To examine the impact of digital technology on the integration of industry and town, this study constructs the following fixed effects model for empirical testing:

$c{p}_{i,t}={\beta }_0+{\beta }_{1}d{e}_{i,t}+{\beta }_c{X}_{i,t}+{\mu }_i+{\delta }_t+{\epsilon }_{i,t}$ (1)

Among these, $c{p}_{i,t}$ represents the level of industry-town integration in county i in year t; de represents the level of digitalization in county i in year t; X represents a series of control variables that may influence industry-town integration; ${\mu }_i$ and ${\delta }_t$ represent unobservable individual fixed effects and time fixed effects, respectively; ${\epsilon }_{i,t}$ represents the random error term; ${\beta }_0$ represents the intercept term of the model; ${\beta }_1$ is the coefficient of the digital technology variable, whose magnitude and sign reflect its impact on industry-town integration.

On the one hand, township industries are rooted in county-level areas through the integrated development of the primary, secondary, and tertiary sectors in rural areas. On the other hand, as the basic administrative units of China’s administrative divisions, township units have relatively limited channels for data acquisition, poor stability and continuity of basic data, and significant challenges in obtaining data. Taking into account the difficulty of data acquisition and the rationality of indicators, this study draws on the approach of ( Lai et al., 2022 ) to position the enabling effects of digital technology on the integration of industries and towns at the county level.

Based on this, to investigate the impact of digital technology on the integration of industry and town under the background of common prosperity, this study, after data collation and screening, has collected and organized panel data from a total of 1508 county-level samples in China from 2018 to 2020. The specific data sources include the Digital Village Index of Counties released by the New Rural Development Research Institute of Peking University, the Statistical Yearbook of China’s Counties (County and City Volume), the CSMAR database, the annual county-level database of CEIC, and statistical bulletins of various counties and cities. Missing values for some county-level data were supplemented using linear interpolation and by querying annual statistical bulletins and five-year planning data of the missing counties.

Following the approach of ( Lai et al., 2022 ), this study uses the average DMSP/OLS light intensity across China’s counties and districts as a proxy indicator for industry-town integration. The reason is that nighttime light intensity is the most direct and comprehensive visible light representation of the intensity of human economic and social activities in the town area, which can effectively capture the symbiotic relationship between “industry” and “town” in the development process of interdependence and mutual promotion. Continuous and high-intensity light patches are usually highly correlated with industrial clusters such as industrial areas, logistics parks, and large commercial facilities. Their brightness and range changes can reflect the vitality and spatial expansion of industrial development ( Gui et al., 2024 ) as well as the wealth level of the region ( Alesina et al., 2016 ). The essence of industry town integration is the organic combination and collaborative symbiosis of industry and urban functions within a specific region. This symbiotic relationship is manifested as industrial development providing employment opportunities, economic power, and tax sources for urban areas, driving urban expansion and prosperity; Urban areas provide labor, infrastructure, public services, and market space for industries, supporting industrial upgrading and agglomeration. This high-intensity and close interactive relationship will manifest as abnormally active and intensive human economic activities within the region. And nighttime light intensity is an ideal proxy variable for capturing this high-intensity economic interaction and factor aggregation. In areas with a high degree of integration between industry and town, the “production” (industry) and “life” (urban) activities are highly overlapping and inseparable in terms of time and intensity. Industrial workers gather to form residential areas, which stimulate consumer spending; The service industry in urban areas, in turn, provides services to the industrial population. The highly mixed and interactive “industry city” functions in this area result in a comprehensive lighting intensity that far exceeds that of a single functional area at night, and is more long-lasting and bright. Therefore, the average brightness of the overall lighting in a county can reflect the superposition level and interaction frequency of the economic energy of the two major systems of “production” and “town” within it, and its intensity directly characterizes the level and activity of the integration of production and town.

