Digital Rural Construction and Agricultural Technological Choice Under Land Constraints: Evidence from China

Abstract

Cultivated land is a basic factor in agricultural production, but it is also subject to increasingly strict resource constraints. Under such conditions, agricultural development needs to rely more on the effective use of capital, labor, and digital technologies on existing farmland rather than on the expansion of land input. This study examines the impact of digital rural construction on agricultural technological choice under land constraints. Using panel data from 28 Chinese provinces from 2011 to 2024, this paper first measures capital–land and labor–land biased technological progress based on a normalized factor-augmenting CES production function, and then constructs a multidimensional index of digital rural construction. The empirical results show that digital rural construction significantly promotes both capital–land and labor–land biased technological progress, with a stronger effect on the capital–land dimension. This indicates that digital rural construction mainly promotes agricultural technological choice through capital-embodied and digitally enabled technologies, while also improving the role of labor in agricultural production and land management. The mechanism analysis shows that the capital–land effect is mainly driven by the price effect, whereas the labor–land effect is mainly driven by the scale effect. Further heterogeneity analysis indicates that the positive effects are more pronounced in eastern China, while the effects in central and western regions are weaker and more differentiated. These findings suggest that digital rural construction reshapes agricultural technological choice by changing how capital and labor are combined with limited land resources. The study provides evidence for promoting agricultural transformation, improving land-use sustainability, and strengthening digital governance in agriculture.

1. Introduction

Cultivated land is the basic carrier of agricultural production, but it is also one of the most rigidly constrained production factors. With rapid urbanization, ecological protection requirements, and the need to ensure food security, agricultural development increasingly faces the challenge of maintaining production capacity under limited land resources [1,2]. For countries with large populations and scarce per capita farmland, this issue is particularly important. In this context, agricultural transformation cannot rely mainly on the continuous expansion of land input. A more realistic path is to improve how capital, labor, technology, and services support production on existing farmland [3]. With the rapid development of rural digitalization, digital technologies are becoming an important means of reorganizing agricultural production factors and improving the use of existing farmland. Therefore, understanding agricultural technological choice under land constraints requires attention not only to land scarcity itself, but also to how digital transformation reshapes the relationship among capital, labor, and land.

Digital rural construction provides a new perspective for understanding this issue. It can be understood as the integration of digital infrastructure, smart agricultural technologies, digital finance, digital platforms, and data-based services into rural production and rural development. These technologies include broadband networks, smartphones, digital payment systems, e-commerce platforms, agricultural machinery service platforms, drones, sensors, GPS, remote sensing, smart irrigation systems, and data-driven decision-support tools [4]. They can improve information access, reduce transaction costs, support machinery and service matching, and strengthen production management in rural areas. However, the adoption of these technologies is not automatic. Farmers and agricultural producers may face barriers such as high initial investment costs, limited digital skills, insufficient rural infrastructure, weak agricultural service systems, fragmented land management, and uncertainty about the economic returns of digital technologies. Compared with traditional agricultural technologies, digital technologies are not only production tools; they also reshape the matching relationship among capital, labor, and land. Under land constraints, this may change the direction of agricultural technological progress and encourage agricultural production to rely more on the effective use of non-land factors rather than additional land inputs [5].

Existing studies have provided important insights into the relationship between digitalization and agriculture. One strand of literature focuses on the effects of digital rural construction, digital finance, and digital infrastructure on agricultural productivity, farmers’ income, agricultural modernization, and rural development. These studies generally find that digitalization can reduce information costs, improve resource allocation, and promote agricultural transformation [6,7]. Another strand of literature examines agricultural land use, land productivity, cultivated land protection, and food security, emphasizing that land scarcity and land-use constraints have become central issues in sustainable agricultural development [8,9]. A third strand of literature studies focus on biased technological progress, showing that technological progress may favor certain production factors and thereby change the relative efficiency of capital, labor, and land [10,11,12,13].

However, the existing literature still leaves several issues insufficiently addressed. First, many studies examine whether digitalization improves agricultural output, total factor productivity, or green efficiency, but pay less attention to how digitalization changes the direction of agricultural technological choice under land constraints [14,15]. Second, studies on land and agriculture often focus on land productivity, land transfer, or land-use efficiency, but rarely analyze how digital technologies enable capital and labor to support production on existing farmland [14,16]. Third, although biased technological progress provides a useful framework for understanding factor-oriented technological change, few studies apply this framework to examine the land-related effects of digital rural construction. As a result, we still know relatively little about whether digital rural construction promotes capital–land and labor–land biased technological progress, and through which mechanisms this influence occurs.

This paper addresses these gaps by examining the impact of digital rural construction on agricultural technological choice under land constraints. The core research question is: Does digital rural construction reshape agricultural technological choice by changing the relative role of capital and labor in relation to land? To answer this question, this study focuses on two land-related dimensions of biased technological progress: the capital–land biased technological progress index and the labor–land biased technological progress index. These two indicators allow us to examine whether digital rural construction promotes a technological path in which capital and labor better support agricultural production on existing farmland.

The theoretical logic of this paper is as follows. Under land constraints, agricultural production cannot rely mainly on the expansion of cultivated land. Digital rural construction may promote agricultural technological choice by improving the way non-land factors are combined with land. On the one hand, digital infrastructure, smart equipment, machinery services, digital finance, and agricultural service platforms may strengthen the role of capital-embodied technologies in agricultural production. This may promote capital–land biased technological progress. On the other hand, digital platforms, technical guidance, production information, and service systems may improve farmers’ decision-making and production organization, enabling labor to participate more effectively in land management and agricultural production. This may promote labor–land biased technological progress.

To provide systematic empirical evidence on this relationship, this study uses provincial panel data from China and constructs a multidimensional index of digital rural construction. Based on the theoretical framework, this paper further builds empirical models to examine both the baseline effect and the mechanism effect of digital rural construction on agricultural technological choice under land constraints. Specifically, the baseline model is used to estimate the impact of digital rural construction on capital–land and labor–land biased technological progress. The mechanism model further examines whether digital rural construction affects agricultural technological choice through the scale effect and the price effect. This research design links the theoretical mechanism with empirical testing and helps clarify how digital rural construction reshapes the relationship among capital, labor, and land in agricultural production.

This paper makes several contributions. First, it introduces a land-constraint perspective into the study of digital rural construction. Instead of only asking whether digitalization improves agricultural productivity, this paper examines how digitalization reshapes agricultural technological choice when land resources are limited. Second, it links digital rural construction with land-related biased technological progress by distinguishing between capital–land and labor–land dimensions. This helps clarify how digital technologies change the relationship between non-land factors and existing farmland. Third, it examines the scale effect and price effect as mechanisms, providing a more detailed explanation of how digital rural construction influences agricultural technological choice. This study provides evidence from China, a major agricultural country facing strong land constraints, thereby offering useful implications for agricultural transformation, cultivated land protection, and land-related digital governance.

2. Theoretical Analysis and Research Hypotheses

2.1. Digital Rural Construction and Agricultural Technological Choice Under Land Constraints

The theory of biased technological progress suggests that technological progress does not improve the productivity of all production factors equally. Instead, it may improve the efficiency of one factor more rapidly than that of another factor [17,18]. In agriculture, biased technological progress reflects changes in the relative productivity of capital, labor, and land. The direction of technological progress is shaped not only by factor endowments, but also by market conditions, institutional arrangements, and the diffusion of new technologies. Under the constraints of cultivated land protection and agricultural security, land has become a relatively rigid factor in agricultural production. Compared with capital and labor, land is more difficult to expand in the short term. Therefore, agricultural development increasingly depends on whether capital, labor, and digital technologies can better support production on existing farmland.

