that farm households may go out to work rises, accelerating the transfer of agricultural labor and thus reducing agricultural productivity. Second, Train significantly increases maize productivity, which is basically in line with the study of Li et al. (2010). Farmers can understand the latest national agricultural policies and regulations by participating in agricultural vocational education, learning cutting-edge agricultural production and management technology, and increasing maize productivity. Again, there is a negative relationship between Farmland and maize productivity. The studies of Gautam and Ahmed (2019), Julien et al. (2019) also showed a negative relationship between farmland size and agricultural productivity. Compared to large-scale households, smallholders have the advantage of intensive farming, and their combined utilization of land, machinery, and other factors is more efficient, resulting in higher maize productivity. At the same time, small farmers can significantly resist natural and market risks due to the diversity of planting structures and the difficulty of planting structure adjustment, thus gaining advantages. Family labor and maize productivity are negatively correlated. Chinese agriculture is in small-scale operation, and in the process of mechanization promotion, maize production often does not need much labor, so the relationship between the number of labor and maize productivity is negative. Finally, Income significantly increases maize productivity, which is consistent with the findings of Ye et al. (2023a). As Income increases, farmers will increase investment in maize, which will promote technology adoption and thus increase maize productivity. In addition, an increase in income will alleviate the financial constraints of agricultural production, thus promoting agricultural production (Yi et al., 2015).

The empirical results above obtained the average effect of MSSR on maize productivity, but it is difficult to reveal the dynamic effect of MSSR. Given that the MSSR is divided into two phases, the first is eliminating the maize purchase price (the first year), and the second is the implementation of the maize producer subsidy (the second year and beyond). Therefore, analyzing the dynamic effects of the MSSR on farmers’ maize productivity has important policy implications. The estimation results are presented in Table 4.

From Model 1, the coefficients for the second to fourth years of MSSR implementation are 0.048, −0.002, and 0.130, with a mean value of 0.059, smaller than the estimated parameters for the first year of MSSR implementation. By comparing the coefficients, we can infer that eliminating maize purchase price enhances maize productivity more than implementing maize producer subsidy. Market-oriented reforms will effectively promote the allocation efficiency of resources, thus significantly increasing maize productivity. Market distortions can reduce factors’ production costs and promote technological progress (Chari et al., 2021). In addition, because China has just started implementing maize producer subsidies, there are problems such as untimely subsidy methods and insufficient disbursement amounts, which can also limit the improvement of maize productivity by producer subsidies. Therefore, continuously optimizing the maize producer subsidy and giving full play to its policy effectiveness is a future reform direction for China’s price support policy.

To understand in which groups the MSSR policy did not have an effect, this paper conducts a series of heterogeneity analyses. In this paper, we analyze the heterogeneity based on different land sizes, levels of specialization in maize cultivation, the degree of part-time work of farmers, and the productivity of farmers.

MSSR is a significant component of China’s market-driven reform of grain support policy. Evaluating its productivity impact can optimize the policy, optimize the policy, and offer insights for policy reform in other developing countries. As such, this paper empirically analyzes the MSSR’s impact on maize productivity by utilizing the PSM-DID model based on China’s official National Rural Fixed Observation Point Farm Household Survey data. The paper’s findings can be summarized as follows: Firstly, MSSR has a significant enhancing effect on farmers’ maize productivity, with results that remain robust after sample and method changes, as well as placebo testing. Secondly, the MSSR is divided into two phases: the elimination of the purchase price of maize and the implementation of the maize producer subsidy, whereby the elimination policy’s effect is greater than that of the producer subsidy’s implementation. Third, the study on heterogeneity found that an increase in farm scale, specialization level, and part-time employment of farmers corresponded to a more pronounced impact of MSSR on their maize productivity. Fourth, unconditional quantile analysis revealed that MSSR lowers maize productivity in the high quartile while raising it in the low quartile. These findings imply that MSSR is a viable method for reducing the gap in maize productivity between farmers.

To enhance national food security, developing the MSSR further and establishing a reasonable subsidy system is recommended. A crucial first move is increasing subsidies to maize producers. These subsidies can alleviate the financial constraints faced by farmers and improve their maize productivity. Secondly, it is necessary to implement differentiated subsidies for maize producers. During this process, we must resolve issues related to unbalanced policy effects, prioritize the needs of vulnerable groups during policy implementation, and guide them toward structural adjustments in their planting practices. Finally, we need to ensure the accuracy of subsidies and increase their matching degree so that subsidies are allocated based on the amount of maize grown. This will effectively safeguard the income of maize growers by providing subsidies per acre of maize produced.

Exploring diversified maize-producer subsidy policies. The policy should shift from single-producer subsidies to diversified support in the future. It is necessary to improve the infrastructure construction of maize-advantaged areas, strengthen support for maize planting technology, and provide favorable external conditions for maize planting. For large-scale maize growers, it is possible to transition from a single producer subsidy to a diversified form of subsidy, such as capital credit and irrigation facility inputs. This would increase the ability of large-scale households to expand reproduction and withstand natural risks. Different producer subsidies should be allocated to various regions. For instance, in areas where maize cultivation is advantageous, the focus should be on supporting the production of large-scale farmers. In arid zones, the focus should be on producing various grains. In the pilot areas of the national “grain-to-fodder” program, the emphasis should be on producing maize seedlings.

The influence of additional subsidies on maize production behavior among farmers must be considered. Empirical evidence indicates that adjustments to soybean policy and direct grain subsidy policy also influence maize target fund payments. In the areas where reform pilots are being conducted, there is a significant substitution between maize, soybeans, and rice. Consequently, when optimizing the subsidy for maize producers, these three crops’ interplay policies must be considered. It is imperative to consider the coordination of subsidies for soybean and maize producers and adjust the ratio of maize and soybean cultivation through these disbursements.

The implementation of MSSR plays a vital role in the modification of China’s agricultural price support system, potentially impacting agricultural production in various ways, including macro-structural adjustments of agriculture and the growth of the agricultural economy. This study exclusively evaluates the policy from a maize productivity standpoint. Future research should expand on this by exploring the above directions. In addition, as the aim of this research paper is to guarantee food security, subjective evaluations are excluded. Hence, only the maize productivity of farmers is taken into account. However, it should be noted that farmers may encounter structural adjustments of food crops and cash crops during actual cultivation, which can potentially impact the farmers’ overall productivity. Future studies can be expanded from this perspective.