How new quality productivity shapes agricultural carbon emissions in China: the masking effect of agricultural mechanization

Agricultural low-carbonization, serving as a critical pathway for achieving the United Nations Sustainable Development Goals (SDGs), extends beyond being a mere technological challenge to fundamentally encompass both the paradigm shift in developmental philosophy and the systemic synergy of institutional frameworks. This study explores how new quality productivity influences agricultural carbon emissions, utilizing Chinese provincial panel data (2012–2022) and a Two-way Fixed Effects model to analyze its impacts through sustainability-oriented lenses. Empirical results demonstrate that new quality productivity exerts a significant reducing effect on agricultural carbon emissions, with each unit increase correlating to a 20.2% reduction. Mechanistic analysis reveals a dual pathway: facility agricultural technology promotion, which drives a 2.8% indirect emission reduction through resource-efficient and environmentally friendly production systems aligned with sustainable development goals. However, this positive effect is partially offset by a “masking effect” from agricultural mechanization, stemming from its fossil fuel dependence that contradicts low-carbon transition objectives. Regional heterogeneity analysis highlights pronounced disparities in sustainability outcomes: the western region benefits most from the emission reduction effects, reflecting adaptive capacity to low-carbon technologies, while the eastern region shows an unexpected promoting effect, attributed to energy path dependencies in intensive production models that hinder sustainable transformation. Production process analysis further indicates that new quality productivity effectively reduces emissions in high-carbon input stages such as fertilizer and pesticide use, but its impact is weaker in Tillage applications, pointing to unaddressed sustainability challenges in these areas Building on these findings, this study underscores that realizing the full decarbonization potential of new quality productivity requires navigating constraints posed by technical trajectories, regional resource endowments, and production process characteristics within a sustainable development framework. It not only enriches the carbon reduction mechanisms of new quality productivity in agriculture, providing clear policy targets for agricultural decarbonization policy-making, but also offers theoretical insights and practical experience for developing countries to achieve agricultural sustainability.

1 Introduction

Agricultural low-carbon transition, as a critical pathway to address the dual challenges of global climate change and food security, holds strategic significance for the synergistic advancement of SDG 2 (Zero Hunger) and SDG 13 (Climate Action). Currently, the extensive development models adopted by most countries have exacerbated resource constraints and environmental pressures (Liu et al., 2025). Data from the Food and Agriculture Organization (FAO) of the United Nations indicates that greenhouse gas emissions from the agricultural and food systems account for 31% of global total emissions, and between 2000 and 2022, such emissions from the agricultural and food systems increased by 10%. This data underscores the dual shortcomings of traditional agricultural models in meeting food demand while controlling carbon emissions.

At present, achieving decoupling between economic growth and resource consumption or environmental pollution is crucial for a country’s green ecological transformation (Liu et al., 2023; Wang et al., 2024). As the world’s largest agricultural country, China’s total carbon emissions from agricultural production in 2022 stood at 828 million tons of carbon dioxide equivalent. At the national level, China’s agricultural carbon emissions have always been higher than those of European and American countries. Nevertheless, China’s agricultural carbon emissions have decreased by 4.53% compared to 2013, with an average annual decline of 0.52%. The intensity of agricultural carbon emissions has also decreased by 35.29% compared to 2013, with an average annual decline of 4.72%, which is already lower than that of many developed countries. These data not only reflect the current situation of China’s agricultural carbon emissions and the progress in emission reduction efficiency but also reveal the urgency of continuous emission reduction.

Against this backdrop, the Chinese government has established high-quality development of productivity as the core development orientation, aiming to promote the achievement of carbon peak and carbon neutrality goals through high-quality development. The concept of New Quality Productivity (NQP) has thus been proposed and received widespread attention. As a new type of productivity driven by scientific and technological innovation, which emphasizes the improvement of resource allocation efficiency and the optimization of energy structure, its green effect has become a research hotspot in academic circles. Current studies have fully confirmed that New Quality Productivity plays a role in the dimensions of technological progress such as digital upgrading of the secondary industry and green technological innovation, especially in the manufacturing and construction industries, where the carbon reduction effect brought by technological progress has become increasingly prominent (Wang et al., 2025a; Liu and He, 2024). Moreover, the catalytic interaction between green finance and New Quality Productivity has further accelerated the green transformation of industries (Wang and Ling, 2024). Then, does the green characteristic of New Quality Productivity extend to the agricultural field and promote the sustainable development of agriculture? What mechanisms does it rely on to promote sustainable agricultural development? What functions do New Quality Productivity play in different agricultural production links? Looking at the existing literature, these issues have not been well answered, and there is a lack of in-depth theoretical exploration and empirical testing on the carbon reduction effect of New Quality Productivity. A few articles explore the path of New Quality Productivity on agricultural carbon reduction from the perspectives of generalized technological progress and productivity improvement. However, these research variables are often generalized. Even if they can explain the mechanism effect at the empirical level, they are difficult to be policy support at the practical level. The black box of some indicator measurements also makes it impossible to observe the specific technical path of carbon reduction, leading to the disconnection between research and reality. In addition, the lack of detailed discussion on regional differences makes it difficult to form targeted policy recommendations. In terms of agricultural carbon emissions, existing studies mostly focus on the direct effects of factors such as the digital economy, environmental regulations, and green total factor productivity on agricultural carbon reduction. Some also explore the impact of new productive forces on agricultural carbon emissions from the perspective of spatial spillover effects (Wang et al., 2025), But systematic investigation into technical pathways remains absent. This deficiency prevents precise identification of how generalized variables specifically influence agricultural decarbonization.

To fill these research gaps, this study will empirically test the impact of New Quality Productivity on agricultural carbon emissions based on panel data of Chinese provinces by using two-way fixed effects (TWFE) model, and analyze the mechanism from two key paths: facility agriculture and agricultural mechanization. By deeply analyzing the relationship between New Quality Productivity and carbon emissions in various agricultural links, this study will reveal the internal mechanism of agricultural low-carbon transformation, which is of great practical significance for optimizing agricultural carbon reduction policies and realizing sustainable agricultural development. Meanwhile, in the global emission reduction action, China’s practical experience and strategies are of great reference significance for other developing countries, which is helpful to promote global agricultural low-carbon development and climate change response actions.

2 Literature review and research hypotheses

2.1 New quality productivity and agriculture carbon emissions

New Quality Productivity (NQP) emphasizes enhancing agricultural production efficiency through technological innovation, digitalization, and green transformation while mitigating ecological impacts. Its carbon reduction effects primarily manifest in optimizing resource allocation efficiency and reducing carbon emissions across production processes. From a political economy perspective, (Luo,2024) argues that NQP restructures agricultural production models by integrating data elements with traditional agriculture, thereby reducing reliance on high-carbon inputs such as fertilizers and pesticides (Luo and Geng, 2024). Ma and Yang (2024), analyzing traditional and emerging factor allocation, demonstrate that NQP fosters innovative resource configurations, drives integrated development, and elevates production capabilities to higher value chain levels, enabling precision, intelligence, and green transformation in agriculture. Chen et al. (2025) construct an NQP evaluation framework for agriculture, revealing that its carbon reduction effects are mediated by improvements in labor and land productivity (Chen et al., 2025). Collectively, these studies underscore NQP’s inherent alignment with green productivity principles.

As a critical component of NQP, digital productivity inherently embodies low-carbon characteristics. Its technical diffusion and application spillovers facilitate effective emission reductions in agricultural practices. (Huang et al., 2024) empirically confirms that rural digitalization strengthens the inhibitory effect on agricultural carbon emissions (Huang et al., 2024). Han and Gong (2025) further reveal that digital rural construction significantly reduces agricultural carbon emissions through direct negative effects and spatial spillover mechanisms. These findings suggest that advancing NQP accelerates digital innovation and dissemination, driving profound transformations in agricultural production paradigms.

Nevertheless, international research presents nuanced perspectives on technology’s role in agricultural decarbonization. The IPCC Fifth Assessment Report cautions that while agricultural technological advancements enhance crop yields and efficiency, they may concurrently increase emissions through intensified fertilizer use and mechanization. Chen et al. (2020) distinguishes between environmental technologies (reducing emissions) and production technologies (potentially increasing emissions). Guan et al. (2023) identifies that agricultural mechanization, despite boosting productivity, may elevate energy consumption and associated emissions, corroborating the energy rebound effect hypothesis - efficiency gains paradoxically increasing total energy use. Regional disparities in technology adoption further complicate emission patterns, with developed regions adopting low-carbon technologies faster than carbon-intensive models persist in underdeveloped areas, exacerbating interregional emission differentials (He and Ding, 2022; Milindi and Inglesi-Lotz, 2022).

