Do Grain Imports Improve Water Use Efficiency in Grain Production? A Cost Competition Perspective

Abstract

Water scarcity poses a major constraint on efficient and sustainable grain production in China. Drawing on the New New Trade Theory and Induced Technological Innovation Theory, this study empirically investigates the relationship between cost competition from grain imports and water use efficiency in grain production from a virtual water trade perspective. The results show the following: (1) From 2003 to 2020, China’s overall grain production water use efficiency exhibited an upward trend, with the Huang-Huai-Hai and Northeast regions increasing by 66.35% and 28.49%, respectively. (2) Cost competition from grain imports can force improvements in water use efficiency. For every 1 billion tons of virtual water saved through imports, water use efficiency increases by 0.008. However, when annual virtual water savings exceed 11 billion tons, import competition surpasses a critical threshold. Due to technological and facility constraints, even previously efficient producers cannot further improve efficiency in the short term, and the allocative efficiency of production factors is undermined, leading to a decline rather than an improvement in water use efficiency. (3) The positive effect of import cost competition on water use efficiency is stronger in northern regions and non-major grain-producing areas. (4) Import cost competition improves water use efficiency by reducing domestic grain production profits. This study validates the applicability of the pro-competitive effect of trade and induced technological innovation in grain trade, expands the research boundaries of virtual water trade, and provides policy insights for improving China’s grain production water use efficiency.

1. Introduction

Food security is an eternal issue concerning national stability and people’s well-being, and ensuring the efficient and sustainable use of water resources is an important prerequisite for achieving food security [1]. China’s per capita water resources are only 26% of the world average, while water used for grain production has always dominated total water consumption. Water scarcity severely constrains China’s sustainable grain production [2]. A particularly striking feature of China’s resource endowment is the pronounced spatial disconnect between cultivated land and freshwater availability: the northern region holds 65% of the nation’s arable land while possessing only 18% of its total water resources [3]. This imbalance makes raising the water use efficiency of grain production a critical lever for securing sustainable food output and long-term food security [4]. Furthermore, the rigid increase in grain demand driven by economic and population growth, along with rising feed grain demand due to dietary structure upgrades, coupled with domestic constraints on water resource endowments and utilization efficiency, mean that relying solely on domestic grain production cannot meet consumption demand. Harnessing global grain resources and international markets has become an increasingly vital strategy for safeguarding China’s food security [5]. As trade theory has evolved, the New New Trade Theory highlights the pro-competitive effect of imports on the productive efficiency of importing nations [6]. While the New New Trade Theory was initially developed for heterogeneous manufacturing firms, its core mechanism—competition-induced efficiency gains—can also apply to agricultural settings where producers exhibit heterogeneity in scale, resource endowment, and productivity. In China, grain production is highly fragmented, but regional differences in farm size, irrigation efficiency, and yield demonstrate substantial heterogeneity. Competitive pressure from imported grains may disproportionately affect lower-efficiency or smaller farms, providing incentives for the adoption of more efficient practices or exit from unprofitable production. This adaptation justifies applying the pro-competitive effect of trade to smallholder-dominated agricultural sectors. Specifically, low-efficiency producers are driven out of the market as their market share erodes, while the surviving high-efficiency firms, facing persistent competitive pressure from imports, are compelled to further boost their efficiency and reduce production costs, thereby fostering the healthy upgrading of the industry [7]. Extending this logic, the Induced Technological Innovation Theory posits that the direction of agricultural technological progress is shaped by the relative scarcity of production factors; innovation tends to conserve the factors that are relatively scarce and make intensive use of those that are relatively abundant [8]. Therefore, in a context defined by constrained water supply and rising grain demand, a pivotal question arises: can grain imports effectively promote the intensive and efficient utilization of agricultural water resources? A rigorous scientific examination of this question can deepen our understanding of how grain imports influence the resource dimension of domestic grain production systems. The aim of this study is to empirically examine how cost competition from grain imports influences water use efficiency in grain production in China, using a virtual water trade perspective. It can help mitigate the spatial mismatch and shortage of land and water resources in China, cater to consumers’ increasingly diverse demand for grain products, and provide both theoretical underpinnings and policy insights for ensuring the nation’s food security.

Specifically, this study addresses the following research questions: Does cost competition from grain imports improve water use efficiency in China’s grain production? Are there regional differences in the impact of import competition on water use efficiency? Although grain imports are primarily an economic phenomenon, they can influence water use efficiency in domestic grain production through market-mediated mechanisms. For instance, competition from lower-cost imports can incentivize domestic producers to adopt more efficient water-saving practices, adjust input use, or restructure production to remain competitive. By linking economic pressures from trade with resource allocation decisions at the farm and regional level, it becomes meaningful to examine how import cost competition may affect water use efficiency. This study thus explores the interaction between economic and environmental dimensions, providing insights into how trade policies can indirectly contribute to sustainable water management. The remainder of the paper is organized as follows: Section 2, Materials and Methods, describes the data, variables, and empirical methodology; Section 3, Results, presents the empirical findings and mechanism analysis; and Section 4, Conclusions and Implications, discusses the policy relevance and summarizes the key contributions of the study.

Under the Ricardian framework and the assumption of perfect competition, traditional classical and neoclassical trade theories predominantly examine the welfare impacts of trade through the lens of technological gaps and factor endowments. At the beginning of this century, the emergence of the New New Trade Theory, pioneered by Melitz, redirected scholarly attention toward the welfare gains stemming from firm-level heterogeneity, among which the pro-competitive effect of trade became a central focus [7]. As a key component of total trade welfare, the pro-competitive effect was empirically isolated by Feenstra, who substituted VES (variable elasticity of substitution) preferences with translog preferences, thus allowing markups to respond to demand elasticity [6]. His estimates indicated that between 1992 and 2005, trade brought gains equivalent to 0.85% of U.S. GDP, with 0.44% arising from product variety and 0.41% from pro-competitive effects.

The concept of virtual water has since been extended from agricultural products to encompass all water consumed in the production of goods and services [9]. Scholars increasingly acknowledge that, alongside physical water, a vast amount of virtual water is embedded in socioeconomic activities and transferred through trade [10]. Virtual water trade aims to relieve pressure on water-scarce regions by importing water-intensive commodities, thereby realizing water savings. Since agricultural goods are typically water-intensive, virtual water trade is concentrated in agricultural exchanges [11]. Consequently, investigations into the welfare implications of grain imports have progressively incorporated a virtual water perspective. By sourcing water-intensive products from water-abundant countries, virtual water trade not only enables water-scarce nations to save water but also contributes to global water conservation [12]. Allan, for example, argued that virtual water trade can effectively assist the Middle East and Africa in achieving domestic water savings. Recognizing the water demands of grain production, many scholars have linked virtual water trade with grain trade, proposing that water-deficient countries can simultaneously secure grain supply and conserve water through grain imports [13]. Zhao et al. also observed that importing virtual water supplements water resources in a virtual form and mitigates China’s water scarcity stress [14]. Nonetheless, the existing literature has largely neglected the potential pro-competitive effect of grain imports on the resource use efficiency of the importing country [15].