This study selected the Peking University County Digital Rural Index as a substitute indicator for digitalization levels. The index comprehensively evaluates the digitalization of rural areas in China from eight aspects: digital production, digital supply chains, digital marketing, digital finance, digital governance, digital consumption, digital culture, education, and healthcare, and digital living. It has strong authority and scientific validity ( Zhao et al., 2022 ) and is therefore suitable for the empirical analysis in this study.

Based on existing research, this study selected the following control variables: 1) Population density (pd): the ratio of the total registered population of a region to its administrative planning area; 2) Agricultural foundation (agrf): per capita grain production; 3) Consumption potential (consume): per capita total retail sales of consumer goods; and 4) Resident income (income): per capita disposable income of urban residents.

The descriptive statistics of the main variables are shown in Table 1 . As can be seen from the results in the table, the average value of the level of industry-town integration is 6.688, with a standard deviation of 8.184, indicating that there are significant differences in the level of industry-town integration across different regions. The average value of the digitalization level is 55.340, the median is 55.340, the minimum value is 17.310, and the maximum value is 87.260, indicating that there is a certain degree of polarization in the digitalization levels across different regions.

Table 1. Descriptive statistics of variables.

Table 2 reports the regression results of digital technology on the level of industry-town integration. Column (1) shows the univariate regression based on controlling for county fixed effects and year fixed effects. The regression results indicate that the regression coefficient of the digital technology treatment variable is significantly positive at the 1% level, preliminarily indicating that digital technology empowerment has a significant positive impact on the level of industry-town integration in the region. Considering that characteristics such as population density, agricultural foundation level, consumption potential level, and residents’

Note: ***, ** and * indicate significance levels of 1%, 5%, and 10%, respectively. The values in parentheses are t-values after controlling for robustness.

income level may influence the industrial-urban integration level of a region, this study further controlled for these variables based on the aforementioned univariate regression. After incorporating these control variables, according to the regression results in Column (2) of Table 2 , it can be observed that, holding other conditions constant, a 1 percentage point increase in digitalization levels can raise the industrial-urban integration level of the corresponding region by 2.2%. This indicates that the application of digital technology can significantly promote the improvement of industrial-urban integration levels in the corresponding regions, thereby supporting the validity of Hypothesis H1.

Considering that outliers in the regression sample may lead to biased estimation results, this study performed tail trimming at the 1% and 99% quantiles for all continuous variables in the benchmark regression, replacing values above the highest quantile and below the lowest quantile, and then re-conducted the regression analysis. The specific regression results are detailed in Column (1) of Table 3 . It can be observed that the conclusions of this study remain unchanged after removing the outliers.

Considering the robustness of the identification strategy, this study changed the dimension of the cluster-robust standard error and performed cluster analysis at the city level. The regression results are listed in column (2) of Table 3 . The regression results show that the coefficient of the digitalization level is still significantly positive, and the research conclusion remains valid.

Given the heterogeneity among different county-level administrative regions in various aspects, this study excluded county-level city samples and re-ran the regression analysis. The specific regression results are shown in Column (3) of Table 3 . The regression results indicate that the level of digitization remains significantly positive at the 1% level, confirming the robustness of the conclusions drawn in the preceding discussion.

Considering that the core explanatory variables may vary at the provincial or city level, this study controlled for provincial fixed effects and city fixed effects in addition to controlling for year and county fixed effects to avoid endogeneity issues caused by omitted variables. The regression results are shown in Table 3 , column (4). The coefficient of digitalization level on town-industry integration is still significantly positive at the 1% level.

After undergoing the aforementioned robustness tests, this study may still be subject to selection bias, which could potentially impact the robustness of the regression results. Compared to traditional causal inference methods, dual machine learning offers significant advantages in terms of variable selection and model specification. Based on this, this study employs a dual machine learning approach to re-examine the causal relationship between digital technology and urban-rural integration. In this paper, the sample splitting ratio for the dual machine learning method is set to 1:5. Under the control of county and year fixed effects, the quadratic terms of the control variable set are included, and the random forest algorithm is used to predict and solve the main regression and auxiliary regression. According to the results in column (5) of Table 3 , the regression coefficient for digitalization remains significantly positive at the 1% level. This indicates that after mitigating potential endogeneity issues using the dual machine learning method, the conclusions of this study remain robust.

