Yang Yu, and Pan Yueping

Chengdu University of Technology

Abstract: A critical method of ensuring grain production is to increase the total factor productivity (TFP), and the key measure to increase the TFP of grain production lies in the construction of agricultural public infrastructure. For this topic, existing literature lacks systematic and empirical analysis. Therefore, research on the influence of agricultural public infrastructure on the TFP of China’s grain production has relatively strong policy implications and theoretical value. For this study, we collected panel data for grain inputs and outputs as well as for agricultural public infrastructure in China’s provinces (autonomous regions/municipalities) from 1990 to 2017, and adopted the stochastic frontier function (SFF) approach to measure the TFP of provincial-level grain production. Through this empirical study, we analyzed the influence of agricultural public infrastructures, such as irrigation, roads, and electric power facilities on the TFP of China’s agriculture. We found that such facilities have a positive influence on the TFP of grain production. Specifically, when the input for irrigation facilities is increased by 1 percent, the TFP of grain production will rise by 5.74 percent. Based on this finding, policy recommendations are proposed for enhancing grain TFP through agricultural public infrastructure construction.

Keywords: agricultural public infrastructure, grain production, total factor productivity, empirical analysis

Introduction

The Chinese government has achieved remarkable results by abiding by the principle of basic food self-sufficiency based on domestic grain production and attaching great importance to improving grain productivity in its economic policies. China’s grain output in 2020 reached 1.339 trillionjin(669.5 billion kg), increasing by 11.3 billionjin(5.65 billion kg) from the previous year, setting a record of good harvests in grain production for 17 years in a row and ensuring a selfsufficiency rate of grain and food security. On the other hand, negative factors are accruing that may affect China’s grain production, including a gradual decrease in arable area, soil degradation, severe insufficiency inland reserves, and an increasing shortage of water resources (Chen Shuai et al., 2016; Ali et al., 2017; Xie et al., 2020). Such constraints are becoming more striking in the context of rural economic transitions and the rapid development of agricultural modernization (Huang Jikun et al., 2019). Therefore, under the constraints of resources and the environment, improving grain productivity has become a key issue for the country. Improving agricultural TFP is also vital for realizing agricultural modernization and sustainable development (Cao & Birchenall, 2013).

The construction of agricultural public infrastructure is the key measure to increase the TFP of grain production. Currently, it is one of the biggest issues that need to be addressed in China’s grain production (Kong Xiangzhi, 2016). Covering irrigation, road, and electric power facilities, agricultural infrastructure can be seen as fundamental investments for large-scale operations and technical progress (Chen Hongwei & Mu Yueying, 2021). The construction of agricultural public infrastructure plays a vital role in supporting the TFP of grain production to contribute to rural economic development, and the Chinese government has emphasized the importance of agricultural public infrastructure construction in its No. 1 central document (the first policy statement released each year) many times (Yang Jun et al., 2019). For instance, the No. 1 central document for 2016 pointed out that efforts should be made to develop high-standard farmland and construct farmland water conservancy facilities on a large scale, to accelerate the construction of infrastructure for rural life and production, and to constantly reinforce the foundation of modern agriculture (Cai Baozhong & Zeng Fusheng, 2017). The construction of agricultural infrastructures, such as irrigation works, mechanical and electric power facilities, roads, and logistics facilities, is of great importance to resisting natural disasters and ensuring production (Zhu Jing & Jin Yue, 2017). Increasing inputs in agricultural infrastructure is actually an effort to reserve grain production capacity (Liang Ziqian & Li Xiaojun, 2006; Teruel, 2005). As early as 1978, Theodore Schultz mentioned in his bookTransforming Traditional Agriculturethat infrastructure for transportation and irrigation facilitate agricultural production by reducing circulation and input costs (Zhu & Jin, 2017). Many scholars also found that agricultural infrastructure can enhance agricultural TFP (Deng Xiaolan & Yan Weibo, 2018; Chen & Mu, 2021; Huang Jinbo & Zhou Xianbo, 2010). The construction and maintenance of farmland irrigation facilities are also important measures to guarantee the growth effect of the TFP (Luan Jian & Han Yijun, 2020).