However, NQP’s endogenous green attributes may counterbalance these risks through three mechanisms: (1) substituting traditional factors with clean alternatives in production configurations, (2) institutional coordination to regulate high-emission agricultural processes, and (3) reshaping energy flow pathways within agricultural ecosystems. From the above research, it can be observed that existing studies have not directly explored the specific impact of the core concept of New Quality Productivity (NQP) on agricultural carbon emissions, lacking targeted analysis of its inherent green attributes. Meanwhile, the adopted holistic measurement methods obscure the carbon emission boundaries of subsystems such as pesticides, chemical fertilizers, and mechanical diesel, and fail to deconstruct carbon emissions across different agricultural production links. This makes it difficult to reveal the differentiated effects of NQP in various production links. Therefore, this study will conduct a heterogeneous discussion on carbon emissions from different agricultural production links to explain the carbon reduction impact of NQP on each production processes, and thus propose the first hypothesis of this paper.

H1: New Quality Productivity exerts significant carbon reduction effects by effectively inhibiting agricultural carbon emissions.

H1a: The carbon reduction effect of New Quality Productivity varies across regions with different geographical conditions.

H1b: The carbon reduction effect of New Quality Productivity also differs across different agricultural production processes.

2.2 New quality productivity and facility agriculture, agricultural carbon emissions

Facility agriculture, characterized by high-efficiency agricultural production within controlled environments, demonstrates strong risk resistance, intensive resource inputs, and technology-intensive operations (Wang et al., 2025b; Zhang et al., 2024a). Empirical studies reveal its complex carbon dynamics: Tao (2017) identified that excessive fertilizer application in facility vegetable cultivation significantly increases short-term carbon emissions, while Yin et al. (2021) emphasized the critical role of technological investments and resource utilization efficiency in modulating emission patterns. Contrastingly, Tong et al. (2024) demonstrated that facility agriculture development substantially enhances agricultural carbon emission efficiency, with accelerated growth rates correlating to improved decarbonization performance.

In the context of green development and agricultural modernization, New Quality Productivity (NQP) aligns with facility agriculture’s technical demands through its dual emphasis on technological innovation and production paradigm shifts. NQP addresses inherent limitations in facility agriculture, including suboptimal decarbonization efficiency and transient carbon surge effects, by providing systematic technological frameworks and theoretical guidance (Lin et al., 2024; Luo et al., 2025). Gao et al. (2022) further argue that conventional agricultural practices, often reliant on empirical farmer knowledge, lack precision in input allocation. NQP-driven digital transformation enables precise input management in facility agriculture, fostering intensive production models that reduce carbon intensity.

Collectively, these findings suggest that facility agriculture’s carbon emissions exhibit dynamic evolutionary patterns. NQP facilitates fundamental technological transitions through breakthroughs in precision control systems and energy infrastructure reconfiguration, thereby extending operational cycles and optimizing technical frameworks. This synergy amplifies the green potential of facility agriculture, progressively unlocking its decarbonization capacity. However, Existing studies have not clarified how NQP transforms the technical foundation of facility agriculture through breakthroughs in precision control technologies and the restructuring of energy systems. Nor have they empirically tested facility agriculture as an intermediary pathway for carbon reduction driven by NQP, which leads to an unclear explanation of the synergistic logic between the carbon reduction potential of facility agriculture and NQP. Consequently, we propose the second hypothesis:

H2: New Quality Productivity can reduce agricultural carbon emissions by promoting the development of facility agriculture.

2.3 New quality productivity and agricultural mechanization, agricultural carbon emissions

In the past, agricultural mechanization usually depended on fossil fuel-driven machinery, such as tractors and harvesters. The promotion of agricultural mechanization itself is a process that increases carbon emissions. Liu and Zhou (2018) found that in regions where the energy structure is dominated by fossil fuels, the energy consumption of agricultural machinery leads to increased carbon emissions, and the carbon footprint of mechanization is highly dependent on the degree of energy cleanliness and the improvement of mechanical efficiency. Agricultural mechanization and large-scale operations promote increased agricultural carbon emissions, and there is a mutual interaction between them. The carbon emissions generated by agricultural mechanization itself would, in turn, promote mechanization technology progress (Guan et al., 2023). Technological innovation can optimize the utilisation of agricultural machinery, increase crop yields, reduce carbon emissions, and promote the integration of food security and low-carbon economic development (Li and Zhang, 2021). Ma et al. (2022) found that promoting advanced seeding techniques and machinery suitable for mountainous areas can help reduce reliance on chemical fertilizers and subsequently reduce carbon emissions. Zhang et al. (2024b) discovered that in Belt and Road Initiative countries, agricultural mechanization can significantly reduce carbon emissions, and technological progress can enhance this effect. Yang et al. (2022) conducted research on green total factor productivity (GTFP) and found that an increase in the level of agricultural mechanization is positively correlated with GTFP growth, thereby indirectly reducing carbon emissions. However, the negative impact of agricultural mechanization on agricultural carbon reduction currently outweighs the positive impact of environmental regulation on agricultural carbon reduction (Jia and Xu, 2024). Gao et al. (2024) and Guo et al. (2020) identified the dual impact of agricultural mechanization on carbon reduction through electrification and intelligence. Empirical research shows that the carbon emissions of electric farm machinery over its entire life cycle are 62% lower than those of diesel machines.78%, with the electrification of the power system contributing 58% and the precision of the work achieved through the reduction of the use of fertilizers and pesticides contributing 24%.

In summary, existing research reveals that there is a complex interplay between agricultural mechanization and carbon emissions. Most studies indicate that the high carbon emissions associated with mechanization primarily stem from inefficiencies in the energy conversion of agricultural machinery. However, the transition to clean energy sources and technological efficiency improvements driven by new quality productivity can transform the energy sources of agricultural machinery. The continuous development of electrified and intelligent agricultural machinery is breaking through the limitations of traditional energy sources and reshaping the carbon emission trajectory of agricultural mechanization. Therefore, the third hypothesis is proposed.

H3: New Quality Productivity can reduce agricultural carbon emissions by promoting agricultural mechanization.

In conclusion, this study constructs a theoretical framework that presents the hypothesis and pathway analysis of how New Quality Productivity (NQP) reduces agricultural carbon emissions, as illustrated in Figure 1.

Figure 1

Figure 1. Theoretical analysis of new quality productivity to reduce carbon emissions from agriculture.

3 Research design

3.1 Data sources

The research scope of this study encompasses 31 provincial-level administrative regions in China (excluding Hong Kong, Macao, and Taiwan) from 2010 to 2022. Data were primarily sourced from authoritative publications including the China Rural Statistical Yearbook, China Statistical Yearbook on Culture and Related Industries, China Agricultural Yearbook, provincial statistical yearbooks, the National Bureau of Statistics (NBS) official website, and the EPS database.

To ensure data comprehensiveness and accuracy, systematic processing was implemented for missing values and outliers in specific years. For unavailable data points in provincial statistical yearbooks, interpolation methods based on adjacent years were employed for data imputation. Additional methodological details regarding data processing procedures are explicitly outlined in subsequent sections of this paper.

3.2 Explanatory variable

The explanatory variable in this study is New Quality Productivity (NQP). Drawing on the measurement framework proposed by Lu et al. (2024), we employ the entropy method to construct a comprehensive evaluation system encompassing three dimensions: technological productivity, green productivity, and digital productivity. Within the secondary indicators, resource-saving and environmentally friendly productivity exhibits a close association with agricultural carbon emissions, while also being influenced by technological advancement and digitalization levels. The enhancement of NQP optimizes agricultural production patterns, improves resource utilization efficiency, and reduces carbon emissions, thereby fostering sustainable agricultural development. This measurement framework aligns with the intrinsic attributes of NQP, enabling a systematic exploration of its impact mechanisms on agricultural carbon emissions. Specific secondary indicators and their definitions are detailed in Table 1.

Table 1

Table 1. Measurement of new quality productivity.

3.2.1 Data standardization

Given the heterogeneity in measurement units, dimensions, and magnitudes across indicators, dimensionless processing is essential to ensure objective weighting. To align with the entropy method’s requirement for positive values, we adopt the min-max normalization technique, transforming data into the [0, 1] interval (Equation 1):

$\begin{array}{ll}{x}_{\mathit{ij}}^{\ast }=\frac{x_{\mathit{ij}}-m_j}{M_j-m_j}& (1)\end{array}$

where

${\mathrm{M}}_{\mathrm{j}}=max\left\{{\mathrm{x}}_{\mathrm{ij}}\right\},{\mathrm{m}}_{\mathrm{j}}=min\left\{{\mathrm{x}}_{\mathrm{ij}}\right\}.$

3.2.2 Entropy weight calculation

The entropy method is particularly suitable for comprehensive evaluation involving multiple dimensions. This is mainly reflected in its ability to assign weights objectively, integrate multi-dimensional indicators, and comprehensively capture the connotation of new quality productivity. Meanwhile, it is well-suited for heterogeneous data. Therefore, the entropy method is adopted in this paper to construct the variable of new quality productivity.

The entropy method is applied to determine indicator weights through the following steps:

Step 1: Feature Proportion Calculation.

Compute the proportional contribution of each sample (Equation 2):

$\begin{array}{ll}{p}_{\mathit{ij}}=\frac{x_{\mathit{ij}}}{{\sum }_{i=1}^m{x}_{\mathit{ij}}}& (2)\end{array}$

where m denotes the number of samples, and $p_{\mathit{ij}}$ represents the feature proportion of the i-th sample under the j-th indicator.