Note: ***, ** and * indicate significance levels of 1%, 5%, and 10%, respectively. The values in parentheses are t-values after controlling for robustness.

Based on the analysis and arguments presented earlier, we can conclude that digital technology enables the symbiotic integration of industry and towns under the backdrop of common prosperity. This conclusion remains robust after undergoing a series of robustness tests. This section will further examine the mechanisms through which digital technology facilitates the symbiotic integration of industry and towns. Theoretical analysis suggests that digital technology can influence this integration through three channels: residents’ wealth accumulation, labor force return, and external demand stimulation. To this end, this paper measures residents’ wealth by constructing residents’ income levels at the county level, measures labor force return rates using the proportion of rural residents’ permanent population in the total regional population, and measures external demand stimulation rates using county export volumes. By examining whether digital technology can influence residents’ wealth accumulation, labor force return, and external demand stimulation, we aim to validate the reliability of the theoretical mechanisms outlined above.

Digital technology has significantly improved residents’ wealth levels by optimizing resource allocation, thereby promoting the integration and symbiosis of industry and towns ( Zheng & Liu, 2025 ). This study uses residents’ income levels across counties to represent their wealth levels. According to the regression results in Column (1) of Table 4 , the regression coefficient for digital technology on residents’ wealth levels is 0.00764 (p < 0.01, t = 2.97), indicating that a 1-unit increase in digital technology leads to a 0.76% increase in residents’ wealth levels. This result indicates that artificial intelligence and digital technology can significantly enhance regional wealth accumulation efficiency by reducing transaction costs and optimizing factor allocation. Specifically, digital technology expands agricultural product sales channels through e-commerce platforms and improves capital turnover efficiency through digital payments, directly promoting farmers’ income growth and wealth accumulation, thereby strengthening the economic foundation of industrial-urban integration. Assuming H2 is verified.

The positive impact of digital technology on labor return rates validates its role in promoting the integration of industry and towns. At the macro level, the stability or increase in the proportion of permanent rural residents in a region usually indicates an enhanced attraction to labor or a slowdown in outflow trends. The creation of job opportunities through the development of local industries driven by digital technology, which attracts out-migrants back to their hometowns, is one of the important factors leading to this increase. Although the increase may also be due to a decrease in the outflow of local labor rather than an increase in active return migration, it still reflects the retention effect of local economic vitality and increased job opportunities on labor, which is consistent with the goal logic of digital technology empowering local industries and promoting the integration of industry and town. Therefore, this paper uses the proportion of permanent rural residents in the total population of the region as an indicator to measure labor return migration. This paper uses the proportion of rural residents’ permanent population in the total regional population as an indicator of labor return. Column (2) of Table 4 shows that the regression coefficient of digital technology on the labor return rate is 0.00246 (p < 0.001, t = 5.16), indicating that a 1-unit increase in digital technology leads to a 0.25% increase in the labor return rate. Specifically, digital technology has significantly increased migrant workers’ willingness to return to their hometowns for employment and entrepreneurship through information dissemination and skill training. Short video platforms (such as Douyin and Kuaishou) provide remote employment information matching services for rural labor, while blockchain technology reduces labor mobility barriers through the digitization of academic credentials and skill certificates. This “digital talent attraction” effect manifests as a gradual return of labor from cities to specialty industrial towns in the context of industry-town integration. Assuming H3 is verified.

The significant positive impact of digital technology on external export demand highlights its crucial role in expanding international markets. By optimizing production processes and enhancing supply chain responsiveness, digital technology significantly boosts a company’s international competitiveness. In this paper, we use county-level export volumes to represent external export demand. To ensure consistent measurement units, we apply logarithmic transformation to the export volumes of each county. As shown in Column (3) of Table 4 , the regression coefficient of digital technology on external export demand is 0.00221 (p < 0.01, t = 3.78), indicating that for every 1-unit increase in digital technology, the contribution rate of external demand to industrial-urban integration increases by 0.2%, thereby validating the leverage effect of digital technology in driving external demand. Assuming H4 is verified.