Digital rural construction provides new technological conditions for agricultural production under land constraints. Through the development of rural digital infrastructure, digital production systems, digital finance, digital transactions, and digital public services, digital rural construction changes the way agricultural producers obtain information, allocate resources, adopt technologies, and organize production [19]. Rural networks, smart terminals, agricultural meteorological information, and digital platforms reduce the cost of information acquisition and transmission [20]. Digital finance and online transactions improve the accessibility of agricultural capital and production services [21]. Smart agricultural equipment, machinery services, sensors, irrigation systems, drones, and digital management platforms further promote the integration of digital technologies with agricultural production [22].

From the perspective of land constraints, the key role of digital rural construction is not to expand land resources, but to reshape how non-land factors interact with land. Digital technologies can improve the ability of capital to support production on existing farmland. For example, smart machinery, digital equipment, agricultural machinery services, and platform-based production services allow capital to be more effectively embedded in land-related production processes [22,23]. This may lead agricultural technological progress to become more oriented toward capital relative to land. In this sense, capital-related technological progress reflects a pathway in which agricultural production responds to land constraints by strengthening the role of capital-embodied and digitally enabled technologies.

At the same time, digital rural construction may also improve the role of labor in agricultural production under land constraints. Digital platforms, technical guidance, market information, weather information, and production management tools can improve farmers’ information access and decision-making capacity [24]. Labor is not only a traditional production input, but also an important carrier of digital information, production coordination, and land management. When farmers and agricultural operators are better connected with digital services and agricultural service systems, labor can more effectively support agricultural production on existing farmland. Therefore, digital rural construction may also promote labor–land biased technological progress.

Based on the above analysis, this study proposes the following hypotheses:

Hypothesis 1a.

Under land constraints, digital rural construction promotes capital–land biased technological progress in agriculture.

Hypothesis 1b.

Under land constraints, digital rural construction promotes labor–land biased technological progress in agriculture.

2.2. Scale Effect, Price Effect, and Biased Agricultural Technological Progress

Digital rural construction may affect agricultural technological choice under land constraints through changes in factor allocation and relative factor returns. In agricultural production, capital, labor, and land are not independent of each other. The effect of digital technologies depends on how they change the relationship among these factors. Since cultivated land is difficult to expand, the adjustment of capital and labor around existing farmland becomes an important way for agriculture to respond to land constraints. In this process, the scale effect and the price effect are two important channels through which digital rural construction may influence biased technological progress.

First, digital rural construction may affect agricultural technological progress through the scale effect. The improvement of digital infrastructure, agricultural information services, and digital production systems can reduce information asymmetry and transaction costs in agricultural production. It can also improve the efficiency of land transfer, machinery service matching, and production service coordination. These changes may promote more organized forms of agricultural production and improve the matching between land, capital, labor, and agricultural services [25,26]. Under more organized production conditions, machinery services, digital management, standardized production, and platform-based agricultural services are more likely to be adopted. Therefore, the scale effect does not only refer to the expansion of operation size; it also reflects the reorganization of production factors around land. By improving factor coordination, digital rural construction may change the relative efficiency of capital, labor, and land, thereby affecting the direction of agricultural technological progress.

Second, digital rural construction may affect agricultural technological progress through the price effect. According to the theory of induced technological change, changes in relative factor prices can influence the direction of technological progress [17,18,27]. When the use cost of a factor declines or its relative return increases, technological progress is more likely to improve the efficiency of that factor. Digital rural construction can affect relative factor prices by reducing information costs, transaction costs, financing constraints, and service access costs. For example, digital finance can improve access to agricultural capital; digital platforms can reduce the cost of matching machinery services and production services; and online transactions can improve the efficiency of agricultural input and output markets. These changes may alter the cost–benefit conditions of capital, labor, and land use, thus influencing technological choice.

From the perspective of land constraints, both the scale effect and the price effect are closely related to how agriculture uses existing farmland. The scale effect changes the organizational conditions under which capital and labor are combined with land, while the price effect changes the economic incentives for adopting different technologies. Together, these two channels may guide agricultural technological progress toward different factor directions. Therefore, digital rural construction may influence capital–land and labor–land biased technological progress by changing both factor organization and relative factor returns.

Based on the above analysis, this study proposes the following hypothesis:

Hypothesis 2.

Digital rural construction affects biased agricultural technological progress through both the scale effect and the price effect.

3. Research Design

3.1. Measurement of Biased Agricultural Technological Progress

According to classical economic growth theory, production depends on both factor inputs and technology. Since this study focuses on agricultural production, land is introduced as a key production factor in addition to capital and labor. The general production function can be written as follows:

$$Y=F({\Gamma }_t^{K}K,{\Gamma }_t^{N}N,{\Gamma }_t^C{C}_t)$$

(1)

where $Y$ denotes agricultural output; $K$, $L$, and $C$ represent capital, labor, and land inputs, respectively; ${\Gamma }_t^K$, ${\Gamma }_t^N$ and ${\Gamma }_t^C$ denote capital-augmenting, labor-augmenting, and land-augmenting technological progress, respectively. Equation (1) is a general form that can describe different types of technological progress. When ${\Gamma }_t^K\ne {\Gamma }_t^N\ne {\Gamma }_t^C$, technological progress is neutral. When the growth rates of ${\Gamma }_t^K$, ${\Gamma }_t^N$ and ${\Gamma }_t^C$ differ, technological progress is factor-biased.

To construct an operational measurement of biased technological progress and avoid measurement bias caused by differences in index benchmarks, this study adopts a normalized factor-augmenting CES production function. The specific form is as follows1:

$$Y_t=Y_0{\left[\alpha {\left({\displaystyle \frac{K_t}{K_0}}{\displaystyle \frac{{\Gamma }_t^K}{{\Gamma }_0^K}}\right)}^{{\displaystyle \frac{\sigma -1}{\sigma }}}+\beta {\left({\displaystyle \frac{N_t}{N_0}}{\displaystyle \frac{{\Gamma }_t^N}{{\Gamma }_0^N}}\right)}^{{\displaystyle \frac{\sigma -1}{\sigma }}}+\gamma {\left({\displaystyle \frac{C_t}{C_0}}{\displaystyle \frac{{\Gamma }_t^C}{{\Gamma }_0^C}}\right)}^{{\displaystyle \frac{\sigma -1}{\sigma }}}\right]}^{{\displaystyle \frac{\sigma }{\sigma -1}}}$$

(2)

where the subscript 0 denotes the normalization benchmark; $Y_0$, $K_0$, $L_0$, and $C_0$ denote output, capital input, labor input, and land input at the benchmark point, respectively. $\alpha ,\beta \mathrm{and} \gamma$ are distribution parameters, and $\sigma$ denotes the elasticity of substitution among factors.

Biased technological progress reflects changes in the relative marginal products of factors caused by changes in relative factor efficiencies. To capture this relationship, this study first examines the response of the relative marginal product of two factors to changes in their relative efficiency:

$$R{P}_{ij}^{\prime }={\displaystyle \frac{𝜕{RP}_{ij}}{𝜕({\Gamma }_t^i/{\Gamma }_t^j)}}{\displaystyle \frac{d({\Gamma }_t^i/{\Gamma }_t^j)}{dt}}$$

(3)

where $R{P}_i$ denote the marginal products of factors $i$ and $j$, respectively. $R{P}^{\prime }$ reflects the extent to which changes in relative factor efficiency induced by technological progress affect the relative marginal product of the two factors. Following the relevant literature [18], this study uses the growth rate of the relative marginal product between factors $i$ and $j$ to measure the direction and degree of biased technological progress:

$$BT{C}_t^{ij}={\displaystyle \frac{R{P}^{\prime }}{RP}}={\displaystyle \frac{1}{RP}}{\displaystyle \frac{𝜕RP}{𝜕({\Gamma }_t^i/{\Gamma }_t^j)}}{\displaystyle \frac{d({\Gamma }_t^i/{\Gamma }_t^j)}{dt}}={\displaystyle \frac{\sigma -1}{\sigma }}({\displaystyle \frac{{\dot{\Gamma }}_t^i}{{\Gamma }_t^i}}-{\displaystyle \frac{{\dot{\Gamma }}_t^j}{{\Gamma }_t^j}})$$

(4)

In this study, the main focus is on two land-related technological bias indices: the capital–land biased technological progress index and the labor–land biased technological progress index. These two indicators are used to identify how digital rural construction changes the relative roles of capital and labor in supporting agricultural production under land constraints.