Step 2: Entropy Value Determination.

Calculate the entropy value $e_j$ for each indicator (Equation 3):

$\begin{array}{c}{e}_j=\frac{1}{ln\left({\sum }_{i=1}^m{p}_{\mathit{ij}}\right)}ln\frac{1}{p_{\mathit{ij}}}\end{array}\text{ (3)}$

where ${\mathrm{I}}_{\mathrm{ij}}=ln\frac{1}{{\mathrm{p}}_{\mathrm{ij}}}$ denotes the information quantity, ${\mathrm{I}}_{\mathrm{j}}=\sum _{\mathrm{i}=1}^{\mathrm{m}}{\mathrm{p}}_{\mathrm{ij}}ln\frac{1}{{\mathrm{p}}_{\mathrm{ij}}}$ represents the total information quantity, and ${\mathrm{e}}_{\mathrm{j}}$ is the entropy value.

Step 3: Calculation of Weights.

Differentiation Coefficient (Equation 4):

$\begin{array}{c}{d}_j=1-e_j\end{array}\text{ (4)}$

Weight Assignment (Equation 5):

$\begin{array}{c}{w}_j=\frac{d_j}{{\sum }_{j=1}^n{d}_j}\end{array}\text{ (5)}$

where a higher weight indicates greater contribution of the indicator to New Quality Productivity (NQP).

Step 4: Calculation of New Quality Productivity (NQP) (Equation 6)

$\begin{array}{c}{\mathrm{NQP}}_i={\sum }_{j=1}^n(w_j·z_{\mathit{ij}})\end{array}\text{ (6)}$

${\mathrm{NQP}}_i$ ∈[0,1], with higher values reflecting superior New Quality Productivity levels.

3.3 The dependent variable

The dependent variable in this study is agricultural carbon emissions. The measurement of carbon emissions primarily follows the methodology established by Fang et al. (2024). On this basis, draw on the life cycle carbon emission assessment method of Nian et al. (2025), and the agricultural carbon emission was divided into different agricultural production processes for calculation. The specific calculation formula and reference indicators are as follows (Equation 7):

$\begin{array}{c}\mathit{\text{TotalCarbonEmissions}}={\sum }_{i=1}^n(w_i\times e_i)\end{array}\text{ (7)}$

wherein: wi is the weight of the i-th activity, ei is the carbon emission of the i-th activity, and n is the number of different agricultural activities, as shown in Table 2.

Table 2

Table 2. Carbon emission calculation.

3.4 Control variables

To enhance the scientific rigor and accuracy of the research findings, this study incorporates multidimensional control variables that align with the comprehensive nature of New Quality Productivity (NQP). The selected controls are categorized into three hierarchical dimensions to mitigate estimation bias:

1. Rural-level Controls

• Rural income growth rate (current/previous per capita net income - 1)

• Rural Engel coefficient (total food expenditure/total expenditure amount)

2. Provincial-level Controls

• Transport infrastructure (road mileage/year-end population)

• Tax burden (tax revenue/GDP)

• Urbanization rate (registered population in urban areas/total population)

• Human capital (enrollment in regular institutions of higher learning/registered population)

3. Macro-level Government Controls

• Openness to the outside world (total import–export volume/GDP)

• Government intervention (government expenditure/GDP)

The inclusion of these multidimensional controls enhances the model’s capacity to isolate NQP’s independent effect on agricultural carbon emissions by mitigating confounding factors. Furthermore, the hierarchical integration of controls across governance levels provides actionable insights for multi-tiered policymaking.

3.5 Mechanism variables

Existing studies on the mediating mechanisms of agricultural carbon emissions have mostly focused on the reconstruction of traditional factors, such as agricultural industrial structure, generalized technological progress (Yu et al., 2025), and rural labor productivity (Chen et al., 2025). These variables suffer from vague conceptual boundaries and high measurement heterogeneity. Meanwhile, the “black-box” treatment makes it difficult to decompose the mediating mechanisms. This study focuses on the concrete carriers of new productive forces, taking agricultural mechanization (measured by per capita total power of agricultural machinery) and protected agriculture (measured by the level of protected agriculture) as mechanism variables.

From the perspective of actual production, the level of protected agriculture is directly related to resource use efficiency. For instance, precision agricultural facilities can reduce excessive fertilizer application, thereby lowering carbon emissions. The level of agricultural mechanization directly reflects the low-carbon upgrading of production tools; for example, the application of electric agricultural machinery can reduce carbon emissions caused by diesel consumption during production. As mediating variables, both can transform the green attributes of new productive forces into observable and quantifiable production practices, making the mediating transmission chain more specific and verifiable.

This design not only avoids the “generalization” defect of traditional mediating variables but also converts the abstract concept of new productive forces into observable and intervenable real-world carriers, enhancing the academic increment and practical value of the research. At the data processing level, to address the issue of missing observations for some mediating variables in 2021–2022, the study adopts linear interpolation and moving average methods for scientific extrapolation. This ensures the temporal continuity of the data while minimizing the interference of estimation bias on causal inference to the greatest extent possible.

3.6 Model setting

3.6.1 Construction of the baseline regression model

Since the data is panel data with two dimensions of year and province, and different provinces have different levels of economic development and resource endowments, they are also affected by macroeconomic fluctuations and policy cycles at the time level. If the inherent characteristic variables of these provinces and the common trends over time are not controlled, error terms may be mixed in, causing the measurement coefficients to deviate from the true values and making it impossible to accurately measure the true impact. Therefore, in order to ensure the reliability of the experimental results, this paper adopts a two-way fixed effects (TWFE) model for model construction. The specific mathematical model and symbol meanings are as follows:

$\begin{array}{c}ln\left({\mathrm{ACE}}_{\mathit{it}}\right)={\alpha }_0+{\alpha }_1{\mathrm{NQP}}_{\mathit{it}}+\beta {\mathrm{X}}_{\mathit{it}}+{\mu }_i+{\lambda }_t+{\in }_{\mathit{it}}\end{array}\text{ (8)}$

To linearize relationships, stabilize variance, or mitigate the impact of heteroscedasticity, logarithmic transformation is employed. In Equation 8, $ln\left({\mathrm{ACE}}_{\mathit{it}}\right)$ represents the natural logarithm of agricultural carbon emissions in province i at time t. ${\mathrm{\alpha }}_0$ denotes the intercept term. ${\mathrm{\alpha }}_1$ is the coefficient of NQP, indicating the marginal effect of NQP on agricultural carbon emissions. ${\mathrm{NQP}}_{\mathit{it}}$ represents the level of NQP in province i at time t. β is the vector of coefficients for the control variables (X), reflecting the impact of these variables on agricultural carbon emissions. ${\mathrm{X}}_{\mathit{it}}$ denotes the control variables, which may include other factors affecting agricultural carbon emissions, such as economic development level and policy support. ${\mathrm{\mu }}_i$ represents the individual fixed effect, capturing the specific effects of province i that do not change over time, controlling for individual characteristics that are invariant over time. ${\mathrm{\lambda }}_t$ is the time fixed effect, representing the specific effects at time t, controlling for time trends that do not vary across individuals. ${ϵ}_{\mathit{it}}$ is Idiosyncratic error term, representing the unexplained portion of the model.

3.6.2 Construction of the mediation model

In mediating effect analysis, to examine the extent to which mediating variables exert their mediating effects, this paper adopts a two-way fixed effects model to generate empirical results, consistent with the setup of the earlier baseline regression. It also employs the causal steps approach for mediation analysis, which involves sequential regression to test, in turn, the effect of the independent variable on the mediating variable and the joint effect of the mediating variable and independent variable on the dependent variable, thereby determining the presence of a mediating effect. The specific mathematical models and symbol meanings are as follows:

First, we test the effect of new quality productivity (NQP) on the mediating variable of facility agriculture:

$\begin{array}{c}\mathit{Ln}\left(\mathit{\text{Facilit}}{y}_{\mathit{it}}\right)={\gamma }_0+{\gamma }_1{\mathrm{NQP}}_{\mathit{it}}+\delta {\mathrm{X}}_{\mathit{it}}+{\mu }_i+{\lambda }_t+{\in }_{\mathit{it}}\end{array}\text{ (9)}$

Second, after controlling for NQP, we test the effect of facility agriculture (the mediating variable) on agricultural carbon emissions (LnACE):

$\begin{array}{c}\begin{array}{l}\mathit{Ln}\left(\mathit{AC}{E}_{\mathit{it}}\right)={\theta }_0+{\theta }_1\mathit{NQ}{P}_{\mathit{it}}+{\theta }_2\mathit{Ln}\left(\mathit{\text{Facilit}}{y}_{\mathit{it}}\right)\\ +\eta X_{\mathit{it}}+{\mu }_i+{\lambda }_t+{\in }_{\mathit{it}}\end{array}\end{array}\text{ (10)}$

In Equation 9, ${\text{Facility}}_{\mathit{it}}$ denotes the level of facility agriculture development in province i in year t; ${\gamma }_0$ represents the intercept term, i.e., the baseline level of facility agriculture when all explanatory variables are zero; ${\gamma }_1$ indicates the change in facility agriculture level corresponding to a one-unit increase in NQP; ${\mathrm{NQP}}_{\mathit{it}}$ stands for the level of new-quality productivity in province i in year t; ${\mathrm{X}}_{\mathit{it}}$ refers to control variables, with $\delta$ representing the corresponding coefficient vector for these control variables; ${\mu }_i$ denotes provincial fixed effects, ${\lambda }_t$ denotes year fixed effects, and ${ϵ}_{\mathit{it}}$ represents the unexplained portion of the model.