Note: ***, ** and * indicate significance levels of 1%, 5%, and 10%, respectively. The values in parentheses are t-values after controlling for robustness.

Digital technology can break through the path dependence of traditional industrial development in impoverished areas by optimizing resource allocation and reducing information asymmetry, thereby creating a “digital catch-up” effect. The promotional role of digital technology in the integration of industry and townships exhibits significant differentiation between impoverished and non-impoverished areas. This paper collects the list of 832 national-level impoverished counties nationwide, takes the median value, and classifies regions with fewer impoverished counties than the median as non-priority impoverished regions, while regions with more impoverished counties than the median are classified as priority impoverished regions. According to the regression results in columns (1) and (2) of Table 5 , the digital technology coefficient for key poverty-stricken areas is 0.0594 (t = 4.66***, p < 0.001), while that for non-key poverty-stricken areas is only 0.00343 (t = 0.31, p > 0.1). This indicates that the enabling effect of digital technology on the integration of industry and township development is more significant in key poverty-stricken areas than in non-key poverty-stricken areas.

Digital technology lowers barriers to entry through inclusive access, thereby reinforcing the comparative advantages of low-income regions. Based on this, this paper categorizes regions into those with lower and higher resident income levels based on the median resident income level at the end of the year. As shown in columns (3) and (4) of Table 5 , the digital technology coefficient for regions with lower resident income is 0.0515 (t = 3.98***, p < 0.001), while the coefficient for high-income regions is only 0.0104 (t = 1.00, p > 0.1). This gradient difference confirms the pattern that digital technology is more likely to benefit low-income groups, and the impact of digital technology on town-village integration exhibits a significant “pro-poor” characteristic.

Note: ***, ** and * indicate significance levels of 1%, 5%, and 10%, respectively. The values in parentheses are t-values after controlling for robustness.

This chapter conducts theoretical analysis and proposes research hypotheses, then uses panel data at the county level in China from 2018 to 2020 to empirically test the enabling effects of digital technology on the integration of industry and towns. A series of robustness analyses are conducted to validate the empirical results of this chapter. The research findings indicate: First, digital technology can effectively enable the integrated development of industry and towns, and this result remains reliable after a series of robustness analyses. Second, digital technology can promote the symbiotic development of industry-town integration through the wealth accumulation effect of residents, the labor return effect, and the external demand stimulation effect. Third, heterogeneity analysis indicates that the inclusiveness of digital technology makes its enabling effect on the symbiotic development of industry-town integration in key poverty-stricken areas and regions with lower resident income levels more significant. The impact of digital technology on the symbiotic development of industry-town integration exhibits a pronounced “pro-poor” characteristic.

The “pro-poor” characteristics presented in the heterogeneity analysis of this study merit further exploration into the underlying mechanisms. This paper suggests that this “pro-poor” effect may stem from the combined effects of the following two aspects:

1) The “catch-up” potential created by the relatively low baseline level of digitalization and industrial integration development. Key poverty-stricken counties and low-income areas often face development bottlenecks such as weak infrastructure, a single industrial base, a shortage of talent, and limited market access. Their initial levels of digitalization and the degree of integration between industry and town are generally low. A lower starting point means there is greater room for marginal improvement. When digital technology intervenes as a powerful exogenous driving force, the effects it brings, such as relative efficiency improvement, market connection expansion, and accelerated factor flow, are more likely to produce significant “marginal incremental” effects in these regions with lower starting points. This high marginal return brought by a low starting point is a direct manifestation of the “catch-up” potential and an important source of “pro-poor” characteristics.

