China's push toward digitally enabled agriculture is doing more than adding sensors, machinery and data systems to farms. It is changing how technological progress is distributed across the basic factors of production, with capital and labor becoming increasingly important relative to land where cultivated area is constrained.
The finding comes from "Digital Intelligence–Agricultural Economy Integration and Agricultural Technological Choice Under Land Constraints: Evidence from China," a study published in the MDPI journal Land by Qi Zhang, Zewen Yuan and Wanping Yang of Xi'an Jiaotong University. Using provincial data from 28 Chinese provinces between 2011 and 2024, the study examines how the integration of digital intelligence with the agricultural economy influences the direction of technological change under land scarcity.
The researchers find that stronger digital intelligence–agricultural economy integration is associated with technological progress that favors capital and labor relative to land. The effect is concentrated in regions where cultivated land is scarce, suggesting that resource constraints do not merely limit agricultural expansion; they also influence the kinds of technologies farmers and production systems adopt.
Land scarcity is redirecting technological change
Cultivated land occupies a unique position in agriculture because its supply cannot expand freely with production demand. Once land becomes increasingly difficult to add, raising agricultural output depends more heavily on improving the effectiveness of the other factors operating on that fixed base.
The study approaches this problem through the concept of biased technological progress. Instead of treating technological change as a uniform force that lifts productivity across the board, it examines whether new technologies disproportionately strengthen capital, labor or land.
This framework allows the researchers to move beyond a simple question of whether digitalization improves productivity. Precision irrigation, remote sensing and soil monitoring can directly improve land efficiency, while intelligent machinery and automated equipment can strengthen capital. Digital production systems can also increase the value of workers who operate machinery, interpret information and manage increasingly complex production processes.
To capture this transformation, the authors construct an index of digital intelligence–agricultural economy integration, or DIAI. The index combines indicators of digitalization and intelligent development with measures of agricultural infrastructure, economic structure, green development, output and rural living conditions.
The empirical results point in a clear direction. Higher DIAI is associated with stronger capital-augmenting and labor-augmenting technological progress relative to land. After the study introduces its full set of controls, the estimated coefficient is 1.078 for capital relative to land and 0.796 for labor relative to land, with both relationships statistically significant.
These results do not suggest that land is becoming less relevant, but indicate that when the land base cannot expand easily, adjustment occurs increasingly through machinery, information systems, skills and new forms of production organization.
Capital and labor are strengthened through different channels
Digital transformation does not affect all production factors in the same way. The study finds that the technological shift toward capital operates through both changes in relative input scale and changes in relative prices.
As intelligent machinery, sensing equipment, automated facilities and digital production systems become more deeply embedded in agriculture, more capital is deployed on a given land base. At the same time, changes in the costs and expected returns associated with capital and land alter the incentives surrounding technology adoption.
The mechanism results support both channels. The estimated coefficient for the capital–land scale effect is 0.896, while the price effect is 2.527, with both statistically significant. This suggests that capital-oriented technological change is shaped by how much equipment and infrastructure is used relative to land as well as by the economic incentives governing that deployment.
Labor follows a different pattern. The scale effect is statistically significant, while the labor–land price effect is not. The study links this to the way digital agriculture changes the content and organization of work rather than simply changing wages or labor costs.
Workers in digitally transformed agricultural systems may be required to operate intelligent equipment, process information, monitor production conditions and implement data-supported decisions. Labor therefore becomes more technologically capable not because its price rises relative to land, but because its role within production is reorganized.
The asymmetry has direct consequences for policy design. Capital-oriented strategies need to address financing, equipment access, utilization rates, service pricing and the economic conditions surrounding investment. Labor-oriented strategies require attention to skills, production organization and the ability of workers to function effectively within digital production systems.
Scarce land makes digital integration more consequential
The geographical results sharpen the study's argument. When the sample is divided according to cultivated land availability, the technological effects of digital integration appear mainly in land-scarce regions.
For the capital–land relationship, the DIAI coefficient reaches 1.611 in cultivated land-scarce regions and is statistically significant. In land-abundant regions, the corresponding estimate is not significant. The pattern is similar for labor: the coefficient is 1.483 in land-scarce regions, while no significant effect appears in land-abundant areas.
The researchers also test this relationship using the full sample and find that the effect of digital integration weakens as cultivated land per capita increases. In other words, the tighter the land constraint, the stronger the tendency for digital intelligence to shift technological progress toward capital and labor.
The economic logic is straightforward. Where land is relatively abundant, producers retain more flexibility to expand or adjust production through land input. Where land becomes scarce, improving machinery, management, skills and production organization offers a more viable route to increasing productive capacity.
The findings challenge uniform approaches to agricultural digitalization. Land-scarce regions may require greater emphasis on precision machinery, sensing technologies, intelligent irrigation and field-level digital management. Land-abundant areas may benefit more from machinery coordination, service platforms and digital systems capable of managing larger or more dispersed production areas.
For developing countries, the broader lesson is not that China's policy model can simply be transferred elsewhere. The study does not test conditions outside China. Its relevance lies in showing that the impact of digital agriculture is likely to depend heavily on local factor endowments, land availability and production structures.
Policy has to focus on fit, not just technology uptake
The study's policy recommendations move away from judging digital transformation through investment volumes, equipment installation or infrastructure coverage alone. Those indicators can show whether technology is present, but not whether it is reorganizing production effectively.
For capital-intensive technologies, the authors argue that policy should combine investment support with better allocation and use. Shared machinery services, leasing models, digital agricultural platforms and financing support can improve access, but equipment also has to match plot size, crop structure, irrigation conditions and the degree of land fragmentation.
Poorly matched investment creates a different risk: farms may acquire sophisticated machinery without the operational scale or land configuration needed to use it efficiently. Capital deepening, in that case, can produce underutilized assets rather than higher productivity.
Labor policy requires a separate approach. Because the study does not find a significant labor-price channel, direct wage interventions are less closely aligned with the mechanism observed in the data. Training in intelligent machinery operation, field data collection, remote sensing, production monitoring and digital decision support is more consistent with the way labor adjustment appears to occur.
The study also carries several limitations. DIAI is a composite index and may still contain measurement overlap despite multiple robustness tests. Provincial data cannot capture differences between farms, crops, landholding sizes or local production arrangements, while variables such as farmer age, digital skills and land fragmentation are not fully incorporated.
The measurement of farmland rent is another constraint. Consistent long-term provincial rental data are unavailable, so the study relies partly on survey-based rental proportions derived from 669 rural households across 25 provincial-level regions, combined with agricultural output information.
These limitations leave several questions unresolved. Farm-level evidence is still needed to identify which technologies produce the strongest effects, which producers benefit most, how smallholders respond, and whether the observed shift toward capital and labor ultimately improves land-use efficiency and environmental performance.
The broader takeaway is that digital transformation cannot be assessed by counting devices, platforms or investment projects. Its economic significance depends on how technology interacts with the resource constraints already shaping agriculture.