We focused our analysis on the influence of agricultural public infrastructure on the TFP of grain production. But considering that there are various kinds of agricultural public infrastructure that can enhance TFP in a variety of ways and to different extents, it was difficult to make an exhaustive study. Therefore, in addition to theoretical analysis, we used panel data from 25 provinces (autonomous regions/municipalities) in China to make an empirical study on the influence of three major types of agricultural public infrastructure (farmland water conservancy, roads, and electric power facilities) on the TFP of grain production to offer empirical evidence for the proposal of strengthening agricultural infrastructure construction to improve grain productivity.

Literature Review

Many scholars have conducted research on the influence of agricultural public infrastructure relative to the TFP of grain production. However, much of the research studied theories on the TFP, and only a few delved into the actual TFP and the mechanisms that directly influence the TFP of grain production. After a thorough reading of these papers, we found that they can be roughly classified into two categories.

Studies on the TFP of Grain Production

On the measurement of the TFP of grain production.

As China is faced with growing resource and environmental constraints on agricultural production, there is less and less room for improving agricultural output by increasing inputs of natural resources and the factors of production. In this case, the TFP is of more research significance for the sustainable development of China’s agriculture. Robert Solow was the first scholar to define the portion that could not be accounted for by the explanatory variable in the production function as the residual of technological advanced (“the Solow residual”), which he defined as the TFP (Tian Hongyu & Zhu Zhiyong, 2018). Input indicators of grain the TFP generally covers include five aspects: sown area, agricultural machinery power, irrigation area, the input of draft animals, and consumption of chemical fertilizers (Wang Chen et al., 2017). The theory of the TFP was later introduced to China. There are generally four types of methods for TFP measurements: the production function approach, the growth accounting index, the Malmquist productivity index based on data envelopment analysis (DEA), and the stochastic frontier approach (SFA) (Yu Kang et al., 2012). Many domestic scholars thought DEA can measure the efficiency of a production system featuring many inputs and outputs and is more in line with production reality, so they used a DEA-based Malmquist approach to measure China’s grain TFP (Ma Linjing et al., 2014; Li Gucheng et al., 2015; Jiang Songying et al., 2016; Xie Feng, 2016; Zhang Liguo & Bao Bingfei, 2016; Chen Qiufei and Tan Xiaodong, 2017; Zhuo Yue & Zeng Fusheng, 2018; Tian & Zhu, 2018; Deng & Yan, 2018). Other scholars used a slack-based measure (SBM) in combination with the Malmquist approach to calculate and decompose the TFP (Fan Lixia, 2017; Song Haifeng & Liu Yingzong, 2019). Still, other scholars used SFA functions to measure the TFP of China’s grain production (Huang & Zhou, 2010; Xiao Saili et al., 2015). Among them, Li Gucheng et al. (2007) used an SFA approach based on Trans-log functions to make a systematic analysis in the TFP and technological efficiency of rural family businesses through microdata from rural households in Hubei Province from 1999 to 2003. In recent years, the Torgvist-Theil index has also been widely used to measure the changes of the TFP. Chen Weiping and Ding Ying (2007) used such an index to measure the TFP of China’s grain production and then calculated the contribution of the TFP to the output.

On growth sources of the TFP of grain production.