In Equation 10, $\mathit{Ln}\left(\mathit{AC}{E}_{\mathit{it}}\right)$ denotes the natural logarithm of agricultural carbon emissions in province i in year t; ${\theta }_0$ represents the baseline logarithmic value of agricultural carbon emissions when all explanatory variables are zero; ${\theta }_1$ is the coefficient of the direct effect; ${\theta }_2$ is the coefficient of the mediating effect; and the remaining variables have the same meanings as above.

After testing the mediating role of facility agriculture, we further examine the mediating effect of agricultural mechanization in the impact of new-quality productivity on agricultural carbon emissions. Following the same logic, we first test the effect of new-quality productivity (NQP) on agricultural mechanization:

$\begin{array}{c}\mathit{Ln}\left(\mathit{\text{Mechanizatio}}{n}_{\mathit{it}}\right)={\kappa }_0+{\kappa }_1\mathit{NQ}{P}_{\mathit{it}}+\xi X_{\mathit{it}}+{\mu }_i+{\lambda }_t+{\in }_{\mathit{it}}\end{array}\text{ (11)}$

Second, after controlling for NQP, we test the effect of agricultural mechanization on agricultural carbon emissions (LnACE):

$\begin{array}{c}\begin{array}{l}L\mathrm{n}\left({\mathrm{ACE}}_{\mathit{it}}\right)={\rho }_0+{\rho }_1{\mathrm{NQP}}_{\mathit{it}}+{\rho }_2\mathrm{Ln}\left({\text{Mechanization}}_{\mathit{it}}\right)\\ +\nu {\mathrm{X}}_{\mathit{it}}+{\mu }_i+{\lambda }_t+{\in }_{\mathit{it}}\end{array}\end{array}\text{ (12)}$

In Equation 11, $\mathrm{Ln}\left({\text{Mechanization}}_{\mathit{it}}\right)$ denotes the level of agricultural mechanization in province i in year t; ${\kappa }_0$ is the baseline level of agricultural mechanization when all explanatory variables are zero; ${\kappa }_1$ indicates the change in agricultural mechanization level corresponding to a one-unit increase in NQP; $\mathit{NQ}{P}_{\mathit{it}}$ stands for the level of new-quality productivity in province i in year t; $X_{\mathit{it}}$ refers to control variables, with $\zeta$ representing the corresponding coefficient vector for these control variables; ${\mu }_i$ denotes provincial fixed effects, ${\lambda }_t$ denotes year fixed effects, and ${ϵ}_{\mathit{it}}$ represents the unexplained portion of the model.

In Equation 12, $L\mathrm{n}\left({\mathrm{ACE}}_{\mathit{it}}\right)$ denotes the natural logarithm of agricultural carbon emissions in province i in year t; ${\rho }_0$ represents the baseline logarithmic value of agricultural carbon emissions when all explanatory variables are zero; ${\rho }_1$ is the coefficient of the direct effect; ${\rho }_2$ is the mediating effect coefficient of agricultural mechanization; and the remaining variables have the same meanings as above.

4 Empirical test and result analysis

4.1 Descriptive statistics

Table 3 and Figure 2 depict the dynamic evolution trends of new-quality productivity and agricultural carbon emissions during 2010–2026. New-quality productivity increased continuously from an initial value of 0.16308 to a projected final value of 0.23341, with an average annual growth rate of approximately 3.5%, showing an overall upward trajectory despite periodic fluctuations: it grew steadily from 0.16308 to 0.21362 during 2010–2016, reflecting the effects of technological accumulation or initial investment; experienced a slight correction to 0.20662 in 2017–2018, possibly associated with structural adjustments or short-term policy realignments; and accelerated growth in the projection period (2022–2026), surging from 0.20504 to 0.23341, which may signal technological breakthroughs and large-scale application.

Table 3

Table 3. Descriptive statistics.

Figure 2

Figure 2. Chinese agricultural carbon emissions and new quality productivity annual averages. (The data for 2022–2026 are forecasted values, and the data of 2010 and 2011 were not included in the model due to differential processing).

Agricultural carbon emissions exhibited an inverted U-shaped curve of “first increasing, then decreasing”: rising from 3.1596 million tons in the early stage to a peak of 3.4489 million tons in 2016, likely linked to agricultural production expansion and technological path dependence; and declining continuously after 2017, with the projected value dropping to 2.3710 million tons by 2026, representing an average annual decline of approximately 3.2%. This decline is hypothesized to stem from green agricultural policies (such as emission reduction subsidies and carbon trading mechanisms), clean technology substitutions (e.g., bioenergy, precision fertilization), and the promotion of eco-agricultural models.

The relationship between the two variables shows that in the early stage (2010–2016), productivity improvements were accompanied by increased carbon emissions, reflecting the high-resource-consumption characteristics of initial technological upgrades. In the later stage (2017–2026), productivity growth decoupled from carbon emissions, demonstrating a “technology-environment” synergistic effect consistent with the Environmental Kuznets Curve hypothesis—i.e., after reaching a certain stage of economic development, green innovation and policy intervention offset environmental pressures. The accelerated decline in carbon emissions during the projection period may benefit from the implementation of carbon taxes, the popularization of digital agriculture, and the deepening of circular economy models.

Overall, the transition from a “contradictory” to a “synergistic” relationship between new quality productivity and agricultural carbon emissions highlights the core role of technological innovation and policy guidance in driving agricultural low-carbon transformation.

For panel data, due to differences in data characteristics, non-stationary data may exhibit unit root discrepancies, which can easily lead to issues like spurious regression. Therefore, we conducted panel unit root tests. The results showed that in the LLC test, most variables were stationary; however, some variables such as social consumption level displayed instability. Based on the current data characteristics and the LLC test results, we applied differencing to non-stationary variables to generate new variables: first-order differencing was performed on tax burden level, human capital level, and government intervention level, while second-order differencing was applied to social consumption level. After data processing, the panel data became stationary, indicating that it is suitable for further panel data benchmark regression analysis. The specific descriptive statistics of the processed data and the unit root test results of the processed data are provided in the Supplementary materials.

4.2 Baseline regression

Before conducting baseline regression, we first perform the variance inflation factor (VIF) test to evaluate multicollinearity among variables. The results show that the average variance inflation factor is 2.61, while the variance inflation factor of the degree of openness to the outside world reaches 7.66. The addition of this variable has brought certain collinearity risks to the model. Theoretically, the degree of openness to the outside world may have a certain intrinsic correlation with other control variables. At the same time, there may be a certain degree of variable overlap when included. Therefore, the variable of the degree of openness to the outside world was excluded in the benchmark regression. In the variance inflation factor test after exclusion, the average vif of the model was 1.74. This indicates that the corrected model does not have significant multicollinearity problems, has good stability and explanatory power, and can provide stability for subsequent regressions.

The regression results (Table 4) reveal that in the baseline Model I, New Quality Productivity (NQP) significantly reduces agricultural carbon emissions at the 5% significance level, with a one unit increase in NQP corresponding to a 0.164 unit decrease in the logarithm of agricultural carbon emissions. Upon incorporating rural level control variables—the growth rate of rural per capita income and the rural Engel coefficient—the coefficient of NQP in Model II strengthened to −0.296, indicating an enhanced carbon reduction effect. This suggests that controlling for rural income growth and consumption structure further amplifies NQP’s emission-reducing impact. Rural per capita income growth exert statistically significant negative effects on agricultural carbon emissions at the 1% level. The research by Chang et al. and Li et al. indicates that the increase in per capita income in rural areas can prompt them to adopt innovative low-carbon technologies and promote the transformation of household energy structure, thereby reducing agricultural carbon emissions (Li and Jiang, 2024; Chang et al., 2022).

Table 4

Table 4. Baseline regression analysis.

Level of Urbanization significantly increases agricultural carbon emissions at the 1% level. In Model III, a one unit rise in urbanization elevates emissions by 2.303 units. From a labor substitution perspective, urbanization-driven rural to urban migration shifts labor-intensive farming toward mechanized production, escalating energy use and emissions. Tang and Chen (2022) substantiate this mechanism, noting that below a certain threshold, rural population transfer heightens reliance on labor- and land-saving technologies, directly driving energy consumption and emissions. Tax Burden also exhibit positive and statistically significant coefficients (1.365 units in Model III), implying that higher tax intensifies agricultural carbon emissions.