Extensive research has been carried out on agricultural water use efficiency, centering on indicator selection and measurement methodologies as well as on influencing factors and empirical models. Regarding measurement, the two dominant approaches are the parametric method represented by Stochastic Frontier Analysis and the non-parametric method represented by Data Envelopment Analysis. Stochastic Frontier Analysis requires an explicit production function to determine the frontier and subsequently compute efficiency, whereas Data Envelopment Analysis identifies the frontier and evaluates the relative efficiency of decision-making units without imposing a functional form or assuming a distribution for the inefficiency term; it can also handle multiple outputs and is more readily extended [16]. With respect to influencing factors, studies have been conducted at both regional [17] and national [18] levels. The factors considered include natural conditions (e.g., precipitation, sunlight), general agricultural inputs (e.g., fertilizer application, cropping patterns), irrigation technology inputs (e.g., effectively irrigated area), water resource management indicators (e.g., water prices, administrative arrangements), and individual characteristics (e.g., education) [19,20]. Enhancing the water environment efficiency of agricultural production not only raises yields and quality but also curbs pollution and improves the ecological environment, aligning with the imperative of sustainable development [21]. Furthermore, scholars commonly employ multiple linear regression, obstacle factor analysis, spatial econometric models, Tobit models, and similar techniques to investigate the determinants of agricultural water use efficiency [22].

In sum, existing scholarship has thoroughly explored grain import trade welfare, virtual water trade, and water resource use efficiency, laying a firm theoretical and methodological foundation for this study. Nevertheless, important gaps remain. On the one hand, while prior work has sought to quantify the total welfare gains from grain imports and to trace all possible sources—such as the regulation of domestic supply and demand by imports or the enhancement of consumer surplus through greater product variety—few studies have investigated whether grain import trade can improve the resource allocation efficiency of domestic grain production [23]. Research on virtual water flows in grain trade partly addresses this gap, yet it primarily describes the virtual resource flows generated by imports, without empirically testing how the accompanying market shocks compel improvements in grain production efficiency in the importing country [10]. Even less attention has been paid to the regional heterogeneity in the effect of these virtual resource flows on water use efficiency. On the other hand, most existing studies use only grain output or output value as the output variable, overlooking undesirable outputs such as agricultural non-point source pollution (e.g., grey water). This omission tends to overestimate the measured water use efficiency in grain production. The present study addresses these shortcomings by more comprehensively incorporating pollution emissions and ecological outputs, thereby better reflecting real-world grain production and enhancing measurement accuracy. To this end, the study employs the DEA-GML index (Data Envelopment Analysis—Global Malmquist-Luenberger index), treating the ecological value of grain cultivation as a desirable output and agricultural non-point source pollution as an undesirable output, to measure water use efficiency in grain production. Drawing on the New New Trade Theory and the Induced Technological Innovation Theory, It empirically examines the impact of grain import cost competition on water use efficiency and its regional heterogeneity from a virtual water trade perspective. In doing so, this study innovatively verifies the applicability of the pro-competitive effect of trade and the induced technological innovation effect in the domain of grain trade, extends the boundaries of virtual water trade research, and offers new insights into the mechanism through which grain imports force improvements in domestic grain production.

2. Materials and Methods

2.1. Measurement Method of Grain Import Cost Competition

Adopting a water demand perspective for grain production, this study quantifies the virtual water content embedded in various domestically produced grain crops using the water footprint methodology. As noted by Chapagain et al. [11] while the concepts of virtual water and water footprint share similar connotations, the water footprint has a broader scope of application; however, for quantifying crop water requirements, the two notions are equivalent. Accordingly, virtual water imports are primarily assessed through the water footprint approach [24]. The crop water footprint captures the total freshwater consumed during the crop growth cycle and is disaggregated into blue and green water footprints according to water source. The CROPWAT model, as recommended by the Natural Resources Conservation Service (NRCS), is applied in this study. By incorporating daily climate records from provincial meteorological stations, along with altitude, latitude, crop coefficients, and sowing/harvesting dates for different crops and regions, the blue and green water footprints of each crop in each province are derived. The detailed computational procedures are presented as follows:

$$WF{P}_{itk}=WF{P}_{itk,g}+WF{P}_{itk,b}=10{\displaystyle \frac{E{T}_{itk,g}}{Y_{itk}}}+10{\displaystyle \frac{E{T}_{itk,b}}{Y_{itk}}}$$

(1)

$$E{T}_{itk,g}=\min (E{T}_{itk,c},P_{itk,e})$$

(2)

$$E{T}_{itk,b}=\max (E{T}_{itk,c}-P_{itk,e},0)$$

(3)

In Equations (1)–(3), WFP_itk denotes the water footprint per unit mass of grain crop k in province i in year t (m³/kg); WFP_itk,g and WFP_itk,b represent the green water footprint and blue water footprint of grain crop k in province i in year t (m³/kg), respectively; ET_itk,g is the green water evapotranspiration of grain crop k in province i in year t (mm), and ET_itk,b is the corresponding blue water evapotranspiration (mm); Y_itk stands for the yield per unit area of grain crop k in province i in year t (kg∙hm⁻²); 10 is the unit conversion factor; ET_itk,c designates the evapotranspiration of grain crop k in province i in year t (mm), estimated via the Penman-Monteith method advocated by the Food and Agriculture Organization (FAO) of the United Nations; P_itk,e is the effective precipitation during the growing season of grain crop k in province i in year t (mm). The subsequent formulas are as follows:

$$SW{D}_{ltk}=SW{D}_{ltk,2000}\times {\displaystyle \frac{Y_{ltk,2000}}{Y_{ltk}}}$$

(4)

In Equation (4), SWD_ltk represents the water footprint per unit mass of grain crop k in country l in year t (m³/kg); SWD_ltk,2000 is the corresponding water footprint in the base year 2000 (m³/kg); Y_ltk,2000 is the yield per unit area of grain crop k in country l in 2000 (kg); Y_ltk is the yield per unit area of grain crop k in country l in year t (kg).

$$VW{I}_{tk}=SW{D}_{ltk}\times {\displaystyle \frac{I{M}_{ltk}}{{\displaystyle \sum _{l=1}^{n}I{M}_{ltk}}}}$$

(5)

In Equation (5), VWI_tk indicates the average water footprint of imported grain crop k in year t (m³/kg); IM_ltk is the quantity of grain crop k imported from country l in year t (kg); and n denotes the total number of grain-exporting source countries.

$$FIW{S}_{itk}=I{M}_{itk}\times (WF{P}_{itk}-VW{I}_{tk})$$

(6)

In Equation (5), FIWS_itk measures the volume of water saved through imports of grain crop k in province i in year t (m³); IM_itk refers to the quantity of grain crop k brought into province i in year t (kg). The net import water savings of grain crops for province i simultaneously capture two aspects: the amount of global water resources conserved through net grain imports, and the differential between the water cost of domestic grain production and that of grain production in net-importing countries. Viewed through the lens of resource use efficiency, this variable captures the market shock exerted by net grain imports on the domestic grain sector—that is, the grain import cost competition engendered by these imports.