3.2. Parameter Estimation

This study estimates the parameters of the normalized CES production function using a normalized supply-side system. The system consists of the normalized CES production function, the first-order conditions under profit maximization, and the corresponding constraint equations. Different from studies that assume perfect competition [28], this paper introduces a markup parameter $\mu$ into the supply-side system to capture imperfect competition in agricultural production. This setting is more consistent with the reality of agricultural factor markets and can improve the explanatory power of the estimation results.

Following the factor-augmenting technological progress framework, this study further assumes that factor-augmenting technology evolves in an exponential form with a variable growth rate:

$${\Gamma }_t^i={\Gamma }_0^i\exp ({\gamma }_i{t}_0{\lambda }_i^{-1}({(t/t_0)}^{{\lambda }_i}-1))$$

(5)

where ${\gamma }_i$ is the first-order growth coefficient of factor-augmenting technological progress, reflecting the direction and speed of technological progress; and ${\lambda }_i$ is the technological curvature parameter, reflecting the dynamic trend of the growth rate of factor-augmenting technology.

Based on the normalized CES production function and the profit maximization conditions, the normalized supply-side system can be derived as follows2:

$$\left\{\begin{array}{l}\ln r_t=\ln ({\displaystyle \frac{\alpha }{1+\mu }}{\displaystyle \frac{\xi \overline{Y}}{\overline{K}}})+{\displaystyle \frac{1}{\sigma }}\ln ({\displaystyle \frac{Y_t/\xi \overline{Y}}{K_t/\overline{K}}})+{\displaystyle \frac{\sigma -1}{\sigma }}{\epsilon }_K\overline{t}{\lambda }_K^{-1}[{({\displaystyle \frac{t}{\overline{t}}})}^{{\lambda }_K}-1] \\ \ln w_t=\ln ({\displaystyle \frac{\beta }{1+\mu }}{\displaystyle \frac{\xi \overline{Y}}{\overline{N}}})+{\displaystyle \frac{1}{\sigma }}\ln ({\displaystyle \frac{Y_t/\xi \overline{Y}}{N_t/\overline{N}}})+{\displaystyle \frac{\sigma -1}{\sigma }}{\epsilon }_N\overline{t}{\lambda }_N^{-1}[{({\displaystyle \frac{t}{\overline{t}}})}^{{\lambda }_N}-1] \\ \ln e_t=\ln ({\displaystyle \frac{\gamma }{1+\mu }}{\displaystyle \frac{\xi \overline{Y}}{\overline{C}}})+{\displaystyle \frac{1}{\sigma }}\ln ({\displaystyle \frac{Y_t/\xi \overline{Y}}{C_t/\overline{C}}})+{\displaystyle \frac{\sigma -1}{\sigma }}{\epsilon }_C\overline{t}{\lambda }_C^{-1}[{({\displaystyle \frac{t}{\overline{t}}})}^{{\lambda }_C}-1] \\ \ln {\displaystyle \frac{Y_t}{\xi \overline{Y}}}={\displaystyle \frac{\sigma }{\sigma -1}}\ln \left[\begin{array}{l}\alpha {\left({\displaystyle \frac{K_t}{\overline{K}}}{e}^{{\epsilon }_K\overline{t}{\lambda }_K^{-1}\left({(t/\overline{t})}^{{\lambda }_K}-1\right) }\right)}^{{\displaystyle \frac{\sigma -1}{\sigma }}}+\beta {\left({\displaystyle \frac{N_t}{\overline{N}}}{e}^{{\epsilon }_N\overline{t}{\lambda }_N^{-1}\left({(t/\overline{t})}^{{\lambda }_N}-1\right) }\right)}^{{\displaystyle \frac{\sigma -1}{\sigma }}}\\ +\gamma {\left({\displaystyle \frac{C_t}{\overline{C}}}{e}^{{\epsilon }_C\overline{t}{\lambda }_C^{-1}\left({(t/\overline{t})}^{{\lambda }_C}-1\right) }\right)}^{{\displaystyle \frac{\sigma -1}{\sigma }}}\end{array}\right]\end{array}\right.$$

(6)

where $r_t$, $w_t$ and $e_t$ denote the prices of capital, labor, and land in period $t$, respectively. $\xi$ denotes the scale factor. This system links agricultural output, factor inputs, factor prices, and factor-augmenting technological progress, allowing the estimation of capital-, labor-, and land-augmenting efficiency changes.

The nonlinear seemingly unrelated regression method is used to estimate the above system. The initial value of the substitution elasticity is set as an arithmetic sequence from 0.02 to 2.02 with a tolerance interval of 0.05. The model is estimated repeatedly, and the optimal result is selected according to the Log-Det statistic and model stability. The benchmark values of output, factor inputs, and factor prices are set using the sample geometric mean and arithmetic mean, while the initial values of the remaining parameters are set with reference to previous studies [29].

3.3. Baseline Empirical Model

To examine the impact of digital rural construction on agricultural technological choice under land constraints, this study constructs the following baseline model:

$$BT{C}_{\omega t}^{ij}={\beta }_0+{\beta }_{1}D{R}_{\omega t}+{\phi }_{\omega }{\displaystyle \sum _{\omega }Contro{l}_{\omega t}+}{\mu }_{\omega }+{\eta }_t+{\epsilon }_{\omega t}$$

(7)

where $\omega$ denotes province and $t$ denotes year. BTC represents the biased technological progress index between factors $i$ and $j$. In the empirical analysis, this study mainly focuses on the capital–land and labor–land biased technological progress indices. DR denotes the level of digital rural construction. Control represents a set of control variables. Except for the biased technological progress indices, which may take negative values, all other variables are expressed in logarithmic form. Β is the coefficient of interest, capturing the impact of digital rural construction on agricultural technological choice under land constraints. $\mu$ denotes province fixed effects, which control for time-invariant provincial characteristics; $\eta$ denotes year fixed effects, which control for common time shocks; and $\epsilon$ is the random disturbance term.