Fan (2017) holds that based on the significant differences in the growth of grain in the provinces concerned, the TFP of the provinces can be classified into three types. The first type features “continuous decline,” including eight provinces, such as Anhui, whose TFP value is lower than 1. The drop in their grain TFP is due to efficiency deterioration and technology decline. The second type features “slow rise,” including 11 provinces such as Guangdong, whose TFP value is between 1 and 1.01. The growth in their grain TFP is mostly due to technological progress. The third type features “continuous rise,” including nine provinces such as Jilin, whose TFP value is greater than 1.01. The growth in their grain TFP is driven by both technological efficiency improvements and progress. In China, many scholars think that technological progress is the major force to promote the growth of the TFP of grain production, while technological efficiency improvement does not matter much or even impedes the growth of agricultural TFP (Min Rui, 2012; Xie Feng, 2016; Chen & Tan, 2017). Other scholars hold that technological efficiency and progress are the two factors that promote increases in the TFP (Gao Shuai & Wang Zhengbing, 2012; Xiao Hongbo & Wang Jimin, 2012; Ma Linjing et al., 2014). Ma et al. (2014) hold that the dual driving forces are vital to the sustainable development of agriculture and grain production in the future.

On influence factors of grain TFP growth.

There are still insufficient studies that have taken human capital into the framework for analyzing agricultural TFP growth and expanding production limits (Ma et al., 2014). Chen and Tan (2017) evaluated TFP growth from four input indexes: total power of agricultural machinery, effective irrigation area, consumption of chemical fertilizers, and sown area of grain crops. Tian and Zhu (2018) hold that chemical fertilizers, agricultural machinery, irrigation, and grain price are key factors to promote the increase in grain productivity and their degrees of contribution can be shown as: price ＞ chemical fertilizers ＞ agricultural machinery ＞ irrigation. Ma et al. (2014) think that agricultural economic development levels, rural education levels, land quality, grain sowing proportions, level of exposure to natural disasters, per capita scale of operations, and the level of mechanization for grain growing have a positive or negative influence on grain TFP growth.

Studies on the Influence of Agricultural Public Infrastructure on the TFP of Grain Production

Infrastructure, like the factors of production such as labor and capital, influences production, and can influence economies of scale, products, and markets, promote technological progress, and enhance agricultural productivity. Zhang Yan et al. (2011) hold that rural irrigation facilities are fundamental for guaranteeing food security and production. Many scholars analyzed the relationships between agricultural public infrastructure and grain output in a quantitative way. For instance, Cai and Zeng (2017) proved by using simultaneous equations that inputs in agricultural public infrastructure have increased grain yields in 28 provinces of China over 13 years. Xie Xiaorong et al. (2014) made an empirical study on the relationships between agricultural public infrastructure and grain output by using panel data from the regions. Ma Peiheng (2011) conducted an empirical analysis on the effect of agricultural public infrastructure inputs on increasing grain yield in Henan province using the DEA approach and making the amount of rural social fixed asset investment a substitute variable for agricultural public infrastructure input. Based on the influence of agricultural public infrastructure on grain productivity, Zhuo and Zeng (2018) found that the lagged variables of farmland water conservancy and transport facilities had a significant positive influence on grain TFP in their study. Romeo G. Teruel and Yoshimi Kuroda (2005) also hold that agricultural public infrastructure, especially water conservancy and transport infrastructure, make huge contributions to the increase in grain productivity. Liu Qiong and Xiao Haifeng (2021) found that rural transport infrastructure had a noticeable positive spatial spillover effect on agricultural productivity in their study.

Some scholars delved into the path by which agricultural public infrastructure affected the increase in grain TFP. Zhu and Jin (2017) found that agricultural infrastructure enhanced TFP by increasing returns on investments in other private factors, optimizing the input structures of the factors of production, and serving as public investments to replace private input. The research by Chen and Mu (2021) shows that water-saving irrigation facilities improved grain TFP through technological progress and efficiency improvements, and the positive influence spiraled up with the rise in rural human capital, the per capita grain-growing scale, and the per capita GDP. Onofri and Fulginiti (2008) hold that infrastructure affects agricultural production by stimulating long-term needs for private capital through increasing returns to the scale of input factors. In addition, improvements in agricultural public infrastructure are also investments made to promote scaled operations of agriculture and to save costs, thus enhancing grain productivity and competitiveness (Mamatzakis, 2003; Teruel & Kuroda, 2005; Zeng Fusheng & Li Fei, 2015; Zhu & Jin, 2017). But some scholars have different opinions. For instance, Wu Qinghua et al. (2015) put forward the possibility that infrastructure construction may indeed promote optimal configurations of the factors of agricultural production but lead to an increase in agricultural production costs.