2.2. Measurement Method of Water Use Efficiency in Grain Production

Given that the GML index is well-suited to simultaneously pursue the green development objectives of maximizing desirable outputs while minimizing undesirable outputs and input factors, this study constructs a GML index model, following Oh [25], to evaluate the dynamics of grain production water use efficiency for 30 provincial-level administrative regions in China over the period 2003 to 2020. The detailed computational specification is as follows:

$$\begin{array}{c}GM{L}^{t,t+1}(x^t,y^t,b^t,x^{t+1},y^{t+1},b^{t+1})={\displaystyle \frac{1+D^g(x^t,y^t,b^t)}{1+D^g(x^{t+1},y^{t+1},b^{t+1})}}={\displaystyle \frac{1+D^g(x^t,y^t,b^t)}{1+D_t(x^t,y^t,b^t)}}\times \\ {\displaystyle \frac{1+D_{t+1}(X^{t+1},Y^{t+1},B^{T+1})}{1+D^g(x^{t+1},y^{t+1},b^{t+1})}}\times {\displaystyle \frac{1+D_t(x^t,y^t,b^t)}{1+D_{t+1}(x^{t+1},y^{t+1},b^{t+1})}}=\\ {\displaystyle \frac{1+D_t(x^t,y^t,b^t)}{1+D_{t+1}(x^{t+1},y^{t+1},b^{t+1})}}\times \left[{\displaystyle \frac{\frac{1+D^g(x^t,y^t,b^t)}{1+D_t(x^t,y^t,b^t)}}{\frac{1+D_t^g(x^{t+1},y^{t+1},b^{t+1})}{1+D_{t+1}(x^{t+1},y^{t+1},b^{t+1})}}}\right]=\\ {\displaystyle \frac{T{E}^{t+1}}{T{E}^t}}\times {\displaystyle \frac{BP{G}_{t+1}^{t,t+1}}{BP{G}_t^{t,t+1}}}=GE{C}^{t,t+1}\times GT{C}^{t,t+1}\end{array}$$

(7)

In Equation (7), the vectors x^t, y^t, and b^t respectively capture the input factors, desirable outputs, and undesirable outputs in year t; x^t+1, y^t+1, and b^t+1 are defined analogously for year t + 1. D_t(x^t, y^t, b^t) and D_t+1(x^t+1, y^t+1, b^t+1) denote the output distance functions for the input-output bundles (x, y, b) in periods t and t + 1, while D_g(x^t, y^t, b^t) represents the directional vector of the reference set. TE^t and TE^t+1 measure the comprehensive technical efficiency in years t and t + 1; BPG^t and BPG^t+1 indicate the distance between the technology reference set and the production frontier in those years. The GML index from period t to t + 1 is further decomposed into two components: GEC^t,t+1, which captures technical efficiency change, and GTC^t,t+1, which captures technological progress in grain production water use. Values of GML, GEC, or GTC exceeding 1 imply an improvement in water use efficiency, technical efficiency, or technological progress from t to t + 1; values below 1 indicate a decline; and values exactly equal to 1 signify no change. To obtain a cumulative series, 1993 is taken as the base year with its grain production water use efficiency normalized to 1, and the efficiency values of subsequent years are cumulatively multiplied.

In measuring grain production water use efficiency, the input factors considered are water, labor, fertilizer, pesticide, machinery, and plastic film used in grain production. The desirable outputs consist of grain output and the ecological value of grain cultivation, while the undesirable outputs include agricultural non-point source pollution and carbon emissions [26]. The specific indicators and their measurement methods are summarized in

Table 1. Measurement Indicators of ecological total factor productivity of grain.

Indicators Category	Indicators	Calculation Method	Unit
Expected output	Ecological value of grain cultivation	Measured by ESV method [27]	None
	Grain production	Statistics	×104 t
Undesired output	Carbon Emissions	Agricultural carbon emissions × A
	Non-point source pollution	Agricultural non-point source pollution	×104 t
Input element	Water input	Amount of water used in agriculture × A	×108 m3
	Labor input	Number of employees in primary industry × B	×104 persons
	Machinery input	Total power of agricultural machinery × A	×104 kW
	Pesticide input	Amount of pesticides used × A	×104 t
	Fertilizer input	Amount of agricultural fertilizer applied × A	×104 t
	Plastic film input	Amount of agricultural plastic film used × A	×104 t

A = grain sown area/total crop sown area; B = (A × (agricultural output value/total agricultural, forestry, animal husbandry and fishery output value)).

.

2.3. Selection of Variables

For the model specification, appropriate dependent, independent, instrumental, and control variables are chosen, with their corresponding indicators constructed based on the extant literature. The definitions and measurement procedures for all variables are summarized in

Table 2. Variables and Calculation Methods.

Variable Category	Variable	Calculation Method	Unit
Dependent Variable	Grain Production Water Use Efficiency	Measured using the Global-Malmquist-Luenberger index	None
Core Explanatory Variable	Grain Import Cost Competition	(Domestic grain virtual water content—Imported grain virtual water content) × Grain import volume	×109 t
Instrumental Variable	Livestock Scale	Milk production + Beef production + Poultry production + Mutton production + Pork production	×108 t
Control Variables	Technological Environment	Technology market transaction value/Internal R&D expenditure	None
	Water Use Structure	(Total industrial water use + Total domestic water use)/Total water use	None
	Irrigation Ratio	Effective irrigated area/Total sown area of crops	None
	Financial Support for Agriculture	Government expenditure on agriculture, forestry, and water affairs	×1010 RMB
	Disaster Rate	Affected crop area/Total sown crop area	None
	Agricultural Mechanization Level	Total agricultural machinery power/Number of primary industry employees	kW/person
	Specialization Degree of Crop Planting	Herfindahl index of wheat, rice, corn, beans, and tubers	None
	Environmental Regulation	Investment in environmental pollution control as a percentage of GDP	%

.

Dependent variable: water use efficiency in grain production. Drawing on Chung et al., this study adopts the GML index method with selected input-output indicators [28]. Following Zhou et al. [29], undesirable outputs are incorporated, and the ecological value of grain cultivation is treated as a desirable output. This framework seeks to maximize the combined economic and ecological value generated by grain production while minimizing agricultural non-point source pollution, carbon emissions, and factor inputs, thereby capturing grain production water use efficiency in a more rigorous and realistic manner.

Core explanatory variable: grain import cost competition. Because the water footprint of a given grain product differs across countries or regions, the virtual water content per unit of output also varies. According to the pro-competitive effect of trade, nations that produce grain with lower virtual water content enjoy lower production costs and prices. When a country imports grain from such lower-cost sources, the market share of domestic producers is compressed and domestic grain prices are depressed, thereby delivering a shock to the importing country’s grain market. Accordingly, this study operationalizes “grain import cost competition” as the difference between the virtual water content of China’s domestically produced grain and that of imported grain, multiplied by the volume of grain imports. From a water resource perspective, this measure captures the intensity of the market shock exerted by grain imports on domestic grain production. While virtual water content is an environmental metric rather than a direct cost measure, it serves as a proxy for relative production efficiency across countries. Imported grains with lower virtual water content generally correspond to lower input requirements per unit of output, which can exert competitive pressure on domestic producers.