3.4. Mechanism Model

This study further constructs a mechanism model to examine how digital rural construction affects biased agricultural technological progress. Based on the normalized CES production function, the marginal products of capital, labor, and land can be derived as follows:

$$\begin{array}{l}{\displaystyle \frac{𝜕Y_t}{𝜕K_t}}=\alpha {\displaystyle \frac{Y_0}{K_0}}{\left({\displaystyle \frac{{\Gamma }_t^K}{{\Gamma }_0^K}}\right)}^{{\displaystyle \frac{\sigma -1}{\sigma }}}{\left({\displaystyle \frac{Y_t/Y_0}{K_t/K_0}}\right)}^{{\displaystyle \frac{1}{\sigma }}}\\ {\displaystyle \frac{𝜕Y_t}{𝜕N_t}}=\beta {\displaystyle \frac{Y_0}{N_0}}{\left({\displaystyle \frac{{\Gamma }_t^N}{{\Gamma }_0^N}}\right)}^{{\displaystyle \frac{\sigma -1}{\sigma }}}{\left({\displaystyle \frac{Y_t/Y_0}{N_t/N_0}}\right)}^{{\displaystyle \frac{1}{\sigma }}}\\ {\displaystyle \frac{𝜕Y_t}{𝜕C_t}}=\gamma {\displaystyle \frac{Y_0}{C_0}}{\left({\displaystyle \frac{{\Gamma }_t^C}{{\Gamma }_0^C}}\right)}^{{\displaystyle \frac{\sigma -1}{\sigma }}}{\left({\displaystyle \frac{Y_t/Y_0}{C_t/C_0}}\right)}^{{\displaystyle \frac{1}{\sigma }}}\end{array}$$

(8)

The relative marginal product between factors $i$ and $j$ can then be expressed as:

$$RP={\displaystyle \frac{𝜕Y/𝜕i}{𝜕Y/𝜕j}}={\displaystyle \frac{{\alpha }_i}{{\alpha }_j}}{\displaystyle \frac{j_0}{i_0}}{({\displaystyle \frac{j_t/j_0}{i_t/i_0}})}^{{\displaystyle \frac{1}{\sigma }}}{\left({\displaystyle \frac{{\Gamma }_t^i/{\Gamma }_0^i}{{\Gamma }_t^j/{\Gamma }_0^j}}\right)}^{{\displaystyle \frac{\sigma -1}{\sigma }}},{\alpha }_i=(\alpha ,\beta ,\gamma )$$

(9)

Accordingly, the expression for biased technological progress can be obtained as:

$$BT{C}_t^{ij}={\displaystyle \frac{1}{RP}}{\displaystyle \frac{𝜕RP}{𝜕({\Gamma }_t^i/{\Gamma }_t^j)}}{\displaystyle \frac{d({\Gamma }_t^i/{\Gamma }_t^j)}{dt}}={\displaystyle \frac{\sigma -1}{\sigma }}{\displaystyle \frac{1}{({\Gamma }_t^i/{\Gamma }_t^j)}}{\displaystyle \frac{d({\Gamma }_t^i/{\Gamma }_t^j)}{dt}}$$

(10)

Under the profit maximization condition of the normalized CES production function, the factor-augmenting technological progress coefficient can be further derived:

$$\begin{array}{l}{\Gamma }_t^K={\displaystyle \frac{Y_t}{K_t}}{\left[{\displaystyle \frac{r_t{K}_t}{(w_t{N}_t+r_t{K}_t+e_t{C}_t)}}\right]}^{{\displaystyle \frac{\sigma }{\sigma -1}}}\\ {\Gamma }_t^N={\displaystyle \frac{Y_t}{N_t}}{\left[{\displaystyle \frac{w_t{N}_t}{(w_t{N}_t+r_t{K}_t+e_t{C}_t)}}\right]}^{{\displaystyle \frac{\sigma }{\sigma -1}}}\\ {\Gamma }_t^C={\displaystyle \frac{Y_t}{C_t}}{\left[{\displaystyle \frac{e_t{C}_t}{(w_t{N}_t+r_t{K}_t+e_t{C}_t)}}\right]}^{{\displaystyle \frac{\sigma }{\sigma -1}}}\end{array}$$

(11)

And the relative factor efficiency can be expressed as a function of relative input changes and relative price changes. Taking capital and land as an example, the relative factor efficiency can be written as:

$${\displaystyle \frac{{\Gamma }_t^K}{{\Gamma }_t^C}}={\left({\displaystyle \frac{K_t}{C_t}}\right)}^{{\displaystyle \frac{1}{\sigma -1}}}{\left({\displaystyle \frac{r_t}{e_t}}\right)}^{{\displaystyle \frac{\sigma }{\sigma -1}}}$$

(12)

Substituting this expression into the biased technological progress index (taking capital and land as an example):

$$\begin{array}{l}BT{C}_t^{KC}={\displaystyle \frac{\sigma -1}{\sigma }}{\displaystyle \frac{d\ln ({\Gamma }_t^K/{\Gamma }_t^C)}{dt}}\\ \hspace{1em}\hspace{1em}={\displaystyle \frac{\sigma -1}{\sigma }}{\displaystyle \frac{d\left[\frac{1}{\sigma -1}\ln \left(\frac{K_t}{C_t}\right)+\frac{\sigma }{\sigma -1}\ln \left(\frac{r_t}{e_t}\right)\right]}{dt}}\\ \hspace{1em}\hspace{1em}={\displaystyle \frac{d\left[\frac{1}{\sigma }\ln \left(\frac{K_t}{C_t}\right)+\ln \left(\frac{r_t}{e_t}\right)\right]}{dt}}\\ \hspace{1em}\hspace{1em}={\displaystyle \frac{\frac{1}{\sigma }d\ln \left(\frac{K_t}{C_t}\right)+d\ln \left(\frac{r_t}{e_t}\right)}{dt}}\end{array}$$

(13)

Substituting this expression into the biased technological progress index shows that the direction of biased technological progress theoretically depends on two channels: changes in relative factor inputs and changes in relative factor prices. The former is defined as the scale effect, and the latter is defined as the price effect. Therefore, the effect of digital rural construction on agricultural technological choice under land constraints may be realized by changing factor input structures and relative factor price relationships.

Based on the above theoretical mechanism, this study further constructs the following empirical models:

$$Scal{e}_{\omega t}={\gamma }_0+{\gamma }_{1}D{R}_{\omega t}+{\theta }_{\omega }{\displaystyle \sum _{\omega }Contro{l}_{\omega t}+}{\nu }_{\omega }+{\rho }_t+{\tau }_{\omega t}$$

(14)

$$Pric{e}_{\omega t}={\alpha }_0+{\alpha }_{1}D{R}_{\omega t}+{\chi }_{\omega }{\displaystyle \sum _{\omega }Contro{l}_{\omega t}+}{\nu }_{\omega }+{\rho }_t+{\tau }_{\omega t}$$

(15)

where $Scale$ is measured by the input ratio between factors, and $Price$ is measured by the price ratio between factors. In this paper, the scale effect is mainly captured by the ratio of capital input to land input, and the price effect is captured by the ratio of capital price to land price. The control variables are consistent with those in the baseline model. $\nu$ and $\rho$ denote province and year fixed effects, respectively, and $\tau$ is the random disturbance term.

3.5. Data Sources

This study uses panel data for 28 provinces in China from 2011 to 2024. Due to data availability, Hong Kong, Macao, Taiwan, and Tibet are not included in the sample. These provinces are distributed across eastern, central, and western China, providing broad geographic coverage of China’s major agricultural regions. To improve the transparency of the sample, the provinces included in the study and their regional classifications are reported in

Table 1. Provinces Included in the Study.

Region	Provinces
Eastern China	Beijing, Tianjin, Hebei, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, Hainan
Central China	Shanxi, Anhui, Jiangxi, Henan, Hubei, Hunan, Jilin, Heilongjiang
Western China	Inner Mongolia, Guangxi, Sichuan, Guizhou, Yunnan, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang

. The data are mainly obtained from the China Statistical Yearbook, Historical Data of China’s Gross Domestic Product Accounting, China Population and Employment Statistical Yearbook, China Agricultural Yearbook, China Rural Statistical Yearbook, China Tertiary Industry Statistical Yearbook, and provincial statistical yearbooks [30,31,32,33,34]. Missing values are supplemented and adjusted using interpolation methods and relevant statistical materials.

3.6. Variable Description

The dependent variables are the agricultural biased technological progress indices. Based on the measurement model described above, this study calculates the capital–labor, capital–land, and labor–land biased technological progress indices. Since the focus of this paper is agricultural technological choice under land constraints, the empirical analysis mainly uses the capital–land and labor–land biased technological progress indices to examine how digital rural construction changes the relationship between non-land factors and land.