The above literature shows that previous domestic studies mostly focus on the influence of agricultural public infrastructure on agricultural production but with less attention to the effects of infrastructure on grain production output. But since grain production constitutes a major part of all agricultural production, and because of the substitution effect, these papers shed much light on our research as well. Agricultural public infrastructure construction is conducive to improving grain productivity and reducing production costs and thus promoting output growth. Based on the above research findings, we assumed that certain correlations might exist between agricultural public infrastructure and the TFP of grain production. Although scholars have studied the influence of agricultural public infrastructure on grain TFP, they generally chose the data of only one province for analysis. There are few papers making analysis of nationwide data in this regard. Therefore, we present a detailed study on the influence of three key facilities (farmland water conservancy, transport, and electric power) of agricultural public infrastructure in 25 provinces (autonomous regions/municipalities) on grain TFP in China. We then analyzed our findings and offered recommendations for improving the TFP of grain production in China.

Measurement of Grain TFP

Modeling

The SFA model is used to study the maximum output achievable under established technological conditions. It can be applied to measure technological efficiency and progress of grain production. A mathematical SFA model can be expressed using the following equation:

Where:yitstands for the output of theithunit in thetthyear;f(xit,t) stands for the maximum economic yield achievable under a perfectly efficient state;xitstands for the investment amount of economic factors of theithunit in thetthyear;tmainly refers to the time trend variable;vitrefers to errors that are generated in the economic process and are not subject to human intervention, including statistical errors or errors out of market changes;μitmainly refers to a non-negative variable for technical loss of theithunit in thetthyear. The word “non-negative” here means that the actual output of the decision-making unit (DMU) cannot exceed the production frontier and must be above or below it. TE refers to the variable for technological efficiency.TEit= exp(-μit) refers to the gap between the expected yield and the actual yield due to unordered and inefficient production. The following mathematical equation is obtained by expressing the SFA model in a logarithmic form.

A translog function mode was selected to build the stochastic frontier production model for the grain production of 25 provinces (autonomous regions/municipalities) in China. It was also assumed that technological progress was neutral. The production function was expressed as a translog function, and factor inputs of the production function were set to consist of physical capital inputs (Sfor average seed input permu①One mu is equal to about 667 square meters.,Ffor average chemical fertilizer input permu,Mfor machinery input) and labor input (L). In this way, the SFA logarithmic production model in this article can be expressed as:

Where:Sitstands for average seed input permuof DMUiin thetthyear;Fitstands for fertilizer input permuof DMUiin thetthyear;Mitstands for average large equipment input permuof DMUiin the th year;Litstands for labor input of DMUiin thetthyear;Tstands for the time variable;β0,β1,β2, andβ3mainly refer to each predictor variable; the stochastic error term (vit) presents the normal distribution ofrefers to the technical inefficiencies. Supposethen the influenceηin this equation refers to the specific influence of the time variable onμit.

In terms of γ, one of the most important reference values for judging the suitability of the SFA model, some scholars hold that covariance parameters should be expressed by γ=σ2μ/σ2μ+σ2v, where γ∈[0, 1] mainly refers to a situation in which there is a large gap between the actual yield and the expected yield due to low production efficiency. Where γ is approaching 0, the gap between the actual yield and the possible maximum yield is mainly due to the variablevit. If γ=1, it means that the deviation of the actual yield from the production frontier is fully due to production inefficiency and irrelevant to stochastic errors. Relevant estimators can be obtained through related SFA equations. Then the TFP of each province (autonomous region/municipalities) in the years concerned can be found by using these equations.