Instrumental variable: livestock production scale. The sum of China’s output of milk, beef, poultry meat, mutton, and pork is used to represent “livestock production scale” and serves as the instrumental variable. The rationale is that the extent of grain import cost competition hinges on the degree of shock that grain imports impart to domestic grain production. Presently, the average virtual water content of imported soybeans and corn is substantially lower than that of their domestically produced counterparts, generating a pronounced shock to domestically grown soybeans and corn—both of which exhibit relatively high virtual water content. The primary domestic consumer of imported feed grains (soybeans and corn) is the livestock sector. Hence, conceptually, livestock production scale is correlated with the endogenous variable grain import cost competition, while remaining unrelated to the contemporaneous error term or to grain production water use efficiency. For these reasons, livestock production scale is selected as the instrument.

Control variables: Informed by prior research on the determinants of grain production water use efficiency [30], eight control variables are incorporated into the model: technological environment, water use structure, irrigation ratio, fiscal support for agriculture, disaster rate, agricultural mechanization level, specialization degree of crop planting, and environmental regulation. We include livestock production scale and financial support for agriculture as control variables because they may indirectly affect water use efficiency in grain production. Livestock production can influence land allocation, input use, and crop choice, which in turn affect water demand and efficiency. Financial support for agriculture can alter production incentives, encourage technology adoption, and support water-saving practices. Including these variables helps to isolate the effect of import cost competition on water use efficiency while accounting for other factors that may influence production efficiency.

2.4. Empirical Model Design of the Impact of Grain Import Cost Competition on Water Use Efficiency in Grain Production

We hypothesize that higher import cost competition is associated with higher water use efficiency in domestic grain production, and that this effect may vary across regions and production scales.

Given that the dependent variable—water use efficiency in grain production (Wefp), measured by the GML index—is continuous but strictly greater than zero, a Tobit regression model is adopted to account for the non-negativity of the dependent variable in parameter estimation, as specified in Equation (8).

$$Wefp=\left\{\begin{array}{ll}Wef{p^{\ast }}_{it}=\sigma +{\alpha }_{1}Fiw{c}_{it}+{\alpha }_2{X}_{z,it}+{\mu }_i+{\phi }_t+{\epsilon }_{it}\hfill & Wef{p^{\ast }}_{it}>0\\ 0\hfill & Wef{p^{\ast }}_{it}\le 0\end{array}\right.$$

(8)

In Equation (8), Wefp^*_it denotes the water use efficiency in grain production for region i in year t; Fiwc_it serves as the core explanatory variable, measuring the grain import cost competition confronting region i in year t; and X_z,it represents the set of control variables that also affect grain production water use efficiency, with the subscript z ranging from 1 to 8, corresponding respectively to the eight control variables: technological environment, water use structure, irrigation ratio, fiscal support for agriculture, disaster rate, agricultural mechanization level, specialization degree of crop planting, and environmental regulation. The parameter σ is the constant term, α denotes the coefficient associated with each explanatory variable, μ_i captures unobservable province-specific effects, φ_t accounts for time trend fixed effects, and ε_it is the random disturbance term. Equation (8) constitutes the benchmark model of this study and is employed to examine the linear relationship between grain import cost competition and water use efficiency in grain production.

$$Wefp=\left\{\begin{array}{ll}Wef{p^{\ast }}_{it}=\sigma +{\alpha }_{1}Fiw{c}_{i(t-1)}+{\alpha }_2{X}_{z,it}+{\mu }_i+{\phi }_t+{\epsilon }_{it}\hfill & Wef{p^{\ast }}_{it}>0\\ 0\hfill & Wef{p^{\ast }}_{it}\le 0\end{array}\right.$$

(9)

To further account for the possible time lags and dynamic feedbacks inherent in grain import cost competition, market adjustments, and changes in water use efficiency in grain production, Equation (9) incorporates the one-period lagged term of grain import cost competition, Fiwc_i(t−1), in order to investigate the lagged effect of grain import cost competition on grain production water use efficiency.

$$Wefp=\left\{\begin{array}{ll}Wef{p^{\ast }}_{it}=\sigma +{\alpha }_{1}Fiw{c}_{it}+{\alpha }_{2}Fiw{c^2}_{it}+{\alpha }_3{X}_{z,it}+{\mu }_i+{\phi }_t+{\epsilon }_{it}\hfill & Wef{p^{\ast }}_{it}>0\\ 0\hfill & Wef{p^{\ast }}_{it}\le 0\end{array}\right.$$

(10)

Equation (10) extends the specification by including the quadratic term of grain import cost competition, thereby allowing for a test of whether the relationship between grain import cost competition and water use efficiency in grain production is non-linear.

2.5. Data Sources and Data Processing

This study utilizes a panel dataset of 30 provincial-level administrative regions in China from 2003 to 2020, excluding Hong Kong, Macao, Taiwan, and Tibet due to data unavailability. Trade data for grain products are obtained from the General Administration of Customs of the People’s Republic of China, the Food and Agriculture Organization (FAO) of the United Nations database, and the United Nations Commodity Trade Statistics Database (UN Comtrade). Other statistical information is compiled from various official yearbooks and bulletins.

3. Results

3.1. Analysis of Grain Import Cost Competition Measurement Results

Between 2003 and 2020, the water footprint of grain production in China varied considerably across crop types and regions. Soybean exhibited the highest water footprint, with a national average of 2044.595 m³/t, while maize registered the lowest at 655.713 m³/t, roughly one-third of the soybean figure; wheat and paddy rice fell in between. Regionally, South China and Southwest China recorded relatively high water footprints for all grain crops, whereas the Northeast and Northwest regions showed relatively low values; the Huang-Huai-Hai and Middle-Lower Yangtze River regions occupied intermediate positions. These spatial disparities are primarily attributable to differences in crop coefficients, yields per unit area, and growing-season precipitation. Over the study period, the national average virtual water content of the four major grain crops followed a steady downward trajectory, driven by continuous improvements in irrigation technology and progressive gains in grain production efficiency. Specifically, the virtual water content of wheat, paddy rice, and maize declined consistently, while that of soybean exhibited a fluctuating but overall decreasing pattern.

The net water-saving effect of grain imports also evolved markedly across China’s regions during this period. At the national level, the average regional net water savings from grain imports leaped from 336.2 million tons in 2003 to 2809.8 million tons in 2020, albeit with distinct phases: a relatively stable fluctuation before 2011, a rapid surge from 2011 to 2017, a sharp drop of 25.68% in 2018, and a subsequent gradual recovery. Pronounced regional heterogeneity was evident: South China registered the largest and most volatile net water-saving effect, the Huang-Huai-Hai region experienced a steady upward trend turning from negative to positive, the Northeast and Middle-Lower Yangtze River regions exhibited moderate levels and upward trajectories (interrupted by sharp declines in 2018–2019 in the latter), and the Southwest and Northwest regions, although starting from low bases, maintained steady increases.