The core explanatory variable is digital rural construction. Digital rural construction reflects the level of rural digital development, but there is still no unified evaluation system in the literature. Based on existing studies and the practical characteristics of agricultural digitalization, this study constructs a digital rural construction index from four dimensions: rural digital infrastructure, rural digital production, rural digital economy, and rural digital life. The index contains 15 secondary indicators, as shown in

Table 2. Evaluation Indicator System for the Level of Digital Rural Construction.

Primary Indicator	Secondary Indicator	Indicator Description
Rural Digital Infrastructure	Rural smartphone penetration rate	Number of mobile phones owned by rural households at year-end per 100 households/unit
	Rural radio and television penetration rate	Television access rate of rural households per 100 households/%
	Rural computer penetration rate	Number of computers owned by rural households at year-end per 100 households/unit
	Rural network construction level	Number of rural broadband access households/10,000 households
	Agricultural meteorological observation stations	Number of agricultural meteorological observation services/unit
Rural Digital Production	Agricultural mechanization level	Total power of agricultural machinery/(104) kW
	Agricultural digital production level	Number of employees in the primary industry × number of enterprises using Internet technologies for production and operation/unit
	Agricultural production investment intensity	Fixed asset investment in agriculture, forestry, animal husbandry, and fishery/100 million yuan
	Rural online payment level	Digital financial inclusion index/dimensionless
	Rural digital transaction level	Number of employees in the primary industry × e-commerce purchase and sales value/100 million yuan
	Rural digital technology purchasing capacity	Per capita disposable income of rural residents/yuan
Rural Digital Life	Rural digital service consumption level	Per capita transportation and communication consumption expenditure of rural households/yuan
	Rural digital cultural development level	Number of township cultural stations/unit
	Rural postal delivery level	Average number of postal deliveries per 100 rural households/times
	Rural electricity consumption level	Rural electricity consumption/rural population/(109) kW·h per 10,000 persons

. Compared with existing measurements, this index pays more attention to the application of digital technologies in agricultural production, factor allocation, and rural economic activities. Therefore, it can better reflect the basis through which digital rural construction affects agricultural technological choice under land constraints.

Specifically, rural digital infrastructure reflects the level of digital infrastructure construction in rural areas and is measured by indicators such as rural smartphone penetration, rural radio and television penetration, rural computer penetration, rural broadband access, and agricultural meteorological observation stations. Rural digital production reflects the application of digital technologies in agricultural production and is measured by agricultural mechanization, digitalized agricultural production, and agricultural production investment. Rural digital economy reflects the development of digital transactions and digital finance in rural areas and is measured by rural online payment, rural digital transactions, and rural digital technology purchasing capacity. Rural digital life reflects rural residents’ digital consumption and public service conditions and is measured by rural digital service consumption, digital cultural development, postal delivery, and electricity consumption.

To measure the overall level of digital rural construction, this study adopts an improved entropy weight method [35]. The traditional entropy weight method determines indicator weights according to the amount of information contained in each indicator. Although this method is clear and easy to implement, it may assign an excessively large weight to an indicator when its values show a high degree of dispersion, causing a single indicator to have a disproportionate influence on the final evaluation result. To address this issue, this study improves the entropy weight method by drawing on the logic of the analytic hierarchy process. Specifically, the difference coefficients of indicators are compared pairwise, and the comparison results are mapped onto a 1–9 scale to construct a judgment matrix based on information entropy. The eigenvector corresponding to the maximum eigenvalue of this matrix is then calculated and normalized to obtain the final indicator weights. This improved method retains the information-based weighting advantage of the entropy method while reducing the excessive influence of highly dispersed individual indicators on the digital rural construction index.

The mechanism variables are the scale effect and the price effect. According to the theoretical model and mechanism analysis, the scale effect is measured by the ratio of factor inputs, while the price effect is measured by the ratio of factor prices. These variables are used to examine whether digital rural construction affects the direction of agricultural technological progress by changing factor input structures and relative price relationships.

The control variables include agricultural infrastructure, rice–wheat planting structure, agricultural electricity use, agricultural research input, industrial structure, and agricultural mechanization. Agricultural infrastructure is measured by the proportion of irrigated area in total sown area, reflecting the impact of agricultural production conditions on factor allocation efficiency. Rice–wheat planting structure is measured by the proportion of rice sown area in the total rice and wheat sown area, controlling for the effect of crop structure on factor demand and productivity. Agricultural electricity use is used to reflect basic agricultural production conditions. Agricultural research input is measured by the stock of agricultural research investment calculated using the perpetual inventory method, reflecting the influence of agricultural research and development (R&D) activities on factor productivity. Industrial structure is measured by the share of primary industry value added in regional GDP, controlling for the effect of structural change on agricultural factor efficiency. Following previous studies on agricultural mechanization in China, this study measures agricultural mechanization by the total power of agricultural machinery, which reflects the level of agricultural capital input and production technology conditions.

For the data used to calculate the biased technological progress indices, agricultural output is measured using the income approach according to the requirements of the normalized supply-side system. Specifically, agricultural gross output is adjusted by deducting net production taxes, and missing components are estimated using the proportional structure of regional input–output tables. The output series is deflated using the value-added index of the primary industry. Capital input is measured by agricultural capital stock, which is estimated using the perpetual inventory method. The depreciation rate and related parameters follow existing studies [36], and the agricultural means of production price index is used for deflation. Labor input is measured by the number of employees in the primary industry. Land input is measured by crop sown area, taking multiple cropping into account. Capital price is measured by the ratio of agricultural fixed capital depreciation to capital input. Following Wang and Yuan [37], this study estimates land rent by converting the average output value of major crops per mu of cultivated land according to the proportion used in the literature. Labor income is obtained by subtracting capital income and land income from agricultural output, and agricultural labor price is then calculated by combining labor income with labor input.

4. Empirical Results and Analysis

4.1. Baseline Regression Results

Table 3. Baseline Regression.

	(1)	(2)	(3)	(4)
	Capital–Land Biased Technological Progress Index		Labor–Land Biased Technological Progress Index
VARIABLES	Without controls	With controls	Without controls	With controls
Digital	0.961 ***	0.984 ***	0.515 ***	0.407 ***
	(0.233)	(0.218)	(0.119)	(0.128)
Observations	392	392	392	392
R-squared	0.132	0.159	0.211	0.232
Controls	NO	YES	NO	YES
Province FE	YES	YES	YES	YES
Year FE	YES	YES	YES	YES

Note: Standard errors are reported in parentheses. *** indicates significance at the 1% level.

reports the baseline regression results for the impact of digital rural construction on agricultural technological choice under land constraints. Columns (1) and (2) use the capital–land biased technological progress index as the dependent variable. Column (1) reports the result without control variables, while column (2) reports the result after adding control variables. Columns (3) and (4) use the labor–land biased technological progress index as the dependent variable. Column (3) reports the result without control variables, while column (4) includes control variables. The results show that digital rural construction has a significantly positive effect on both the capital–land and labor–land biased technological progress indices, indicating that rural digital transformation plays an important role in shaping agricultural technological choice under land constraints.