Model IV indicates that the coefficient of NQP to the logarithm of agricultural carbon emissions is −0.202, which is significant at the 1% level, demonstrating the robustness of NQP in suppressing agricultural carbon emissions. Meanwhile, although the coefficient of government intervention level to agricultural carbon emissions is negative, it has no significant meaning.

4.3 Endogeneity and robustness tests

Potential bidirectional causality between New Quality Productivity (NQP) and carbon emission intensity—where changes in emissions may inversely affect NQP—raises endogeneity concerns. To address this issue, we employ an Instrumental Variable (IV) approach, utilizing lagged terms of NQP as instruments. Both Ordinary Least Squares (OLS) and Two-Stage Least Squares (2SLS) regressions are conducted, with results presented in Table 5.

Table 5

Table 5. Endogeneity test.

The endogeneity test results demonstrate that the first-order lag of New Quality Productivity (Lag1_NQP) effectively predicts contemporaneous productivity levels. In the first-stage regression, the coefficient estimates for the lagged term are 1.007 and 0.920, both statistically significant at the 1% level, indicating strong correlation between the instrument and the endogenous variable. The first-stage F-statistics (7321.26 and 2,249) substantially exceed the conventional threshold for instrument relevance (F > 10), confirming the validity of the lagged term as a robust instrumental variable.

The second-stage regression further validates the significant inhibitory effect of NQP on agricultural carbon emissions. Without controlling for covariates, a one-unit increase in NQP reduces carbon emissions by 38.2%. When control variables are included, the emission reduction effect attenuates but remains statistically significant at the 5% level, with a coefficient of −0.169. These results statistically confirm that NQP enhancement effectively drives agricultural decarbonization, and the conclusion holds robustly across alternative model specifications.

In the robustness tests, this paper employs three approaches to assess robustness. First, the regression analysis is performed by reducing the sample size and excluding observations from developed and underdeveloped regions, Such as Beijing, Shanghai, Xizang. Second, informatization level is added as an additional control variable to mitigate omitted variable bias. Theoretically, informatization level may concurrently influence both NQP and agricultural carbon emissions. Regions with higher informatization levels are more likely to adopt NQP technologies (e.g., smart agricultural machinery, precision agriculture), thereby reducing carbon emissions through optimizing resource utilization efficiency. Controlling for informatization level thus enables the isolation of the independent effect of NQP, enhancing the accuracy of causal inference. Third, due to the high-dimensional nature of the TWFE model, which may give rise to heteroscedasticity and autocorrelation that distort regression results, the GLS (Generalized Least Squares) model is substituted for the TWFE model in the robustness check. The specific robustness check results are presented in Table 6.

Table 6

Table 6. The robustness tests.

Table 6 presents the results of the robustness tests. Using three methods—excluding samples from economically extreme provinces, adding the omitted variable of informatization level, and replacing the estimation model with GLS, all results show that new quality productivity (NQP) has a significant negative impact on agricultural carbon emissions (LnCO₂). After excluding economically extreme provinces, the coefficient of NQP is −0.210, significant at the 1% level, indicating that carbon emission patterns in economically extreme provinces are heterogeneous, and excluding them allows for a clearer capture of the universal emission reduction effect of NQP. When informatization level is added, the NQP coefficient is −0.201, significant at the 1% level. Although the absolute value of the coefficient decreases slightly compared to the original model, the significance remains substantially unchanged, suggesting that NQP’s emission reduction effect is not driven by informatization level and retains independent explanatory power. After controlling for panel heteroscedasticity and common AR(1) autocorrelation using generalized least squares (GLS), the NQP coefficient is −0.165, significant at the 5% level. All three tests support the theoretical hypothesis that NQP achieves agricultural low-carbon transformation through technological innovation and factor restructuring, enhancing the credibility of causal inference. In summary, the robustness tests further validate the reliability of the research conclusions and provide empirical evidence for agricultural carbon neutrality pathways.

4.4 Mechanism analysis

In the mechanism analysis, this study employs causal mediation analysis approach to test mediating effects, while utilizing a parallel mediation analysis to systematically examine the intermediary roles of facility agriculture and agricultural mechanization in the relationship between new quality productivity and agricultural carbon emissions. The initial phase focused on verifying the mediating effect of facility agriculture as a key variable, with detailed test outcomes presented in Table 7 and Figure 3.

Table 7

Table 7. Mediating effect of facility agriculture.

Figure 3

Figure 3. The mediating effect of facility agriculture.

Table 7 and Figure 3 show that the coefficient of NQP for facility agriculture is 0.826, and it is significant at the 5% level, indicating that the improvement of new quality productivity will significantly promote the development of facility agriculture. For every 1 unit increase in new quality productivity, facility agriculture increases by an average of 0.826 units. Meanwhile, the regression coefficient of facility agriculture on agricultural carbon emissions is −0.034, which is significant at the 1% level. This indicates that the development of facility agriculture can significantly curb agricultural carbon emissions. For every 1 unit increase in facility agriculture, the logarithmic average reduction of agricultural carbon emissions is 0.034 units, reflecting the positive role of facility agriculture in carbon reduction. It conforms to its resource-intensive utilization model (such as precise environmental control), but the direct effect coefficient of NQP on agricultural carbon emissions is −0.118, which fails the significance test, indicating that the emission reduction effect of NQP is more dependent on the intermediary transmission of specific technical paths such as facility agriculture. It reveals that apart from facility agriculture, there are also independent emission reduction approaches such as clean energy substitution or digital management systems. Further calculation of the mediating effect indicates that the indirect effect of NQP on agricultural carbon emissions is −0.028. In conclusion, the research results confirm the “technology-institutional synergy” theory, that is, NQP achieves emission reduction through the technological substitution for the scale expansion of facility agriculture and the complementary institutional optimization required to release greater synergy effects. This provides empirical evidence for the multi-path advancement of carbon neutrality in agriculture. Therefore, Hypothesis H2 is validated.

As shown in Table 8. The results of the mediating effect test show that NQP has multiple pathways of influence on agricultural carbon emissions. The direct impact coefficient of NQP on agricultural mechanization is 1.005, which is significant at the 1% level. That is, for every 1 unit increase in NQP, the level of agricultural mechanization increases by 1.005 units. The impact coefficient of agricultural mechanization on carbon emissions is 0.192 (significant at the 1% level), which means that for every 1 unit increase in mechanization, agricultural carbon emissions will increase by 0.192 units. This positive connection may stem from the carbon-intensive nature of agricultural machinery that relies on fossil fuels.

Table 8

Table 8. Mediating effect of agricultural mechanization.

Meanwhile, as can be seen from Figure 4, the direct impact coefficient of NQP on agricultural carbon emissions is −0.395 (p < 0.01), indicating that the reduction effect is significant through non-mechanized approaches such as adopting clean technologies or improving management efficiency. Further calculations show that the indirect effect of NQP caused by agricultural mechanization is 0.193, and the pure direct effect is −0.588. This indicates that mechanization partially offsets the emission reduction potential of NQP as a mediating variable, verifying the “technology-carbon lock-in paradox,” that is, the improvement of technological efficiency is restricted by the rigid energy structure, thereby limiting the outcome of decarbonization. Suppose H3 is thus rejected.

Figure 4

Figure 4. The mediating effect of agricultural mechanization.

The above analysis of the mediating effect reflects the inherent contradiction of NQP in reducing carbon emissions in agriculture. At the level of synergy, NQP itself has the underlying logic to directly or indirectly reduce emissions through technological innovation and management optimization. However, at the level of conflict, different production methods such as facility agriculture or mechanization have extremely different impacts on carbon emissions. This demonstrates the significance of providing targeted guidance for the development of intermediaries at the specific practical level.

4.5 Heterogeneity analysis

Given the significant regional disparities in the development of New Quality Productivity (NQP) across China, shaped by distinct geographical characteristics and resource endowments that influence technological pathways and factor allocation efficiency, this study conducts subgroup regression analyses based on China’s four major geographical divisions (eastern, central, western, and northeastern regions). As shown in Table 9, the impact of NQP on agricultural carbon emissions (lnCO₂) exhibits marked regional heterogeneity:

• Eastern Region: The NQP coefficient for agricultural carbon emissions is 0.595***, which is significant at the 1% level. This indicates that the development of NQP has instead increased agricultural carbon emissions, reflecting the characteristics of high input and high emissions in agricultural development. This phenomenon is consistent with the technological carbon lock-in effect of highly intensive agricultural mechanization and digitalization. In the development of NQP, the popularization of intensive agricultural production has increased productivity, but it has also strengthened the reliance on chemical fertilizers, pesticides and fossil fuels, leading to excessive consumption of resources.

• Western Region: The NQP coefficient is −0.916***, which is significant at the 1% level, indicating that the carbon reduction effect of NQP is significant. From the perspective of the geographical conditions and resource endowments of this region, the local agricultural development has long been constrained by the priority given to ecological protection over output expansion. Therefore, under this policy orientation, the “ecological attribute” of the western NQP is far stronger than its “efficiency attribute.” This result also reflects the synergy between environmental supervision and efficiency improvement.