3.2. Analysis of Water Use Efficiency in Grain Production Results

An examination of China’s grain production water use efficiency over the period 2003–2020 reveals that the majority of observations fall within the 0–1 range, with the kernel density curve displaying a concentrated peak of high intensity but relatively narrow span, suggesting limited overall dispersion in water use efficiency levels (Figure 1). As time progresses, however, the concentration weakens, the degree of differentiation among regions rises, and the peak shifts rightward. This pattern points to a general upward trend in national grain production water use efficiency, accompanied by faster improvements in certain regions. These gains are attributable both to the strengthening of government ecological conservation measures and heightened awareness of water saving, as well as to the progressive refinement of water-saving irrigation practices and rapid advances in irrigation technology.

Pronounced regional disparities in grain production water use efficiency emerge under the influence of climatic conditions, socioeconomic development, and water resource endowments (Figure 2). At the national average level, water use efficiency fluctuated with a mild downward tendency from 2003 to 2011, then shifted to a fluctuating upward path from 2012 to 2020. The Huang-Huai-Hai and Northeast regions stood out with relatively high and fast-rising water use efficiency, recording increases of 66.35% and 28.49%, respectively, between 2003 and 2020. The Northwest and Southwest regions occupied an intermediate position, exhibiting a decline followed by a rapid rebound; specifically, Northwest water use efficiency fell by 22.38% from 2003 to 2011 before climbing by 45.27% from 2012 to 2020, while Southwest water use efficiency rose by 16.78% over the entire period. The Middle-Lower Yangtze River and South China regions displayed relatively low water use efficiency and a pattern of initial steep decline followed by slow recovery: the former dropped by 18.68% from 2003 to 2013 and then grew by 32.98% from 2014 to 2020, whereas the latter contracted by 30.97% from 2003 to 2015 and regained 18.57% from 2015 to 2020. Overall, the spatial distribution indicates that regions endowed with scarcer water resources are driven more forcefully to improve grain production water use efficiency, whereas those with more abundant water resources face weaker incentives to enhance efficiency.

3.3. Empirical Analysis of the Impact of Grain Import Cost Competition on Water Use Efficiency in Grain Production

3.3.1. Benchmark Regression

The estimation results for the effect of grain import cost competition on water use efficiency are displayed in

Table 3. Benchmark regression results of the impact of grain import cost competition on water use efficiency in grain production.

Variable	Grain Production Water Use Efficiency
	Model 1	Model 2	Model 3	Model 4	Model 5
Grain Import Cost Competition		0.008 ***	0.198 ***		0.008 **
		(3.019)	(3.571)		(2.188)
Grain Import Cost Competition Lag 1				0.007 ***
				(2.687)
Technological Environment	0.030 **	0.031 **	0.057	0.030 **	0.031 **
	(2.352)	(2.467)	(1.289)	(2.239)	(2.037)
Water Use Structure	0.005	−0.004	−0.201	0.011	−0.004
	(0.096)	(−0.077)	(−1.121)	(0.223)	(−0.113)
Irrigation Ratio	0.371 ***	0.363 ***	0.225	0.364 ***	0.363 *
	(3.765)	(3.699)	(0.581)	(3.612)	(1.744)
Financial Support for Agriculture	0.013 ***	0.011 **	−0.036 *	0.010 **	0.011 *
	(2.835)	(2.403)	(−1.701)	(2.220)	(1.775)
Disaster Rate	−0.156 ***	−0.163 ***	−0.345 *	−0.136 **	−0.163 ***
	(−2.873)	(−3.028)	(−1.792)	(−2.480)	(−3.279)
Agricultural Mechanization Level	0.003	0.002	−0.019	0.002	0.002
	(0.851)	(0.559)	(−1.612)	(0.542)	(0.199)
Specialization Degree of Crop Planting	0.084	0.083	0.094	0.080	0.083
	(1.008)	(1.014)	(0.316)	(0.966)	(0.637)
Environmental Regulation	−0.002	0.005	0.187 ***	0.004	0.005
	(−0.191)	(0.415)	(2.740)	(0.311)	(0.423)
Time Fixed Effects	Yes	Yes	Yes	Yes	Yes
Individual Fixed Effects	Yes	Yes	Yes	Yes	Yes
First-Stage F Statistic			11.02
Wald Test Statistic			195.18 ***
Constant	0.848 ***	0.848 ***	0.827 ***	0.854 ***	0.848 ***
	(12.263)	(12.269)	(3.598)	(12.196)	(7.449)
N	540	540	540	510	540

***, **, and * denote significant at the p < 1%, p < 5%, and p < 10% levels; the numbers in parentheses are t-values.

(based on Equation (8)). Column (1) provides the baseline regression with control variables only. Turning to Column (2), the coefficient on grain import cost competition is positive and statistically significant at the 1% level (coefficient = 0.008), suggesting that a one-unit rise in import cost competition is associated with a 0.008-unit improvement in water use efficiency. For grain producers, who operate under thin profit margins, production efficiency is a crucial determinant of their survival and development, particularly given that water resources impose a tight constraint on grain production. The virtual water content of China’s major imported grain products (e.g., soybeans and corn) lies below the domestic production average. Under the pro-competitive effect of trade, rising grain imports tend to favor large-scale, higher-water use efficiency producers who enjoy lower costs, while small-scale, lower-water use efficiency producers with higher costs face market exit [31]. While direct farm-level exit data are unavailable, previous studies suggest that competitive import shocks can reduce the profitability of smaller, lower-efficiency producers, leading to a reallocation of production towards more efficient farms [30]. In our analysis, provincial-level water use efficiency improvements serve as an aggregate indicator of such reallocation, acknowledging that the micro-level mechanisms are inferred rather than directly measured. International differences in water use efficiency thus manifest as differences in virtual water content. Through pro-competitive forces, growing imports erode the market share of domestic producers and compel the domestic grain sector to pursue a higher level of water resource use efficiency. More concretely, import cost competition reduces water input per unit of grain output via economies of scale and fosters both horizontal and vertical division of labor through the outsourcing of productive services, leading to more precise factor application. Over time, increased specialization allows producers to accumulate crop-specific management experience and adopt more scientific field practices, while concentrated grain cultivation spurs the generation of new technologies and accelerates technological progress. These shifts enhance water resource use efficiency, conserve water and other inputs that would otherwise be wasted, and cut pollutant emissions. Furthermore, consistent with the Induced Technological Innovation Theory, China’s per capita water endowment is merely 25% of the global average, rendering water a relatively scarce factor. This scarcity induces technological change in grain production to move in a water-saving direction, thereby promoting water use efficiency gains [32]. While we do not directly measure technological innovation at the farm level, improvements in water use efficiency may reflect the adoption of more efficient production practices consistent with the direction predicted by Induced Technological Innovation Theory.