For the capital–land biased technological progress index, the coefficient of digital rural construction remains significantly positive after the inclusion of control variables. As shown in column (2), the coefficient is 0.984. This result suggests that digital rural construction promotes a technological choice pattern in which capital becomes more effective in supporting agricultural production on existing land. Under the constraint of limited cultivated land, agricultural growth cannot rely mainly on the expansion of land input. Digital rural construction provides new technological conditions through digital infrastructure, smart equipment, agricultural machinery services, digital finance, and platform-based production services. These digital and capital-embodied technologies help agricultural producers use capital more effectively in the production process, so that existing farmland can carry a higher level of agricultural production [38]. This result shows that the role of digital rural construction is not simply to introduce digital tools into rural areas, but to change the way capital is combined with land. In traditional agricultural production, capital input may be limited by information barriers, service mismatch, financing constraints, and weak access to machinery or technical services. Digital rural construction can reduce these barriers by improving information flow, service matching, and the accessibility of capital-embodied technologies [39]. As a result, capital can be more effectively embedded in agricultural production, and agricultural technological choice becomes less dependent on land expansion and more dependent on the effective use of machinery, equipment, digital platforms, and production services.

For the labor–land biased technological progress index, the coefficient of digital rural construction is also significantly positive after adding control variables. As reported in column (4), the coefficient is 0.407. This result indicates that digital rural construction also promotes a technological choice pattern in which labor becomes more effective in supporting production on existing land. Digital technologies can improve farmers’ access to production information, enhance agricultural decision-making, facilitate technical guidance, and strengthen the connection between farmers and agricultural service systems [39]. Through these channels, labor can participate more effectively in production organization, land management, and the coordination of agricultural activities. Compared with the coefficient in the capital–land model, the coefficient in the labor–land model is smaller. This suggests that the impact of digital rural construction on agricultural technological choice under land constraints is more strongly reflected in the capital-related pathway. This is consistent with the practical process of rural digital transformation. Many digital agricultural applications are closely connected with capital-embodied technologies, such as smart agricultural equipment, agricultural machinery services, sensors, irrigation systems, drones, and digital management platforms. Labor also benefits from digital transformation, but this effect may depend more on farmers’ digital skills, production experience, and organizational capacity. Therefore, the labor-related pathway exists, but it is weaker than the capital-related pathway.

Overall, the baseline results show that digital rural construction reshapes agricultural technological choice under land constraints by improving the way non-land factors support existing farmland. The stronger capital–land effect indicates that digital rural construction mainly works through capital-embodied and digitally enabled technologies, while the positive labor–land effect suggests that digital tools also improve the role of labor in agricultural production and land management. These findings provide initial evidence that digital rural construction helps agricultural development move away from a land-expansion-dependent path and toward a technological choice pattern in which capital, labor, and digital services are more effectively combined with limited land resources.

4.2. Robustness Tests

To further examine the robustness of the baseline regression results, this study conducts robustness tests from three aspects: replacing the core explanatory variable, changing the model estimation method, and adding additional control variables. These tests are used to verify whether the impact of digital rural construction on agricultural technological choice under land constraints depends on the measurement of digital rural construction, the choice of estimation method, or the setting of control variables.

First, this study replaces the core explanatory variable. Considering that the measurement of digital rural construction may affect the estimation results, two alternative indicators of digital rural construction are constructed. On the one hand, while keeping the original indicator system unchanged, this study uses the vertical-and-horizontal scatter degree method to recalculate the digital rural construction index, denoted as Digital1. On the other hand, this study further constructs a new digital rural construction indicator system, denoted as Digital2 [40,41]. The alternative indicator system includes four dimensions: digital rural information infrastructure, digital rural financial infrastructure, digital rural service platforms, and digital rural life scenarios. Specifically, digital rural information infrastructure is measured by rural broadband access users and rural smartphone penetration; digital rural financial infrastructure is measured by the coverage breadth, usage depth, and digitalization level of digital financial inclusion; digital rural service platforms are measured by rural delivery route length and the number of Taobao villages; and digital rural life scenarios are measured by rural residents’ transportation and communication consumption expenditure and total telecommunication business volume. Compared with the original indicator system, this alternative measurement places more emphasis on the basic infrastructure, financial support, platform services, and living scenarios of rural digital transformation. It can therefore be used to test whether the baseline results are sensitive to the construction of the digital rural construction index. The estimation results are reported in columns (1)–(2) and columns (3)–(4) of

Table 4. Robustness Test: Replacing the Explanatory Variable.

	(1)	(2)	(3)	(4)
	vertical-and-horizontal scatter degree method		Alternative Indicator System
VARIABLES	Capital–Land Biased Technological Progress Index	Labor–Land Biased Technological Progress Index	Capital–Land Biased Technological Progress Index	Labor–Land Biased Technological Progress Index
Digital1	0.535 ***	0.320 ***
	(0.162)	(0.124)
Digital2			0.562 ***	0.242 ***
			(0.132)	(0.082)
Observations	392	392	392	392
R-squared	0.150	0.232	0.159	0.232
Controls	YES	YES	YES	YES
Province FE	YES	YES	YES	YES
Year FE	YES	YES	YES	YES

Note: Standard errors are reported in parentheses. *** indicates significance at the 1% level.

, respectively. The results show that after replacing the digital rural construction index, the effects of digital rural construction on both the capital–land and labor–land biased technological progress indices remain generally consistent with the baseline results in terms of direction and significance. This indicates that the baseline findings are not driven by a specific measurement method or indicator system of digital rural construction.

Second, this study changes the model estimation method. The baseline model may be affected by the choice of estimation method, especially when panel data contain within-group autocorrelation, cross-sectional heteroskedasticity, and contemporaneous correlation. Therefore, following Chen [42], this study adopts the feasible generalized least squares method to re-estimate the baseline model. The results are reported in columns (1)–(2) of

Table 5. Robustness Test: Changing the Estimation Method and Adding Control Variables.

	(1)	(2)	(3)	(4)
	Changing the Estimation Method		Adding Control Variables
VARIABLES	Capital–Land Biased Technological Progress Index	Labor–Land Biased Technological Progress Index	Capital–Land Biased Technological Progress Index	Labor–Land Biased Technological Progress Index
Digital	0.967 ***	0.482 ***	0.989 ***	0.629 ***
	(0.041)	(0.019)	(0.191)	(0.152)
Observations	392	392	392	392
R-squared			0.163	0.262
Controls	YES	YES	YES	YES
Province FE	YES	YES	YES	YES
Year FE	YES	YES	YES	YES

Note: Standard errors are reported in parentheses. *** indicates significance at the 1% level.

. The estimation results show that digital rural construction still has a consistent impact on the two land-related biased technological progress indices. This suggests that the main conclusions of this study do not depend on a specific model estimation method.

Third, this study adds additional control variables to further reduce potential omitted variable bias. On the basis of the baseline model, this study further controls for urbanization rate, regional economic development level, natural disaster impact, and government fiscal support for agriculture. The urbanization rate is measured by the proportion of the urban resident population in the total resident population at the end of the year, reflecting the possible impact of non-agricultural labor transfer on agricultural labor efficiency. Regional economic development is measured by real per capita GDP, which controls for the influence of regional economic differences on agricultural factor input structure and technological choice. Natural disaster impact is measured by the ratio of disaster-affected area to total sown area, reflecting the influence of external shocks on agricultural production. Government fiscal support for agriculture is measured by the share of agricultural fiscal expenditure in total fiscal expenditure, capturing the role of public support in agricultural production and technology adoption. The results are reported in columns (3)–(4) of Table 5. After adding these control variables, the effects of digital rural construction on the capital–land and labor–land biased technological progress indices remain stable.

Overall, the robustness tests show that the main conclusions of this study are reliable. Whether replacing the measurement of digital rural construction, changing the estimation method, or adding additional control variables, digital rural construction continues to affect agricultural technological choice under land constraints in a stable manner. This provides further support for the argument that digital rural construction reshapes the relationship between capital, labor, and land in agricultural production, thereby influencing the direction of agricultural technological progress.

4.3. Mechanism Analysis

Table 6. Mechanism Test of the Impact of Digital Rural Construction on Biased Agricultural Technological Progress.