• Central and Northeastern Regions: The coefficients of NQP are 0.0254 and 1.418 respectively, but the significance was not strong. Only in the Northeast region was it significant at the 10% level. In the central region, due to the relatively weak NQP level and the less application of high and new low carbon technologies, the overall effect is lacking and scattered, ultimately manifested as statistically insignificant, indicating that the emission reduction potential of new quality productivity has not been fully released, and the correlation with emissions is in a “neutral state.” In Northeast China, the degree of agricultural scale and mechanization is high. This feature is less affected by NQP. The low-carbon transformation has yet to have an impact. Therefore, there is a high reliance on fossil energy. At the same time, the scale amplifies the emission effect of traditional high-carbon production, thus making the carbon emissions in Northeast China the most intense.

Table 9

Table 9. Regional heterogeneity analysis.

Apart from regional heterogeneity, technological advancements driven by New Quality Productivity (NQP) exhibit varying carbon mitigation potentials across different agricultural production sectors. Conducting heterogeneity analyses by disaggregating agricultural processes enables the formulation of targeted and adaptive decarbonization strategies. This study categorizes agricultural production into six sectors—fertilizers, pesticides, irrigation, diesel, tillage, and plastic film—based on data availability and sector-specific emission profiles. Empirical results, detailed in Table 10, reveal significant differences in NQP’s inhibitory effects on carbon emissions from distinct inputs:

• Fertilizers, Pesticides, and Irrigation: Among these three production processes, NQP has the strongest carbon reduction effect, with coefficients and significance of −55.41***, −28.50***, and −18.89***, respectively. It indicates that the increase in NQP can significantly reduce carbon emissions in these three production processes. This might be because the increase in NQP has led to more precise fertilization techniques, reducing carbon emissions caused by excessive use of chemical fertilizers. At the same time, efficient fertilizer technology has also lowered the carbon emission intensity per unit of chemical fertilizer. These results verify the intensive substitution hypothesis, that is, the potential for optimizing resource utilization efficiency contained in NQP significantly reduces redundant high-carbon inputs.

• Plastic Film: The coefficient is −6.139, which is negative but not significant. It is indicated that NQP may tend to reduce carbon emissions related to agricultural films (such as promoting degradable agricultural films and increasing the recycling rate of agricultural films), but the current effect is not obvious enough. This might be because the low usage cost feature of agricultural films themselves may lead to the carbon reduction in this production process being restricted by factors such as development costs, and no significant impact has been formed statistically.

• Tillage: The coefficient is −0.458 and negative effect is significant at the 5% level, but the impact is relatively small. Perhaps it is because the new quality productivity has promoted the application of low-carbon farming techniques such as no-tillage and less tillage, reducing the frequency of agricultural machinery tillage and thus slightly lowering carbon emissions.

• Diesel Inputs: The influence coefficient of NQP on diesel carbon emissions is 5.740, with no significant sign. This means that the impact of NQP on carbon emissions in the diesel process is not statistically significant, but it may have the effect of increasing carbon emissions. One possible reason is that although NQP may enhance the efficiency of agricultural machinery, the increased frequency of use of agricultural machinery (such as large-scale production) may have offset some of the emission reduction effects, resulting in an overall impact that is not significant. At the same time, it indicates that new quality productivity has not yet driven the low-carbon and electrification transformation of agricultural machinery, and agricultural machinery mainly based on diesel is still widely used. This conclusion echoes the masking effect produced by mechanization in the previous study on the mediating effect.

Table 10

Table 10. Analysis of heterogeneity in agricultural production processes.

These findings highlight that the mitigation efficacy of NQP hinges critically on the technological adaptability specific to each sector. Standardized sectors, such as fertilizers and pesticides, can achieve rapid decarbonization. In contrast, systemic transitions like energy substitution in machinery call for longer-term policy coordination.

To elaborate the dynamic changes in the annual carbon reduction effects across production processes under the role of new quality productivity, this study categorizes production processes into groups, conducts yearly regression analyses, and visualizes the results as a stacked area chart, as illustrated in Figure 4.

As can be seen from the Figure 5, from 2010 to 2022, the impact coefficient of NQP on carbon emissions in each production processes of agriculture showed a significant dynamic change trend, reflecting the synergy effect between technological progress and the transformation of production methods. The coefficient of the traditional high-carbon fertilizer industry has gradually declined and transitioned to a negative value, indicating that green fertilization techniques and clean energy substitution have effectively reduced its carbon emission intensity. This reflects a breakthrough in technological innovation in resource utilization efficiency.

Figure 5

Figure 5. Annual coefficients of various production processes.

The continuous negative deepening of pesticide and irrigation coefficients may stem from the widespread adoption of biopesticides and water-saving technologies, which have significantly reduced the carbon footprint associated with chemical inputs and water consumption. The expanded negative value of the tillage coefficient reveals the emission reduction contribution of conservation agriculture practices such as no-tillage or minimum tillage systems, which is consistent with the production concepts of resource conservation and environmental friendliness.

However, even though the coefficient of NQP on diesel carbon emissions has been decreasing year by year, the coefficient of NQP on diesel carbon emissions remains positive, indicating that the development of new quality productivity has not yet played a role in this traditional production process of diesel.

Overall, the correlation coefficients of various departments tend to be negative, indicating the internal mechanism by which new high-quality productive forces drive the low-carbon transformation of agricultural production through technological innovation, policy incentives, and market orientation. This trajectory not only confirms the theoretical expectations of sustainability research regarding technology spillover effects and the environmental Kuznets curve (EKC), but also highlights the structural adjustment potential of agricultural systems in mitigating climate change. This empirical model emphasizes the crucial role of production factor reconstruction and system optimization in achieving the dual goals of emission control and agricultural modernization under the evolving technological paradigm.

5 Discussion

This study elucidates the inhibitory effect of New Quality Productivity (NQP) on agricultural carbon emissions, aligning with the “green productivity attributes” identified by Wang and Ling (2024) and the “mediating effect of energy consumption structure optimization” proposed by Zhao et al. (2025). Notably, the promotion of facility agriculture technologies achieved a 2.8% indirect emission reduction effect, this indicates that NQP has other specific paths to reduce agricultural carbon emissions. Consistent with Wang et al. (2023), who demonstrated that urban facility agriculture innovations reduce carbon footprints through technological advancements. This reinforces the critical role of precision control technologies in enhancing resource efficiency. However, the “masking effect” of agricultural mechanization reveals a “technology-carbon lock-in” paradox due to fossil fuel dependency in conventional machinery. This phenomenon aligns with Li et al. (2024), who theorized that technological improvements may inadvertently prolong traditional energy systems’ lifecycle, and corroborates Zhuang et al.’s (2025) assertion that renewable energy substitution is pivotal for overcoming agricultural carbon constraints. These findings underscore the untapped potential of electrifying agricultural machinery to structurally replace fossil fuel-dependent systems.

Regional heterogeneity analysis indicates a stronger emission reduction effect of NQP in western China, attributable to “ecologically embedded innovation” models such as photovoltaic agriculture and integrated crop-livestock systems. These practices effectively substitute high-carbon inputs, reflecting the synergy between environmental and efficiency goals under the Porter Hypothesis. Conversely, the paradoxical positive effect observed in eastern China exposes path dependency in intensive production systems. Despite efficiency gains from digital technologies, persistent reliance on fertilizers, pesticides, and diesel fuels induces a rebound effect (Zheng et al., 2022). This regional disparity echoes Milindi and Inglesi-Lotz’s (2022) conclusion that uneven technology diffusion exacerbates carbon inequality, highlighting the urgency for eastern China to transcend “yield-first” technological lock-ins.

Beyond macro-level mechanisms, this study bridges micro-level insights by delineating specific technological pathways through which NQP mitigates agricultural emissions. Heterogeneity analysis across production stages reveals limited effectiveness of NQP in reducing diesel-related emissions. Two technical barriers emerge: First, conventional machinery remains reliant on incremental fossil energy innovations (e.g., smart monitoring), which optimize efficiency but fail to disrupt diesel-based architectures. Second, scaling disruptive technologies like electric agricultural machinery faces triple constraints: (1) energy storage limitations reduce operational endurance compared to fossil fuel counterparts; (2) high upfront costs deter adoption despite long-term savings (Gul et al., 2024); and (3) inadequate charging infrastructure in rural areas (Gao et al., 2024).

Theoretically, this research challenges the linear assumption that “technological progress inherently reduces emissions” by constructing a three-dimensional “technology-pathway–resource-endowment–production-stage” framework. This advances context-specific understandings of NQP’s environmental attributes. At the practical level, firstly, this study reveals the importance of adopting facility agriculture and promoting the electrification transformation of agricultural mechanization in advancing agricultural carbon emission reduction, providing precise references for optimizing technical pathways. Secondly, the heterogeneity analysis clarifies the differences in carbon emission reduction across eastern and western regions as well as among various agricultural production links. This not only offers empirical support for formulating regionally differentiated policies but also provides targeted guidance for emission reduction in specific production processes. Nevertheless, limitations persist: First, the NQP index system incompletely captures frontier technologies like bio-breeding; future studies should integrate gene-editing and smart breeding metrics. Second, the mechanism analysis prioritizes technological variables (e.g., mechanization) over institutional factors like digital finance and carbon markets. Third, regional heterogeneity analysis lacks urban cluster granularity (e.g., Yangtze River Delta). Future research could employ machine learning to simulate multi-pathway carbon reduction scenarios, enabling dynamic policy optimization.