To address potential endogeneity, an instrumental variable approach is adopted. Because the extent of grain import cost competition depends on how strongly grain imports shock domestic production, and given that the average virtual water content of imported soybeans and corn is markedly lower than that of domestic output—imparting a sizeable shock to domestic soybeans and corn—the livestock sector, as the largest consumer of imported feed grains, is theoretically associated with import cost competition but not directly with the error term or water use efficiency. Hence, “livestock production scale” is chosen as the instrument. First-stage diagnostic results confirm the instrument’s validity: the F-statistic is 11.02, exceeding the critical value of 10 for weak instruments, and the Wald endogeneity test rejects exogeneity at the 1% level, confirming that grain import cost competition is indeed endogenous. The two-stage least squares estimates indicate that the impact of import cost competition on water use efficiency remains significant at the 5% level with a coefficient of 0.198, reinforcing the conclusion that import cost competition effectively improves water use efficiency.

Recognizing that the positive effect of import cost competition on water use efficiency may take time to materialize, the first-order lag of import cost competition is added to the model (Equation (9)). As reported in Column (4), the lagged term is significant at the 5% level with a coefficient of 0.007, implying that a one-unit increase in current import cost competition is associated with a 0.007-unit decrease in the following year’s water use efficiency, highlighting the existence of a temporal lag in this relationship.

Because the accuracy of parametric inference hinges on correct model specification, and it is difficult to ascertain solely from theory whether the parametric form is appropriate, a nonparametric Bootstrap estimation (1000 replications) is performed as a robustness check. The results, presented in Column (5) of Table 3, confirm that grain import cost competition remains significant at the 1% level with a coefficient of 0.008, corroborating the baseline finding that import cost competition indeed strengthens water use efficiency. The overall consistency of the estimates points to the robustness of the benchmark regression results.

Furthermore, as shown in Column (6), the quadratic term of grain import cost competition has a significant effect on water use efficiency at the p < 1% level, with a coefficient of −0.002, indicating an inverted U-shaped relationship between grain import cost competition and water use efficiency (based on Equation (10)). There exists an optimal value for the cost competition induced by grain imports between domestic and foreign markets in grain production. Initially, the competitive effect of grain imports on the domestic market remains within an acceptable range, and the market eliminates grain producers with lower water use efficiency through self-adjustment. However, when the annual volume of virtual water saved through grain imports in a region exceeds 11 billion tons, the import shock surpasses a certain threshold, causing the market’s self-adjustment mechanism to fail. Due to constraints in production facilities and technology, the remaining producers cannot further improve water use efficiency to reduce production costs within a short period. At this point, even grain producers with relatively high water use efficiency face the risk of being squeezed out of the market, leading to a decline in the allocative efficiency of grain production resources. Thus, excessively strong competitive effects can severely harm the domestic grain industry, resulting in a situation where water use efficiency does not increase but instead decreases.

3.3.2. Heterogeneity Analysis of the Impact of Grain Import Cost Competition on Water Use Efficiency

To further explore regional heterogeneity, the sample is split into northern and southern subgroups (

Table 4. Heterogeneity of the Impact of Grain Import Cost Competition on Grain Production Water Use Efficiency in Northern and Southern Regions.

Variable	Grain Production Water Use Efficiency
	Northern Region				Southern Region
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
Grain Import Cost Competition		0.037 ***	0.116 ***			−0.00003	−0.110
		(5.930)	(5.941)			(−0.020)	(−0.568)
Grain Import Cost Competition Lag 1				0.046 ***				−0.003
				(7.135)				(−1.425)
Technological Environment	−0.007	0.002	0.020	−0.000	−0.034 *	−0.034 *	−0.062	−0.022
	(−0.471)	(0.169)	(1.002)	(−0.018)	(−1.789)	(−1.788)	(−0.672)	(−1.114)
Water Use Structure	0.032	−0.021	−0.147	−0.024	0.055	0.055	0.196	0.055
	(0.390)	(−0.273)	(−1.321)	(−0.313)	(1.471)	(1.470)	(0.662)	(1.571)
Irrigation Ratio	0.883 ***	0.824 ***	0.789 ***	0.794 ***	−0.212 *	−0.212 *	−0.267	−0.238 **
	(7.126)	(6.954)	(4.578)	(6.783)	(−1.813)	(−1.813)	(−0.503)	(−2.043)
Financial Support for Agriculture	0.040 ***	0.032 ***	0.014	0.030 ***	0.018 ***	0.018 ***	0.055	0.017 ***
	(5.221)	(4.459)	(1.355)	(4.172)	(4.502)	(4.452)	(0.816)	(4.250)
Disaster Rate	−0.135 *	−0.204 ***	−0.356 ***	−0.137 *	−0.133 **	−0.133 **	−0.047	−0.132 ***
	(−1.817)	(−2.889)	(−3.527)	(−1.888)	(−2.534)	(−2.533)	(−0.175)	(−2.630)
Agricultural Mechanization Level	−0.049 ***	−0.045 ***	−0.032 ***	−0.041 ***	0.010 ***	0.010 ***	0.026	0.009 ***
	(−6.064)	(−5.781)	(−2.814)	(−5.280)	(4.775)	(4.730)	(0.846)	(4.875)
Specialization Degree of Crop Planting	0.328	0.178	−0.102	0.206	−0.019	−0.019	0.027	0.001
	(1.468)	(0.826)	(−0.300)	(0.902)	(−0.322)	(−0.322)	(0.113)	(0.022)
Environmental Regulation	−0.076 ***	−0.036 **	0.054 *	−0.032 **	0.009	0.009	−0.096	0.010
	(−4.861)	(−2.205)	(1.846)	(−1.978)	(0.603)	(0.597)	(−0.490)	(0.685)
Time Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Individual Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
First-Stage F Statistic			7.32				10.45
Wald Test Statistic			64.14 ***				5.04 **
	0.769 ***	0.824 ***	0.880 ***	0.838 ***	1.090 ***	1.090 ***	1.149 ***	1.068 ***
Constant	(6.346)	(7.041)	(5.201)	(6.932)	(16.662)	(16.661)	(4.310)	(16.502)
N	270	270	270	255	270	270	270	255

***, **, and * denote significant at the p < 1%, p < 5%, and p < 10% levels; the numbers in parentheses are t-values.

). For the north, Columns (1)–(4) present the estimation results, with Column (1) showing the baseline control-variable regression. Regardless of whether the benchmark specification, the instrumental variable approach, or the first-order lag of grain import cost competition is employed, the coefficient on import cost competition remains consistently positive and statistically significant across Columns (2)–(4), in line with the full-sample findings. For the south, results in Columns (5)–(8) display a different pattern: across all three model forms, the effect of import cost competition on water use efficiency is not statistically significant, departing from the full-sample results.