	(1)	(2)	(3)	(4)
	Capital–Land Biased Technological Progress Index		Labor–Land Biased Technological Progress Index
VARIABLES	Scale Effect	Price Effect	Scale Effect	Price Effect
Digital	−0.635 ***	1.933 ***	1.101 ***	0.490
	(0.176)	(0.239)	(0.407)	(0.456)
Observations	392	392	392	392
R-squared	0.980	0.834	0.879	0.866
Controls	YES	YES	YES	YES
Province FE	YES	YES	YES	YES
Year FE	YES	YES	YES	YES

Note: Standard errors are reported in parentheses. *** indicates significance at the 1% level.

reports the mechanism test results. Based on the baseline findings, this study further examines how digital rural construction shapes agricultural technological choice under land constraints. Two channels are considered: the scale effect and the price effect. The scale effect is measured by the ratio of capital input to labor input, reflecting changes in factor input structure and production organization, while the price effect is measured by the ratio of capital price to labor price, capturing the role of relative factor prices in technological choice. The results indicate that digital rural construction influences agricultural technological choice through differentiated mechanisms in the capital–land and labor–land dimensions.

For the capital–land biased technological progress index, the coefficient of the scale effect is significantly negative (−0.635), whereas the coefficient of the price effect is significantly positive (1.933). These findings suggest that the impact of digital rural construction on capital-related technological choice under land constraints is not primarily driven by changes in the quantity structure of factor inputs. Rather, its key role lies in improving the conditions under which capital is used in agricultural production. Digital rural construction reduces information costs, improves access to agricultural machinery services, expands digital production services, and lowers the threshold for adopting smart equipment and platform-based technologies. These changes improve the cost–benefit conditions of using capital-embodied technologies and enable capital to be more effectively embedded in production on existing farmland. From the perspective of agricultural technological choice under land constraints, this result is particularly important. Since cultivated land is difficult to expand, agricultural producers increasingly rely on machinery, equipment, digital platforms, irrigation systems, sensors, drones, and other capital-embodied technologies to maintain and enhance production [23]. The dominant role of the price effect indicates that digital rural construction facilitates this transformation primarily by improving the accessibility and economic feasibility of such technologies. Therefore, the contribution of digital rural construction is not simply to increase capital input, but to make capital more effective in supporting agricultural production on limited land resources.

The mechanism behind the labor–land biased technological progress index is substantially different. The coefficient of the scale effect is significantly positive (1.101), while the price effect is statistically insignificant, indicating that the labor-related pathway is mainly driven by changes in production organization rather than relative factor prices. Compared with capital, the role of labor depends more heavily on management capacity, production coordination, and the interaction between farmers, land, machinery services, and digital platforms. Digital rural construction improves information access, technical guidance, agricultural service matching, and production coordination, thereby enabling labor to participate more effectively in agricultural production and land management. Under digital transformation, labor functions not only as a direct production input but also as an important carrier of digital information, production decisions, and management activities. As farmers and agricultural operators become better connected with digital platforms, machinery services, technical support, and market information, labor can more effectively support production on existing farmland.

Overall, the mechanism results reveal that digital rural construction reshapes agricultural technological choice under land constraints through two distinct channels. In the capital–land relationship, digital technologies mainly operate by improving the use conditions and economic incentives of capital-embodied technologies. In the labor–land relationship, they mainly operate by strengthening production organization and factor coordination. These findings suggest that the contribution of digital rural construction lies in changing how capital and labor are combined with limited land resources, thereby creating new technological pathways for agricultural development under land constraints.

4.4. Heterogeneity Analysis

Table 7. Heterogeneity Results of the Impact of Digital Rural Construction on Biased Agricultural Technological Progress.

	(1)	(2)	(3)	(4)	(5)	(6)
	Capital–Land Biased Technological Progress Index			Labor–Land Biased Technological Progress Index
VARIABLES	Eastern Region	Central Region	Western Region	Eastern Region	Central Region	Western Region
Digital	1.445 ***	0.687	−0.248	0.767 ***	−0.252 *	−0.120 **
	(0.333)	(2.151)	(0.946)	(0.217)	(0.139)	(0.051)
Observations	140	112	140	140	112	140
R-squared	0.339	0.178	0.202	0.386	0.318	0.286
Controls	YES	YES	YES	YES	YES	YES
Province FE	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES

Note: Standard errors are reported in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

reports the regional heterogeneity results. Considering the substantial differences in digital infrastructure, agricultural production conditions, factor markets, and technology adoption capacity across regions, this study divides the sample into eastern, central, and western China. The results show that the impact of digital rural construction on agricultural technological choice under land constraints varies significantly across regions. For the capital–land biased technological progress index, digital rural construction has a significantly positive effect only in the eastern region, with a coefficient of 1.445, while the coefficients for the central and western regions are not statistically significant. This finding suggests that the role of digital rural construction in promoting capital-related technological choice is mainly concentrated in the eastern region. A possible explanation is that the eastern region possesses more developed digital infrastructure, more mature agricultural machinery service markets, better digital financial systems, and stronger agricultural socialized service networks. Under these conditions, smart machinery, digital platforms, precision agriculture technologies, and agricultural service systems can be more effectively integrated into agricultural production [23]. Consequently, digital technologies become an important means of overcoming land constraints, enabling agricultural producers to rely more on capital-embodied technologies and digital services to support production on existing farmland rather than on additional land inputs. The insignificant coefficients in the central and western regions do not necessarily imply that digital rural construction is ineffective. Instead, they suggest that the transformation from digital investment to capital-related technological choice remains incomplete because the adoption of capital-embodied technologies still depends on complementary conditions such as machinery service systems, financing channels, technical support, and farmers’ capacity to absorb new technologies.

The regional pattern for the labor–land biased technological progress index is substantially different. Digital rural construction exerts a significantly positive effect in the eastern region, with a coefficient of 0.767, indicating that digital technologies also improve labor-related technological choice under land constraints. Through information platforms, technical guidance systems, digital services, and agricultural production networks, labor can participate more effectively in production organization, land management, and agricultural decision-making. Notably, the eastern region is the only region where both the capital–land and labor–land effects are significantly positive, suggesting that digital rural construction has generated a relatively complete technological transformation pathway. Digital technologies simultaneously enhance the roles of capital and labor, allowing both factors to better support production on existing farmland. By contrast, the coefficients for the central and western regions are significantly negative, with values of −0.252 and −0.120, respectively. Given the definition of the index, these results indicate that technological progress in these regions is relatively more oriented toward improvements associated with land than improvements associated with labor. This may reflect the current stage of digital transformation in less-developed regions. Digital rural construction may first affect farmland management, irrigation systems, resource allocation, production planning, and agricultural infrastructure, thereby improving land-related production conditions directly. However, labor-related technological adjustment often requires stronger organizational capacity, digital literacy, agricultural service networks, and human capital accumulation. Since these complementary conditions remain relatively weak in many central and western regions, the improvement of labor-related productivity may lag behind the improvement of land-related production conditions. Moreover, labor outmigration is more common in these regions, which may lead digital technologies to be used primarily to compensate for labor shortages and maintain agricultural production rather than to enhance labor productivity itself.

Taken together, the heterogeneity results reveal that the influence of digital rural construction on agricultural technological choice under land constraints is highly dependent on regional development conditions. In the eastern region, digital technologies have become deeply embedded in agricultural production systems and promote both capital-related and labor-related technological transformation. As a result, digital rural construction provides a relatively complete pathway through which non-land factors can support agricultural production under land constraints. In contrast, the central and western regions have not yet formed such a pathway. Although digital rural construction contributes to agricultural modernization, its effects are more concentrated in land-related production improvements, while the supporting roles of capital and labor remain relatively limited.