6 Conclusions and recommendations

6.1 Conclusion

Based on China’s provincial panel data, this study empirically finds that new quality productivity has a significant inhibitory effect on agricultural carbon emissions with its impacts exhibiting multi-dimensional and complex characteristics, where benchmark regression and endogeneity tests show that a 1-unit increase in new-quality productivity reduces agricultural carbon emissions by 16.4 to 38.2% and the reliability of this conclusion is validated through instrumental variable methods and multiple robustness tests; mechanism analysis reveals that new quality productivity achieves a 2.8% indirect emission reduction effect through the promotion of protected agriculture technologies while the improvement in mechanization generates a masking effect due to fossil energy dependence, partially offsetting the emission reduction effects and forming a technology-carbon lock-in paradox; regional heterogeneity tests show that the western region has the strongest emission reduction effect with a coefficient of −0.916, the eastern region exhibits an anomalous promoting effect (coefficient = 0.595) due to energy path dependence in intensive production, and the central and northeastern regions show no significant effects; heterogeneity analysis across production stages further reveals that new quality productivity has the most prominent emission reduction effects on high-carbon input stages such as chemical fertilizers and pesticides (coefficients of −55.41 and −28.50, respectively) while its effects on stages involving diesel and agricultural films are weaker (coefficients of 5.740 and −6.139, respectively) and its impact on the carbon emissions of the film is negligible(coefficients is −0.458); the study demonstrates that the carbon reduction effect of new quality productivity is constrained by technological pathways, regional resource endowments, and production stage characteristics with its low-carbon transformation potential requiring the synergistic release of technological innovation and systemic reconstruction, thus providing theoretical and empirical evidence for achieving agricultural carbon neutrality goals.

6.2 Recommendations

First, enhance domestic technological innovation and energy system restructuring. Accelerate the integration of facility agriculture with photovoltaic and biomass energy systems, address technical bottlenecks in battery endurance and lightweight design for electric agricultural machinery, and develop a digital management system for carbon footprint monitoring and input optimization. This will drive the low-carbon transformation of agricultural machinery power systems and green technological upgrading in facility agriculture.

Second, adopt region-specific low-carbon pathways. In western regions, promote “pasture-solar hybrid” and “forest-grain intercropping” models with carbon sink revenue compensation mechanisms. In eastern regions, establish dual performance evaluation systems for both production growth and carbon reduction, and pilot “smart farm-carbon trading” integration models. In central and northeastern regions, prioritize smart agricultural machinery adoption to replace chemical inputs, implement protective tillage for black soil conservation, and advance electrification of agricultural machinery.

Third, establish a global collaborative governance framework for carbon emissions. Strengthen international R&D cooperation on agricultural carbon neutrality with partners such as the EU and Brazil, sharing precision control technologies for facility agriculture and promoting the export of electric agricultural machinery technologies. Deploy photovoltaic greenhouse models along the Belt and Road, participate in international mutual recognition of agricultural carbon sink certification systems, and integrate domestic carbon sequestration metrics into global carbon market accounting frameworks.

Finally, deepen international policy coordination and global governance platform development. Domestically, establish an international cooperation fund for new-quality agricultural productivity to support low-carbon technology training and capacity-building in developing countries. Under frameworks like the G20, launch a Global Alliance for Agricultural Carbon Neutrality to facilitate technology transfer of electric agricultural machinery and smart agriculture from developed to developing nations, and institute a performance evaluation system for agricultural emission reduction outcomes.

Data availability statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Author contributions

YX: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Resources, Software, Validation, Visualization, Writing – original draft, Writing – review & editing. KC: Writing – review & editing. HC: Writing – original draft. DZ: Writing – review & editing. YL: Writing – original draft. HX: Writing – original draft. FL: Project administration, Supervision, Writing – review & editing.

Funding

The author(s) declare that no financial support was received for the research and/or publication of this article.

Acknowledgments

The researchers express sincere gratitude to all participants whose dedication made this study possible.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The authors declare that Gen AI was used in the creation of this manuscript. This article uses only gen AI for grammatical editing.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fsufs.2025.1639452/full#supplementary-material

Abbreviations

SDGs, United Nations Sustainable Development Goals; SDG2, Sustainable Development Goal 2 (Zero hungry); SDG13, Sustainable Development Goal 13 (Climate action); FAO, Food and Agriculture Organization of the United Nations; CPC, Communist Party of China; NQP, New Quality Productivity; GTFP, Green Total Factor Productivity; GLS, Generalized Least Squares; TWFE, Two-way fixed effects; ACE, Agriculture Carbon Emissions; G20, Group of 20; R&D, Research and Development.

References

Chang, H., Heerink, N., Wu, W., and Zhang, J. (2022). More use or cleaner use? Income growth and rural household energy-related carbon emissions in Central China. Energy Sustain. Dev. 70, 146–159. doi: 10.1016/j.esd.2022.07.013

Crossref Full Text | Google Scholar

Chen, J., Gao, M., Mangla, S. K., Song, M., and Wen, J. (2020). Effects of technological changes on China's carbon emissions. Technol. Forecast. Soc. Change 153:938. doi: 10.1016/j.techfore.2020.119938

Crossref Full Text | Google Scholar

Chen, P.-B., Huang, K.-Q., and Zhu, C.-Z. (2025). Research on the impact of agricultural new quality productivity on agricultural carbon emission intensity. Mod. Manag. 45, 21–31. doi: 10.19634/j.cnki.11-1403/c.2025.01.003

Crossref Full Text | Google Scholar

Fang, F., Zhao, J., Huang, H. Y., and Xufeng, S. (2024). Urbanization promotion modes and China’s agricultural low-carbon total factor productivity: Evidence from dual-source night-time lights. Statistical Research 41, 111–125. doi: 10.19343/j.cnki.11-1302/c.2024.04.009

Crossref Full Text | Google Scholar

Gao, W. S., Chen, Y. Q., Wang, X. L., and Huang, J. X. (2022). Exploration of technical pathways and countermeasures for carbon neutrality in China's planting industry. Res. Agric. Modern. 43, 941–947. doi: 10.13872/j.1000-0275.2022.0091

Crossref Full Text | Google Scholar

Gao, D., Zhi, Y., and Yang, X. (2024). Assessing carbon emission reduction benefits of the electrification transition of agricultural machinery for sustainable development: a case study in China. Sustain Energy Technol Assess 63:103634. doi: 10.1016/j.seta.2024.103634

Crossref Full Text | Google Scholar

Guan, N., Liu, L., Dong, K., Xie, M., and Du, Y. (2023). Agricultural mechanization, large-scale operation and agricultural carbon emissions. Cogent Food Agric. 9:430. doi: 10.1080/23311932.2023.2238430

Crossref Full Text | Google Scholar

Gul, D., Banday, R. U. Z., and Muzamil, M. (2024). Current status, performance characteristics and future perspective of electric tractor. Adv. Mod. Agric. 5:3078. doi: 10.54517/ama3078

Crossref Full Text | Google Scholar

Guo, X., Huang, S., and Wang, Y. (2020). Influence of agricultural mechanization development on agricultural green transformation in Western China, based on the ML index and spatial panel model. Math. Probl. Eng. 2020:6351802. doi: 10.1155/2020/6351802

Crossref Full Text | Google Scholar

Han, H.-B., and Gong, Y. (2025). Spatial spillover effects of digital village construction on agricultural carbon emissions. Trans. Chin. Soc. Agric. Eng. 41, 298–307. doi: 10.11975/j.issn.1002-6819.202408173

Crossref Full Text | Google Scholar

He, H., and Ding, R. (2022). Spatiotemporal heterogeneity effect of technological progress and agricultural centrality on agricultural carbon emissions in China. Front. Environ. Sci. 10:1078357. doi: 10.3389/fenvs.2022.1078357

Crossref Full Text | Google Scholar

Huang, Y., Li, X., Chi, J. T., and Wang, Y. W. (2024). Spatial spillover and threshold effect of rural digitization on agricultural carbon emission intensity. J. Kunming Univ. Sci. Technol. 49, 180–191. doi: 10.16112/j.cnki.53-1223/n.2024.01.292

Crossref Full Text | Google Scholar

Jia, Y. L., and Xu, H. (2024). Study of the relationship between agricultural mechanization, environmental regulation, and agricultural carbon emissions. J. Chin. Agric. Mech. 45, 208–215.