The pro-competitive effect generated by grain import cost competition on water use efficiency proves markedly stronger in the northern region than in the southern region. On the one hand, this is because livestock production is more developed in the north, generating substantially higher demand for feed grains, especially soybeans and corn [33]. As imports of these crops continue to expand, a large volume of agricultural commodities with virtual water content below that of domestic output flows into the northern market, delivering a pronounced shock and intensifying the pro-competitive effect of trade. Via scale expansion, this competitive pressure reduces the input of marginal grain production factors in the north, while simultaneously raising the degree of specialization in domestic grain cultivation. Over time, deepening specialization allows producers to accumulate crop-specific planting and management expertise, leading to more scientific field practices. This improves water resource use efficiency in grain production and saves water that would otherwise be lost through adjustments in production activities [34]. On the other hand, water resource endowments are lower in the north than in the south, yet grain production requires more water. Combined with the severe spatial mismatch between arable land and water, water use pressure in northern grain production remains persistently high [35]. The relative water scarcity thus strengthens the induced technological innovation effect, accelerating technological progress in grain production toward water conservation [36]. Under the impetus of induced innovation, grain production technologies evolve in a water-saving direction, further enhancing the allocative efficiency of water resources in the north.

Additionally, grain production characteristics may diverge substantially between major grain-producing areas and non-major grain-producing areas. To account for this, the sample is further divided according to this regional attribute (

Table 5. Heterogeneity of the Impact of Grain Import Cost Competition on Grain Production Water Use Efficiency in Different Production Regions.

Variable	Grain Production Water Use Efficiency
	Main Grain Producing Areas				Non-Grain Main Producing Areas
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
Grain Import Cost Competition		−0.001	−0.122			0.011 ***	0.157 ***
		(−0.409)	(−0.975)			(2.687)	(4.513)
Grain Import Cost Competition Lag 1				−0.001				0.010 **
				(−0.386)				(2.421)
Technological Environment	−0.003	−0.004	−0.114	−0.002	0.039 **	0.039 **	0.029	0.042 **
	(−0.144)	(−0.209)	(−0.865)	(−0.083)	(2.364)	(2.362)	(0.698)	(2.369)
Water Use Structure	0.201	0.237	3.685	0.195	0.014	0.011	−0.024	0.025
	(0.960)	(1.043)	(1.070)	(0.880)	(0.251)	(0.189)	(−0.166)	(0.437)
Irrigation Ratio	0.286	0.291	0.966	0.155	0.461 ***	0.454 ***	0.159	0.491 ***
	(1.215)	(1.237)	(1.245)	(0.674)	(3.889)	(3.828)	(0.435)	(4.004)
Financial Support for Agriculture	0.006	0.006	0.026	0.007	−0.000	−0.002	−0.024	−0.003
	(1.099)	(1.131)	(1.168)	(1.318)	(−0.043)	(−0.277)	(−1.090)	(−0.408)
Disaster Rate	−0.296 ***	−0.297 ***	−0.390 *	−0.242 ***	−0.090	−0.089	−0.087	−0.087
	(−4.279)	(−4.291)	(−1.850)	(−3.687)	(−1.128)	(−1.129)	(−0.446)	(−1.075)
Agricultural Mechanization Level	0.008	0.009	0.114	0.003	0.001	0.000	−0.005	0.000
	(0.952)	(1.035)	(1.003)	(0.380)	(0.199)	(0.106)	(−0.560)	(0.115)
Specialization Degree of Crop Planting	0.211	0.218	1.157	0.449	0.023	0.020	0.066	0.007
	(0.796)	(0.820)	(1.377)	(1.612)	(0.227)	(0.198)	(0.263)	(0.073)
Environmental Regulation	−0.046 **	−0.049 **	−0.334	−0.032 *	0.018	0.024	0.092 **	0.020
	(−2.521)	(−2.478)	(−1.092)	(−1.683)	(1.112)	(1.463)	(2.059)	(1.160)
Time Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Individual Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
First-Stage F Statistic			13.45				7.21
Wald Test Statistic			4313.10 ***				2275.32 ***
Constant	0.887 ***	0.872 ***	−0.778	0.808 ***	0.778 ***	0.774 ***	0.765 ***	0.781 ***
	(3.965)	(3.834)	(−0.539)	(3.467)	(9.412)	(9.324)	(3.723)	(9.147)
N	234	234	234	221	306	306	306	289

***, **, and * denote significant at the p < 1%, p < 5%, and p < 10% levels; the numbers in parentheses are t-values.

). In major grain-producing areas, results in Columns (1)–(4)—with Column (1) showing the control-variable-only regression—reveal that, irrespective of whether the baseline, instrumental-variable, or lagged-term specification is used, grain import cost competition exhibits no significant effect on water use efficiency, contrasting with the full-sample pattern. In non-major grain-producing areas, however, estimates in Columns (5)–(8) consistently indicate a significant positive impact across all three model forms, closely resembling the full-sample results.

The water use efficiency-enhancing effect of grain import cost competition is noticeably stronger in non-major grain-producing areas than in major ones. This is because the scale of grain operations in major areas far exceeds that in non-major areas, so the marginal pro-competitive impetus from grain imports on large-scale cultivation is comparatively weak [37]. Moreover, water use efficiency in major areas is already much higher, meaning that the induced technological innovation effect generated by grain imports contributes less at the margin to efficiency gains in these regions. In non-major areas, by contrast, grain production is conducted on a smaller scale and at lower efficiency levels, so the influx of imported grain with low virtual water content inflicts a more intense market shock than in major areas [36]. This shock effectively eliminates smaller, less efficient producers, thereby accelerating the shift toward larger-scale, more efficient production. Scaling up cultivation, in turn, fosters the outsourcing of productive services, deepens both horizontal and vertical division of labor in grain production, enables more precise factor application, and reduces redundant water resource inputs [34]. Hence, grain import cost competition exerts a more powerful water use efficiency-improving effect in non-major grain-producing areas.

3.3.3. Path Test of Grain Import Cost Competition Reducing Domestic Grain Profits and Forcing Improvement in Water Use Efficiency

This study examines the mechanism through which grain import cost competition enhances water use efficiency by squeezing grain production profits. As reported in Column (1) of

Table 6. Pathway Mechanism Test Results of the Impact of Grain Import Cost Competition on Grain Production Water Use Efficiency.