These findings further suggest that the effectiveness of digital rural construction depends not only on digital investment itself but also on the broader institutional and economic environment in which digital technologies are embedded. The ability of digital technologies to reshape agricultural technological choice under land constraints requires complementary conditions such as agricultural service systems, digital skills, production organization, financing channels, and technology adoption capacity. Therefore, regional differences in these supporting conditions largely determine whether digital rural construction can be transformed into sustained technological change in agricultural production. This also explains why digital rural construction generates significantly different technological responses across regions despite sharing the same policy background.

5. Discussion

The empirical results show that digital rural construction is not only related to agricultural productivity improvement, but also to the direction of agricultural technological change under land constraints. Existing studies have mainly examined the effects of rural digitalization on agricultural output, farmers’ income, green efficiency, and rural development [4,5,6,13,14]. This paper extends this literature by showing that digital rural construction changes how capital and labor support agricultural production on existing farmland. In this sense, digitalization does not merely introduce new tools into agriculture; it reshapes the relationship among capital, labor, and land.

The stronger effect on capital–land biased technological progress indicates that digital rural construction is more closely associated with capital-embodied and digitally enabled technologies, such as smart machinery, digital platforms, sensors, drones, irrigation systems, and machinery service platforms. Under land constraints, these technologies help agricultural producers maintain or improve production capacity without relying on additional land expansion. The positive but weaker effect on labor–land biased technological progress suggests that labor also benefits from digital transformation, but this effect may depend more on farmers’ digital skills, production experience, service accessibility, and organizational capacity.

The mechanism results further show that digital rural construction affects agricultural technological choice through different channels. The capital–land effect is mainly driven by the price effect, suggesting that digital rural construction improves the accessibility and economic feasibility of capital-embodied technologies by reducing information costs, service matching costs, and financing constraints. By contrast, the labor–land effect is mainly driven by the scale effect, indicating that the labor-related pathway depends more on production organization and factor coordination. Digital platforms, technical guidance, and agricultural service systems can improve the way labor is connected with land, machinery services, and production information.

The regional heterogeneity results suggest that the effect of digital rural construction is more pronounced in eastern China, where digital infrastructure, agricultural service systems, machinery service markets, and technology adoption capacity are relatively stronger. In central and western regions, the effects are weaker and more differentiated. This indicates that digital rural construction alone may not automatically lead to agricultural technological transformation. Its effectiveness depends on complementary conditions, including digital infrastructure, agricultural service networks, financing channels, farmers’ digital literacy, and local production organization.

6. Conclusions and Implications

6.1. Conclusions

This study examines the impact of digital rural construction on agricultural technological choice under land constraints. In the context of limited cultivated land and increasing pressure on agricultural security, agricultural development needs to rely less on land expansion and more on the effective use of capital, labor, and digital technologies on existing farmland. Using provincial panel data from China, this study investigates how digital rural construction affects capital–land and labor–land biased technological progress. The results show that digital rural construction significantly promotes both capital–land and labor–land biased technological progress, with a stronger effect on the capital–land dimension. This indicates that digital rural construction mainly reshapes agricultural technological choice through capital-related channels, such as smart equipment, machinery services, digital platforms, and other capital-embodied technologies. The positive labor–land effect also suggests that digital rural construction improves the role of labor in agricultural production and land management, although this effect is relatively weaker.

The mechanism analysis shows that digital rural construction affects the two dimensions through different pathways. In the capital–land relationship, the price effect is dominant, suggesting that digital rural construction promotes capital-related technological choice by improving the accessibility and economic feasibility of capital-embodied technologies. In the labor–land relationship, the scale effect is the main channel, indicating that digital rural construction affects labor-related technological choice mainly through production organization, information access, and factor coordination. The heterogeneity analysis further shows that the effect of digital rural construction varies across regions. In the eastern region, digital rural construction significantly promotes both capital–land and labor–land biased technological progress, indicating a relatively complete technological transformation pathway under land constraints. In the central and western regions, the effects are weaker and less balanced, suggesting that the transformation of digital construction into agricultural technological change still depends on agricultural service systems, technology adoption capacity, farmer skills, and production organization.

Overall, the contribution of digital rural construction lies not in expanding land resources, but in reshaping agricultural technological choice under land constraints. By changing how capital and labor interact with land through digital technologies, digital rural construction provides a new pathway for agricultural systems to respond to land scarcity, maintain production capacity, and support long-term land sustainability.

6.2. Policy Implications

First, digital rural construction should be integrated into land-use and farmland management strategies. The results indicate that digital technologies influence agricultural technological choice by changing how production factors interact with land. Therefore, digital rural construction should not be treated only as a digital infrastructure program. More attention should be given to the application of digital technologies in farmland monitoring, soil management, irrigation scheduling, crop growth assessment, land-use planning, and agricultural risk management. These applications can help digital rural construction become more closely connected with agricultural production on existing farmland.

Second, policy should facilitate the adoption of capital-embodied digital technologies used on existing farmland. The mechanism analysis shows that the capital-related pathway mainly operates through the price effect. Therefore, policy support should focus on reducing the cost and threshold of using machinery services, smart equipment, precision agriculture technologies, and digital services. Agricultural machinery service platforms, drone operation services, smart irrigation services, soil testing services, digital farm management systems, and precision farming technologies can be supported through service subsidies, technology vouchers, leasing systems, and digital finance. This can help capital-embodied technologies better support agricultural production under land constraints.

Third, digital rural construction should be combined with innovations in land management and agricultural organization. The labor-related mechanism indicates that the effectiveness of digital technologies depends heavily on production coordination and organizational capacity. Therefore, policies should not rely only on individual farmer training, but should also strengthen cooperatives, family farms, agricultural service organizations, and land trusteeship arrangements. These organizations can help small farmers connect with digital platforms, machinery services, technical guidance, and market information, thereby improving the use and management of cultivated land.

Fourth, regional land conditions should be fully considered when designing digital rural policies. In the eastern region, policy should focus on deepening the integration of digital technologies with precision agriculture, smart equipment, data-driven farmland management, and high-quality agricultural services. In the central region, policy should strengthen machinery service networks, agricultural service platforms, and the connection between digital tools and field production. In the western region, digital rural construction should place more emphasis on low-cost digital tools, land monitoring, water-saving irrigation, disaster warning, ecological protection, and basic technical services. Differentiated regional strategies can help digital rural construction better respond to local land endowments and production conditions.

In summary, digital rural construction should be viewed not only as rural digitalization, but also as a technological response to agricultural development under land constraints. Future policies should focus on embedding digital technologies into land-use practices, reducing the use cost of capital-embodied technologies, improving agricultural organization around land management, and designing differentiated regional strategies.

6.3. Limitations and Future Research

This study has several limitations. First, the analysis is based on provincial panel data, which can reveal the overall relationship between digital rural construction and agricultural technological choice, but cannot fully capture micro-level differences among farmers, farms, and plots. Future research could use household survey data, farm-level data, or plot-level data to examine how digital technologies affect agricultural technological choice at a more detailed level.

Second, digital rural construction is measured by a composite index. Although this approach reflects the overall level of rural digital development, it cannot distinguish the separate effects of specific digital technologies, such as digital finance, e-commerce platforms, smart machinery, drones, sensors, remote sensing, and smart irrigation systems. Future studies could further examine which types of digital technologies play a stronger role in shaping capital–land and labor–land biased technological progress.

Third, this study focuses on China, where land constraints, agricultural organization, and rural digital transformation have specific institutional backgrounds. The applicability of the findings to other countries may depend on differences in land institutions, agricultural service systems, digital infrastructure, and rural development conditions. Future comparative studies could further explore this issue under different land systems and development contexts.