Google Scholar

Li, J., and Jiang, Q. (2024). Rural inclusive finance and agricultural carbon reduction: evidence from China. J. Knowl. Econ. 16, 9806–9829. doi: 10.1007/s13132-024-02261-9

Crossref Full Text | Google Scholar

Li, W., and Zhang, P. (2021). Relationship and integrated development of low-carbon economy, food safety, and agricultural mechanization. Environ. Sci. Pollut. Res. 28, 68679–68689. doi: 10.1007/s11356-021-15465-2

Crossref Full Text | Google Scholar

Li, Z., Zhao, Q., Guo, H., and Huang, R. (2024). Impact of fossil fuel subsidies on energy-saving technological change in China. Energy 286:129265. doi: 10.1016/j.energy.2023.129265

Crossref Full Text | Google Scholar

Lin, L., Gu, T., and Shi, Y. (2024). The influence of new quality productive forces on high-quality agricultural development in China: mechanisms and empirical testing. Agriculture 14:1022. doi: 10.3390/agriculture14071022

Crossref Full Text | Google Scholar

Liu, Y., and He, Z. (2024). Synergistic industrial agglomeration, new quality productive forces and high-quality development of the manufacturing industry. Int. Rev. Econ. Finance 94:103373. doi: 10.1016/j.iref.2024.103373

Crossref Full Text | Google Scholar

Liu, J., Li, Y., and Wang, Z. (2023). The potential for carbon reduction in construction waste sorting: a dynamic simulation. Energy 275:127477. doi: 10.1016/j.energy.2023.127477

Crossref Full Text | Google Scholar

Liu, Z., Wei, Y., Liao, R., Yamaka, W., and Liu, J. (2025). Synergistic effects of dual low-carbon pilot policies on urban green land use efficiency: mechanisms and spatial spillovers through difference-in-differences and spatial econometric analysis. Land 14:882. doi: 10.3390/land14040882

Crossref Full Text | Google Scholar

Liu, L. W., and Zhou, H. Q. (2018). Factors of potato economic growth in Heilongjiang Province from the perspective of low-carbon economy. J. Discret. Math. Sci. Cryptogr. 21, 589–594. doi: 10.1080/09720529.2018.1463677

Crossref Full Text | Google Scholar

Lu, J., Guo, Z. A., and Wang, Y. P. (2024). The development level, regional differences, and improvement pathways of new quality productivity. J. Chongqing Univ. 30, 1–17. doi: 10.13246/j.cnki.iae.2024.04.010

Crossref Full Text | Google Scholar

Luo, B., and Geng, P. (2024). New quality productivity in agriculture: theoretical lineage, basic kernel and promotion path. Issues Agric. Econ. 4, 13–26.

Google Scholar

Luo, W., Zuo, S., Tang, S., and Li, C. (2025). The formation of new quality productivity of agriculture under the perspectives of digitalization and innovation: a dynamic qualitative comparative analysis based on the “technology-organization-environment” framework. Sustainability 17:597. doi: 10.3390/su17020597

Crossref Full Text | Google Scholar

Ma, X., and Yang, X. (2024). Promoting high-quality development of agriculture by accelerating the formation of new quality productivity. Issues Agric. Econ. 4, 4–12. doi: 10.13246/j.cnki.iae.2024.04.009

Crossref Full Text | Google Scholar

Ma, L., Zhang, W., Wu, S., and Shi, Z. (2022). Research on the impact of rural population structure changes on the net carbon sink of agricultural production—take Huan County in the loess hilly region of China as an example. Front. Environ. Sci. 10:911403. doi: 10.3389/fenvs.2022.911403

Crossref Full Text | Google Scholar

Milindi, C. B., and Inglesi-Lotz, R. (2022). Impact of technological progress on carbon emissions in different country income groups. Energy Environ. 34, 1348–1382. doi: 10.1177/0958305X221087507

Crossref Full Text | Google Scholar

Nian, T., Li, Y., Jin, X., Wang, Z., Wang, M., and Wang, Y. (2025). Toward carbon balance in life cycle: the carbon emission assessment for the recycled coarse aggregate concrete. Adv. Civ. Eng. 2025. doi: 10.1155/adce/9184976

Crossref Full Text | Google Scholar

Tang, Y., and Chen, M. (2022). Impact mechanism and effect of agricultural land transfer on agricultural carbon emissions in China: evidence from mediating effect test and panel threshold regression model. Sustainability 14:13014. doi: 10.3390/su142013014

Crossref Full Text | Google Scholar

Tao, B. X. (2017). Characteristics of soil CO₂ emissions in typical northern facility vegetable fields. Res. Environ. Sci. 30, 1927–1933. doi: 10.13198/j.issn.1001-6929.2017.03.56

Crossref Full Text | Google Scholar

Tong, H. T., Xia, E. J., Sun, C., and Huang, J. P. (2024). The impact of facility agriculture development on agricultural carbon emission efficiency. China Environ. Sci. 44, 7079–7094. doi: 10.19674/j.cnki.issn1000-6923.20240822.004

Crossref Full Text | Google Scholar

Wang, C., Chen, Y., Sun, M., and Wu, J. (2023). Potential of technological innovation to reduce the carbon footprint of urban facility agriculture: a food–energy–water–waste nexus perspective. J. Environ. Manag. 339:117806. doi: 10.1016/j.jenvman.2023.117806

PubMed Abstract | Crossref Full Text | Google Scholar

Wang, Z., Hu, T., and Liu, J. (2024). Decoupling economic growth from construction waste generation: comparative analysis between the EU and China. J. Environ. Manag. 353:120144. doi: 10.1016/j.jenvman.2024.120144

Crossref Full Text | Google Scholar

Wang, X., Jiang, X., and Yin, T. (2025). Impact of agricultural new productive forces on agricultural carbon emission reduction in China under the background of "double carbon". Bull. Soil Water Conserv. 45, 295–306. doi: 10.13961/j.cnki.stbctb.2025.03.036

Crossref Full Text | Google Scholar

Wang, Y., and Ling, W. (2024). The impact of new quality productive forces on carbon emissions in manufacturing enterprises. Front. Bus. Econ. Manag. 16, 253–264. doi: 10.54097/tzrcbz33

Crossref Full Text | Google Scholar

Wang, Z., Qin, F., Liu, J., and Jin, X. (2025a). Evolution trajectory and driving mechanism of the synergistic effect on construction waste and carbon reduction: evidence from China. Waste Manag. 203:114891. doi: 10.1016/j.wasman.2025.114891

PubMed Abstract | Crossref Full Text | Google Scholar

Wang, Z., Wei, K., Wen, B., You, K., Wang, H., Ye, C., et al. (2025b). Research on the allocation level of land for agricultural facilities based on green and high-quality development: a case study of Zhejiang Province. Land 14:672. doi: 10.3390/land14040672

Crossref Full Text | Google Scholar

Yang, T., Huang, X., Wang, Y., Li, H., and Guo, L. (2022). Dynamic linkages among climate change, mechanization and agricultural carbon emissions in rural China. Int. J. Environ. Res. Public Health 19:14508. doi: 10.3390/ijerph192114508

PubMed Abstract | Crossref Full Text | Google Scholar

Yin, Y., Xi, F. M., Bing, L. F., Wang, J. Y., Li, J. Y., Du, L. Y., et al. (2021). Carbon emission accounting and carbon reduction pathways in facility agriculture in China. Chin. J. Appl. Ecol. 32, 3856–3864. doi: 10.13287/j.1001-9332.202111.019

Crossref Full Text | Google Scholar

Yu, J. X., Jin, J., and Zhao, B. K. (2025). Research on the impact of agricultural carbon emissions from the perspective of new quality productivity. South China Rural Area 41, 39–46.

Google Scholar

Zhang, J., Trautman, D., Liu, Y., Bi, C., Chen, W., Ou, L., et al. (2024a). Achieving the rewards of smart agriculture. Agronomy 14:452. doi: 10.3390/agronomy14030452

Crossref Full Text | Google Scholar

Zhang, J., Wang, F., and Ding, X. (2024b). Can agricultural mechanization promote carbon reduction in countries along the belt and road? J. Environ. Plan. Manag. 68, 2194–2216. doi: 10.1080/09640568.2024.2311821

Crossref Full Text | Google Scholar

Zhao, H., Li, C., Yang, Z., Zhang, Z., Jin, C., and Guo, S. (2025). How can new quality productivity decrease carbon emission intensity? Evidence from 30 provincial regions in China. Energy 327:136487. doi: 10.1016/j.energy.2025.136487

Crossref Full Text | Google Scholar

Zheng, Y., Xiao, J., Huang, F., and Tang, J. (2022). How do resource dependence and technological progress affect carbon emissions reduction effect of industrial structure transformation? Empirical research based on the rebound effect in China. Environ. Sci. Pollut. Res. 30, 81823–81838. doi: 10.1007/s11356-022-20193-2

PubMed Abstract | Crossref Full Text | Google Scholar

Zhuang, M., Wang, X., Yang, Y., Wu, Y., Wang, L., and Lu, X. (2025). Agricultural machinery could contribute 20% of total carbon and air pollutant emissions by 2050 and compromise carbon neutrality targets in China. Nat. Food 6, 513–522. doi: 10.1038/s43016-025-01163-6

PubMed Abstract | Crossref Full Text | Google Scholar

Appendix

See Table A1

Table A1

Table A1. Results of Levin-Lin-Chu panel unit root test.