Variable	Grain Production Profit	Grain Production Water Use Efficiency
	Model 1	Model 2	Model 3	Model 4
Grain Import Cost Competition	−0.008 ***		0.008 ***	0.007 **
	(−2.862)		(3.019)	(2.510)
Grain Production Profit		−0.190 ***		−0.159 **
		(−2.966)		(−2.447)
Technological Environment	−0.028 *	0.025 *	0.031 **	0.026 **
	(−1.866)	(1.936)	(2.467)	(2.093)
Water Use Structure	−0.008	0.005	−0.004	−0.002
	(−0.352)	(0.099)	(−0.077)	(−0.050)
Irrigation Ratio	−0.068	0.381 ***	0.363 ***	0.372 ***
	(−0.678)	(3.923)	(3.699)	(3.831)
Financial Support for Agriculture	−0.004	0.012 ***	0.011 **	0.011 **
	(−0.687)	(2.704)	(2.403)	(2.354)
Disaster Rate	−0.044	−0.168 ***	−0.163 ***	−0.172 ***
	(−1.102)	(−3.120)	(−3.028)	(−3.210)
Agricultural Mechanization Level	0.001	0.003	0.002	0.002
	(0.462)	(0.917)	(0.559)	(0.656)
Specialization Degree of Crop Planting	−0.225 ***	0.057	0.083	0.060
	(−3.065)	(0.694)	(1.014)	(0.739)
Environmental Regulation	0.016	0.002	0.005	0.007
	(0.944)	(0.136)	(0.415)	(0.593)
Time Fixed Effects	Yes	Yes	Yes	Yes
Individual Fixed Effects	Yes	Yes	Yes	Yes
Constant	0.283 ***	0.886 ***	0.848 ***	0.880 ***
	(6.003)	(12.839)	(12.269)	(12.713)
N	540	540	540	540

***, **, and * denote significant at the p < 1%, p < 5%, and p < 10% levels; the numbers in parentheses are t-values.

, import cost competition exerts a statistically significant negative effect on grain production profits at the 1% level, with a coefficient of −0.008, confirming that intensified import competition depresses domestic grain profitability. Column (2) reveals that grain production profits, in turn, have a significant negative influence on water use efficiency, with a coefficient of −0.190. The direct effect of import cost competition on water use efficiency, presented in Column (3), is positive and significant at the 1% level (coefficient = 0.008). When the profit variable is added in Column (4), the coefficient on import cost competition declines to 0.007 and remains significant only at the 5% level; both the magnitude and the statistical significance of the coefficient attenuate relative to Column (3). This pattern indicates that grain production profits serve as a mediating channel in the relationship between grain import cost competition and water use efficiency.

Prior research has documented that the virtual water content of China’s major imported grains (e.g., soybeans and corn) lies below the average level of domestic production. Under the trade pro-competitive effect, rising grain imports enable low-water-cost grain from international markets to enter the domestic market, exerting downward pressure on grain prices and eroding domestic producers’ profits. Large-scale grain producers, who typically enjoy higher water use efficiency and lower per-unit costs, are better positioned to survive this pressure, whereas small-scale producers with lower water use efficiency and higher costs are more likely to be forced out of the market [34,36]. Hence, cross-country differences in grain production water use efficiency are reflected in differences in virtual water content [38]. Through cost competition, grain imports reduce domestic producers’ market share, depress domestic grain prices, and cut domestic grain production profits, thus compelling domestic water use efficiency to improve toward a higher level [39].

4. Conclusions and Implications

4.1. Discussion and Conclusions

Drawing on a panel dataset covering 30 provincial-level administrative regions in China from 2003 to 2020, this study employs the water footprint method to quantify the cost competition between domestic and foreign grain production. Water use efficiency in grain production is measured through the DEA-GML index, which incorporates the ecological value of grain cultivation as a desirable output and treats agricultural non-point source pollution as an undesirable output. Anchored in the New New Trade Theory and the Induced Technological Innovation Theory and adopting a virtual water trade perspective, the empirical analysis yields the following conclusions: grain import cost competition, by transmitting shocks through the domestic market, erodes the profits of grain producers and thereby compels an improvement in China’s grain production water use efficiency.

For every 1 billion tons of virtual water saved through grain imports, due to the pro-competitive effect of trade, domestic water use efficiency is forced to increase by 0.008. The reason is that meager grain production profits make production efficiency a key factor affecting producer survival. Importing large quantities of grain products with lower water costs than domestic production not only saves certain water resources globally through trade welfare, but also, via the pro-competitive effect of trade, depresses domestic grain producer profits and gradually eliminates less efficient producers. The remaining producers are forced to improve technology and reduce production costs in order to survive, thereby driving up domestic water use efficiency.

The pro-competitive effect of trade caused by grain imports initially stays within the market’s self-adjustment range. However, when the annual volume of virtual water saved through grain imports in a region exceeds 11 billion tons, the import competition shock surpasses a critical threshold. Due to constraints in production facilities and technology, even previously relatively efficient grain producers cannot further improve production efficiency or reduce production costs in the short term, and they also face the risk of being eliminated. Excessive strong import competition instead undermines the allocative efficiency of grain production factors, leading to a decline rather than an improvement in water use efficiency.

The promoting effect of grain import cost competition on water use efficiency is stronger in northern regions than in southern regions. Because the northern region has a more developed livestock industry and a higher demand for feed grains such as soybeans and corn (the virtual water content of which is much lower than that of domestic production), the competitive effect that forces improvement in grain production efficiency is stronger. Moreover, since non-major grain-producing areas have more low-efficiency grain producers with smaller profit margins, the market shock from import cost competition is more likely to eliminate them. Consequently, the effect of grain import cost competition on forcing water use efficiency improvement is also stronger in non-major grain-producing areas.

4.2. Policy Implications

The above conclusions provide some new insights for improving China’s grain production water use efficiency. First, the government should increase investment in R&D of water-saving innovation technologies in regions with high grain production water footprints, such as the South China and Southwest China regions, to further reduce the grain production water footprint in these areas. Second, grain imports should be further expanded in regions with relatively low water use efficiency, such as the South China region and the Middle-Lower Yangtze River region, to exert a stronger pro-competitive effect of trade and induced technological innovation effect, thereby improving water use efficiency in these regions. Third, since the average virtual water content of China’s grain production (especially soybeans and corn) is higher than that of international imports, and the competitive effect caused by grain imports in most regions has not yet reached the critical threshold that would harm the grain industry, China should appropriately increase its imports of grains, particularly soybeans and corn, so as to achieve resource conservation while exerting the competitive effect of grain imports, thereby forcing an improvement in the efficiency of water resource use in domestic grain production. Fourth, to reduce China’s grain trade risks and promote the efficiency of domestic water and land resource use, the government should encourage the expansion of diversified channels to import water-intensive grain products from countries with high water use efficiency (i.e., countries with low virtual water content in grain production), thereby forcing an improvement in domestic water use efficiency.

4.3. Contributions

This study makes several contributions. First, it extends the application of trade and induced technological innovation theories to smallholder-dominated agricultural production, highlighting how import competition can influence resource use efficiency in the agricultural sector. Second, it integrates the concept of virtual water trade with empirical analysis of water use efficiency, providing a novel perspective on the environmental implications of grain imports. Third, by examining regional heterogeneity and potential nonlinear effects, the study offers insights into how the impact of import cost competition may vary across main grain producing and non-main producing areas. Finally, the findings have practical policy relevance, suggesting that trade policies and import management can play a role in promoting more sustainable water use in grain production, while recognizing that strategies may need to be tailored to regional production conditions.

4.4. Limitations

While this study provides new insights into the relationship between grain import competition and water use efficiency in China, some limitations should be noted. The analysis is conducted at the provincial level, which may overlook finer-scale heterogeneity among farms. Our measures, including import cost competition and water use efficiency, rely on available data and proxies, which may not capture all aspects of production and resource use. Additionally, the mechanism analysis is exploratory and does not consider all potential pathways. Future research using more detailed data and additional indicators could further deepen understanding of the link between grain imports and water use efficiency.