2) Digital technology possesses unique advantages in overcoming specific obstacles in low-income areas. The inherent characteristics of digital technology enable it to effectively break through many structural obstacles faced in the traditional development path of impoverished areas. Crossing geographical isolation, the Internet and mobile communication technology have broken the constraints of physical space on information acquisition and market participation in remote areas, enabling impoverished areas to directly access national and even global markets. By reducing transaction costs, digital platforms eliminate intermediary links, thereby reducing transaction costs, search costs, and trust costs for producers in impoverished areas, enhancing their market bargaining power and earnings. Addressing shortcomings in human capital, digital inclusive finance, online education, and remote services can alleviate the shortage of professional talents and high-quality services in impoverished areas to a certain extent, empowering local residents and operators. Stimulating endogenous innovation, low-cost digital tools provide residents in impoverished areas with low-threshold channels to showcase local resources and engage in innovation and entrepreneurship, helping to activate endogenous dynamics.

Therefore, digital technology does not merely “appear” effective because of the low starting point in poor areas. As a general-purpose technology, it possesses breakthrough capabilities in overcoming spatial barriers, lowering the threshold for market participation, providing inclusive services, and empowering small and micro entities. This makes it uniquely and effectively adaptable in addressing core issues that plague low-income areas, such as market failures, information asymmetry, and resource scarcity. The high alignment between this technological characteristic and the key bottlenecks in low-income areas is the underlying reason driving the “pro-poor” effect.

To address the structural contradiction of “having towns without industries” and “having industries without towns”, and to fully leverage the pro-poor empowering role of digital technology in the integration of industries and towns, it is urgent to build a precise empowering system centered around digital technology.

1) Targeting the pro-poor approach: Building a precise delivery system for digital infrastructure in impoverished areas

Implement differentiated digital infrastructure layouts targeting the prominent “pro-poor” effects in impoverished counties and low-income areas. Prioritize the enhancement of digital infrastructure and localization services in impoverished areas, and give priority to the construction of agricultural product blockchain traceability centers and smart warehousing nodes in key impoverished counties. By investing preferentially in hardware such as 5G base stations and cold chain logistics data centers, and supporting township-level “digital service stations” to provide skills training, we aim to overcome the dilemma of “emphasizing hardware over services” in underdeveloped areas.

2) Breaking through the empowerment loop: activating the chain reaction of the three major mechanisms

Deeply activate the three mechanisms of residents’ income increase, labor force return, and external demand stimulation. Leverage e-commerce platforms to build regional agricultural product brands to expand the wealth accumulation effect, and integrate scattered farmers by establishing “one town, one product” live streaming bases, transforming the income increase effect of e-commerce into investment in industries such as cold chain logistics. At the same time, set up digital skill subsidies for returning home employment and training centers in industrial parks, utilizing the “digital translation” function of labor force return to optimize human capital allocation; and implement a “digital apprenticeship system” in areas with weak labor force return, where returning entrepreneurs will train apprentices and provide employment promotion subsidies. For export-oriented short-board counties, deepen the digital upgrade of their export-oriented industries, enhance export response efficiency by building cross-border digital supply chain platforms and blockchain traceability systems, provide low-interest digital transformation loans for county export enterprises, strengthen the supporting role of blockchain traceability technology for export premiums, and amplify the driving effect of external demand markets on the integration of production and towns.

3) Innovative institutional supply: Developing reform tools tailored to address structural contradictions

In response to the significant empowering advantages of digital technology in low-income areas, it is necessary to break through traditional institutional bottlenecks. A hierarchical fiscal incentive system should be established, implementing a central and local funding matching mechanism for digital infrastructure in impoverished counties. This requires local investment to be no less than central funding, and strictly restricts the use of matching funds for digital service operations, preventing a disconnect between hardware investment and talent cultivation. We should deepen inclusive financial reforms, allowing farmers to replace traditional collateral with new credit assets such as e-commerce business data and digital skill certifications. Special low-interest loan products should be developed to effectively alleviate financing constraints for small and micro business entities. Innovative land element allocation should be implemented, prioritizing the allocation of land for digital industries in county development zones. Land price discounts should be offered for the construction of data centers, cloud factories, and other facilities using existing land, promoting the agglomeration of digital elements and the restructuring of industrial space.