Data Source and Variable Selection

Based on the research nature and orientation of this article as well as data availability, input and output panel data for the SFA logarithmic production model herein came from real data for the 25 provinces (autonomous regions/municipalities). As the other three municipalities (Beijing, Shanghai, and Tianjin) are not major grain-producing areas, they do not have data about costs permu. Tibet autonomous region and Qinghai province have little grain cultivation due to their special natural conditions. Therefore, the above areas are not included in the analysis herein. In addition, since it was from 1997 when Chongqing became a municipality, its data before 1997 was included in Sichuan province in our research. Owing to our data statistic specifications, panel data for the 25 provinces (autonomous regions/municipalities) is available only from 1990 to 2017. Unless otherwise specified herein, all the data was taken fromChina Rural Statistical Yearbook(1991-2018),National Agricultural Products Cost-Benefit Data Collection(1991-2018), National Bureau of Statistics, and a few other sources.

Variables were selected appropriately as follows: (a) Output variable. In order to ensure the objectivity and authenticity of this variable, we selected the average output of three major grain crops (rice, wheat, corn) permuin the 25 provinces (autonomous regions/municipalities) during 28 years (from 1990 to 2017) as the output variable in the SFA logarithmic production model. (b) Physical capital investment. To ensure the objectivity and authenticity of this variable, we selected average seed consumption permu, average chemical fertilizer consumption permu, and average agricultural equipment input permuto constitute the variable of physical capital investment. (c) Labor input. We took the average number of the labor force permuas the variable in the model.

Measurement Results and Analysis of the TFP

Based on the SFA panel model, which centers on the Cobb-Douglas production function as designed above, efficient, reliable analyses and verification were conducted on the panel data of the 25 provinces (autonomous regions/municipalities) during the 28 year period (from 1990 to 2017) by using software Frontier 4.1 to measure actual productivity of the TFP of grain production in these areas.

The data in Table 1 shows that:

Table 1 Efficiencies Estimated with SFF

(a) The coefficients of seed, labor, and machinery inputs per unit of land area are all positive but not significant, while the estimated coefficient of chemical fertilizer input is 0.665 and statistically significant at the 0.05 level, meaning that chemical fertilizers promoted an increase in output.

(b) The coefficient of the time variable is 0.0609, proving that with continuous improvements in agricultural public infrastructure, China’s grain production technologies have shown an obvious tendency for upgrading. With quantified data and symbolization, this means that the average annual technological progress is at a rate of 6.09 percent. In addition, the coefficient of the quadratic term of the time variable is -0.00169 and shows statistical significance at the 0.01 confidence level. It can thus be inferred that upgrading and transformations of grain production technologies did not have a significant positive correlation with the time variable. In other words, technological progress did not show a rapid growth tendency over time.

(c) The data also shows that there was a negative relation between the time variable and the variable of chemical fertilizer input. With technological progress, the contribution rate of chemical fertilizers dropped, and technological changes in grain production in the 25 provinces (autonomous regions/municipalities) featured excessive input of chemical fertilizers.

Note: ***, **, * represent the significance levels at 1%, 5%, and 10% respectively. Source: collection and analysis were made based on data from National Agricultural Products Cost-Benefit Data Collection (1991-2018) and China Rural Statistical Yearbook (1991-2018).

Table 2 shows annual average changes in technological efficiency (TE), technological progress (TP), and the TFP of major grain crops in the 25 provinces (autonomous regions/municipalities). The data in Table 2 shows that:

Table 2 Annual Average Changes in TFP, Technological Efficiency and Progress

Table 3 Descriptive Statistics of Data

(a) The technological level of grain production did not have significant development. Although the average actual technological efficiency of grain production in the 25 provinces (autonomous regions/municipalities) reached 0.1453 percent, it should be noticed that some of these provinces (autonomous region/municipalities), such as Yunnan, Jilin, Sichuan (including Chongqing), Shandong, Guangxi, Xinjiang, Hebei, Hubei, Hunan, Guizhou, Shaanxi, and Heilongjiang, still witnessed negative growth in this aspect. Among others, Heilongjiang’s annual average growth in technological efficiency of grain production was the lowest (at -1.605 percent), while Jiangxi’s figure was the highest (at 2.276 percent). The growth rates of technological efficiency of Liaoning, Fujian, Gansu, Hainan, Zhejiang, Jiangxi, Anhui, and Inner Mongolia were higher than the average value, probably due to rapid economic development of these provinces (autonomous regions), which greatly promoted grain production in these areas.

(b) Grain production technologies made rapid progress. Overall, the average annual growth rate of grain production technologies in the 25 provinces (autonomous regions/municipalities) was 0.8584 percent, with Guizhou having the fastest progress (annual average rate at 1.31 percent). Twelve provinces, including Yunnan, Sichuan (plus Chongqing), Anhui, Shanxi, Jiangxi, Henan, Hainan, Hunan, Hubei, Guizhou, Shaanxi, and Heilongjiang, had growth rates above the average value in this aspect.

(3) The TFP of grain production exhibited continuous growth. Except for Heilongjiang, the 25 provinces (autonomous regions/municipalities) have all had continuous growth in the TFP since 1990. For 13 provinces (autonomous regions), including Inner Mongolia, Anhui, Shanxi, Guangdong, Jiangxi, Henan, Zhejiang, Hainan, Gansu, Fujian, Guizhou, Liaoning, and Shaanxi, their TFP growth rates were greater than the average value of the 25 provinces (autonomous regions/municipalities). Except for Heilongjiang, all the other provinces (autonomous regions/municipalities) which had negative growth in technological efficiency, witnessed the same tendency in terms of their the TFP growth and technological progress, so that the latter has become the source of their TFP growth. Such TFP growth is regarded as a growth mode driven by technological progress instead of improvement in existing technological efficiency since technological progress has made up for the slow change in technological efficiency. These provinces (autonomous regions/municipalities) did not make significant achievements in promoting the popularization of agricultural technologies, so that their efforts in this regard should be further strengthened.

(d) The improvement in technological efficiency will be a potential driving force for TFP growth. With their current technological levels, the 25 provinces (autonomous regions/municipalities) did not bring their agricultural technologies into full play so that their technological efficiency applied to grain production should be greatly improved. It is feasible to increase the TFP by enhancing technological efficiency, which will become a potential driving force for TFP growth.

Source: collection and analysis were made based on data from National Agricultural Products Cost-Benefit Data Collection (1991-2018) and China Rural Statistical Yearbook (1991-2018).

Empirical Model, Variable Selection, and Data Processing

Modeling

Based on the research designs of Hulten et al. (2006), and Liu Shenglong and Hu Angang (2010), we developed a dynamic panel econometric model for our analyses:

Where:i(= 1, 2,…25) stands for the 25 provinces (autonomous regions/municipalities);tstands for the year; the explained variable is the logarithm of the TFP;Iis the explanatory variable, standing for the three types of agricultural public infrastructure;uistands for the individual effect;εi,tis the typical stochastic disturbance.

Variable Selection

The explained variable referred to herein is the TFP of grain production calculated by the above equations. Each index is explained here.

Road facilities (Road).

This is represented by the density of rural roads. Subject to data availability, the density was obtained by calculating the lengths of rural roads in the 25 provinces (autonomous regions/municipalities) and dividing them by the corresponding provincial area. For the calculations of road length, we used the agricultural census data released by relevant institutions in China and found that less than half of the towns are accessible to Class II highways or above. Based on this situation, we made a prudent calculation of the lengths of rural roads by utilizing the difference value between Class III and Class IV highways and expressways, Class I and Class II highways, as well as out-ofgrade roads to effectively avoid spatial influence from rural road facilities.

Irrigation facilities (Irri).

In China, irrigation facilities cover a variety of structures that are mainly used to prevent and control droughts, floods, waterlogging, and salinization that affect farmlands and to improve agricultural production conditions. To accurately measure the overall condition of water conservancy facilities, we decided to use the ratio of effective irrigation area and sown area of crops.

Electric power facilities (Elec).

Complete electric power facilities can ensure stable power supplies, provide adequate power sources for crop production, and improve the utilization efficiency of existing factors of production. We selected per capita electricity consumption in rural areas to show the concentration and quality of rural power grids and measure the condition of rural electric power facilities.

Data Source and Descriptive Statistics

Owing to data availability, we used panel data from the 25 provinces (autonomous regions/municipalities) from 1990 to 2017. Unless otherwise specified, all the data was taken fromChina Rural Statistical Yearbook(1991-2018),National Agricultural Products Cost-Benefit Data Collection(1991-2018), National Bureau of Statistics Descriptive Statistics of variables are shown here.

Model Verification and Result Analysis

As the sample data involved in this study has a time span of 28 years, the differentiated generalized method of moments (GMM) was used to test and analyze the model to enhance the reliability of the test results.

Table 4 shows the model assessments. Notice that the first three models have only one independent variable while the fourth model has all the independent variables. The last three lines of the table show the testing of the practicality and validity of the differentiated GMM estimation. The P value was used for substitution. For the other four models, their P values of AR1 are all less than 0.05, while those of AR2 are all above 0.05. This proves that a first-order autocorrelation is present while a second-order autocorrelation is not. Therefore, the models we use comply with the standard for correlation tests.

Table 4 Estimates of Panel Models

In addition, for the first three models, coefficients and significance levels of the model variables are consistent with those of the fourth model variables. This means that the estimates are robust. The estimates show that variables for irrigation, electric power, and road facilities all passed the significance test and corresponding coefficients are positive. This means that such facilities are all conducive to enhancing TFP. Model 4 considers the influence of the three major types of agricultural public infrastructure. The following analysis mainly focuses on the estimates of Model 4.

The model data indicates that the elasticity coefficient is 0.0574, which proves that farmland irrigation facilities have a significant positive influence on the TFP of grain production. Such an influence can be revealed in many aspects. The construction of agricultural irrigation facilities can optimize the planting environment for grain production, thus laying a good foundation for upgrading and transforming agricultural planting technologies. Such facilities can improve agricultural production conditions, enabling wide planting of new crop varieties that have special irrigation requirements, expand the combination of existing technologies, and support the commercialization of agricultural, scientific, and technological achievements, and the inter-regional promotion of agricultural technologies. In the meantime, such facilities can reduce the impact of natural disasters, such as drought and flood on agricultural production, and enhance the efficiency of agricultural technologies in the application process. Both coefficients of irrigation facilities in Model 2 and Model 4 are significantly positive, demonstrating again that water conservancy facilities have a positive effect on grain TFP. This also explains that the gradual completion of China’s water conservancy construction for farmland is one of the key reasons for the rise in grain TFP.

Among the models, both coefficients of road facilities in Model 3 and Model 4 are significantly positive, signifying that road facilities also have a positive effect on grain TFP. The elasticity coefficient of rural road facilities is 0.0135, meaning that road facilities have a less significant effect on grain TFP than irrigation facilities. As the coefficient is still positive, it shows that the more complete the road facilities are, the faster agriculture will develop. Complete rural road facilities can reduce agricultural production costs, such as transportation costs of seeds, chemical fertilizers, and other means of production, depending on the agricultural products. Roads can facilitate the circulation of agricultural equipment and sales of agricultural products. They can also reduce the mobility costs of the factors of agricultural production and enhance the operational efficiency of agricultural machinery and the marketization degree of agricultural products. They affect agricultural TFP by influencing production costs and facilitating production and sales activities. In addition, He Zhongwei et al. (2008) hold that road facilities also have significant positive influence on the development of food markets, and they can improve the transportation efficiency of materials, contribute to a rise in trading volumes or prices, promote technological innovations and the application of new technologies, and enhance productivity. A well-established transportation network can also help popularize new innovations and technologies. Roads can enable agricultural producers to master and apply more advanced technologies and thus entitle them to the economic benefits made possible through scientific and technological progress.

Agricultural electric power facilities ensure the normal operation of agricultural machinery that uses electric power. Some irrigation facilities also require an electric power supply. Complete rural electric power facilities can reduce loss during the operation of agricultural machinery and enhance farmers’ productivity. For our models, when the coefficient of electric power facilities is 0.0205, its influence is relatively noteworthy at the significance level of 1 percent. Wang Xiaoqin et al. (2009) hold that rural irrigation and electric power facilities can overcome negative impacts from the factors of production, resources, and ecological conditions on agricultural productivity and enhance agricultural productivity and output even in the context of labor migration and the decreasing trend of cultivated land.

Conclusions and Policy Recommendations

We collected 1990-2017 data of grain production inputs and three major facilities of agricultural public infrastructure from 25 provinces (autonomous regions/municipalities) in China and measured the TFP of grain production in these areas during the period 1990 to 2017 using an SFA approach. The results indicate that during this period, China’s TFP of grain production presented a constant growth rate of 1.0037 percent annually. After the TFP of grain production was measured, GMM was applied to analyze the influence of agricultural public infrastructures such as roads, irrigation, and electric power facilities on grain crops, including rice, wheat, and corn. Our conclusions are: First, the three types of agricultural public infrastructure can all enhance the TFP of grain production. Specifically, with every 1 percent increase in the expansion of irrigation facilities, China’s TFP of grain production will increase by 5.74 percent. With every 1 percent increase in the construction of electric power facilities, China’s TFP of grain production will increase by 2.05 percent. This shows that among the types of agricultural public infrastructure, irrigation facilities have the biggest positive influence on the TFP of grain production. Second, irrigation and electric power facilities can effectively increase the TFP of grain production through technological progress and efficiency while road facilities can enhance increases through influencing technological efficiency. Third, the measurements of the TFP of grain production show that adjustments to agricultural mechanization, inputs of various factors, and planting structures can enhance the TFP of grain production.

Based on the above conclusions, we put forward the following policy recommendations:

(a) Construction of farmland water conservancy facilities should continuously be strengthened because they can significantly enhance the TFP of grain production. In the process of grain production, such facilities are always the most important part of agricultural public infrastructure. Constant efforts should be made to construct, repair, and maintain such facilities, especially small works like canals, ditches, weirs, and ponds, to develop a well-planned irrigation infrastructure network that is rational in scale, appropriate in location and density, and can play an effective role over the long term. In this way, they are expected to exert a greater positive influence on the TFP of grain production. The government should shift its focus to the central and western regions and the construction of small and mediumsized irrigation facilities according to financial circumstances, actual agricultural development, and localized agricultural public infrastructure demands. It should make innovations in the supply system of agricultural public infrastructure, provide incentives to encourage the participation of non-governmental organizations, and promote the building of democracy and improve democratic decision-making mechanisms for infrastructure in rural areas. It should also implement government subsidy policies for the operation and maintenance of irrigation facilities and adopt diversified management and maintenance mechanisms to ensure the functions of agricultural public infrastructure.

(b) Investments in road and electric power facilities should be increased. Based on our empirical results and related theories, roads, and electric power facilities also played a significant role in China’s TFP of grain production. Since 1990, China has increased its investments in agricultural public infrastructure, but in comparison with developed countries, it still has a long way to go in infrastructure construction. The government should continue to increase investments in rural road construction and explore diversified investments and financing modes with government funding as the major source to accelerate rural road construction, which will reduce farmers’ burdens. It should adjust measures according to local conditions and the electric load of different rural areas and implement a variety of power supply modes to satisfy the needs for rural economic development and household electricity consumption. It should also provide more policy and funding support to rural power grid upgrading and transformation and hydropower renovation for efficiency and expansion. More investments in road and electric power facilities are favorable measures to promote agricultural economic growth, which also benefits from modernized transportation and power networks, reductions in agricultural production costs, and adjustments to investment structures of agricultural infrastructure.

Local governments at various levels should fully recognize the importance of agricultural public infrastructure construction, expand relevant investment channels, attract social capital for infrastructure construction, enhance their management levels for agricultural public infrastructure construction, and play an active role in increasing agricultural TFP.