The impact of digital infrastructure construction on the depressive symptoms among middle-aged and older adults: evidence from China

Abstract

The “Broadband China” strategy has enhanced digital infrastructure through urban fiber upgrades, rural broadband expansion, and improved network performance. This research employs the multi-period difference-in-differences (DID) model using 2011–2020 CHARLS data to explore underlying mechanisms. Findings indicate digital infrastructure reduces depression, mediated primarily by strengthened old-age services and social well-being. Heterogeneity analysis reveals stronger effects among urban residents, individuals aged 45–60, and those with higher cognitive abilities. By demonstrating the mental health benefits of digital infrastructure through the “Broadband China” case, this study provides empirical evidence for developing nations to leverage technological investments for public health objectives, particularly in aging populations. The results underscore the importance of integrating digital development with social support systems to maximize health outcomes.

Introduction

In an era marked by escalating social pressures, rising living costs, and rapid technological transformations reshaping labor patterns, depression has emerged as a critical global mental health challenge [1, 2]. Characterized by its insidious onset, high prevalence, prolonged treatment courses, and frequent comorbidities, depression imposes substantial barriers to individuals’ personal fulfillment and impedes the sustainable development of socioeconomic systems worldwide. According to the World Health Organization, approximately 280 to 322 million people globally suffer from depression, representing around 3.8% to 4.4% of the global population, with older adults disproportionately affected, particularly in countries experiencing rapid demographic aging.

China, home to one of the largest aging populations globally, is undergoing profound demographic shifts [3]. National Bureau of Statistics data from 2021 indicate that in 2020, the population aged 60 and above reached 264 million, constituting 18.7% of the total population, with projections exceeding 30% by 2035 [4, 5]. Within this context, the mental health of middle-aged and older adults represents a critical public health issue. Studies report that the prevalence of depressive symptoms in this demographic in China may reach as high as 30.2%, substantially exceeding the global average [6]. Multifactorial pressures, including evolving social roles, family economic strains, and deteriorating health status not only degrade quality of life but also reduce social participation, thereby exerting profound impacts on social productivity, consumption behaviors, and healthcare expenditures [7, 8].

Unlike other aging-associated chronic conditions, depression is notably more responsive to non-pharmacological interventions such as social support, psychological counseling, and environmental modifications [9]. Consequently, identifying effective mitigation strategies to alleviate depressive symptoms among middle-aged and older adults is a pressing global health priority, with particular urgency for aging societies worldwide, including China. Such efforts are essential for addressing demographic shifts, optimizing health resource allocation, and promoting sustainable socioeconomic development. The determinants of depressive symptoms in middle-aged and older adults are complex and multifaceted. Recent studies underscore the influence of social capital, income disparities, living environments, and healthcare accessibility [10, 11]. Notably, individuals residing in areas with poor environmental quality and insufficient community infrastructure exhibit elevated depression rates, while access to abundant medical resources correlates with reduced prevalence [12].

However, the effectiveness of these factors often hinges on a critical threshold condition: the development of digital infrastructure. In recent years, a growing body of research has begun to explore the role of digital technology in mental health interventions [13]. On the one hand, research has found that providing mental health education through digital platforms can help raise awareness of mental health issues and promote their prevention [14] and using digital tools for mental health screening and assessment can help users identify potential mental health issues and provide a basis for subsequent interventions [15]. On the other hand, a lot of research highlight the transformative potential of digital technologies in mental health care across diverse populations. Among adolescents, digital platforms offer accessible and varied avenues for mental health support, facilitating engagement with educational content and therapeutic interventions via online games and social media, thereby enhancing mental health literacy and early intervention [16]. For individuals with mental illness, these technologies enable improved self-management, optimized treatment adherence, and a consequent elevation in quality of life. Moreover, telemedicine and online counseling effectively overcome logistical challenges such as transportation barriers and scarcity of local resources, expanding access for residents in remote or underserved areas [17]. Advances in digital infrastructure have been pivotal in democratizing access to mental health information, mitigating psychological distress on a broad scale. Crucially, the integration of digital tools with mental health screening protocols transcends traditional temporal and spatial limitations, offering a scalable and flexible model of care [18]. Compared with conventional face-to-face therapy, digital interventions reduce costs and lower entry barriers, enhancing economic efficiency and user convenience by enabling around-the-clock, on-demand access without the need for prior scheduling [19]. Together, these developments signify a paradigm shift towards more inclusive, cost-effective, and accessible mental health services. Nevertheless, the specific mechanisms by which digital infrastructure, especially high-speed broadband and wireless networks, affects mental health have not been fully studied, and existing research has focused little on how digital technology affects the mental health of older adults over the long term.

Against the backdrop of global advancements in new infrastructure and persistent public health challenges, this study examines the implementation of “Broadband China” strategy as a representative policy of digital infrastructure construction. By analyzing its impact on depressive symptoms of middle-aged and older adults, this research aims to contribute valuable insights to global health governance and inform policies in other aging societies. The strategy provides a technological foundation for improving public services and enhancing quality of life. By accelerating the integration of the internet with healthcare, social security, and other services, it facilitates the efficient allocation of public resources.

This study makes several key contributions. Firstly, it analyses the socioeconomic mechanisms through which digital infrastructure construction enhances public well-being, particularly its impact on depressive symptoms of middle-aged and older adults. This aligns with broader objectives of maximizing health benefits under resource constraints, evaluating policy interventions, and promoting health equity. This resonates with the current focus of health economics and the United Nations Sustainable Development Goal 17. Secondly, by examining the influence of network equipment, infrastructure, and services on depressive symptoms, this study enhances extant perspectives and identifies a logical precursor to the mechanisms linking mental health to its direct determinants, thereby complementing prior research [20]. Finally, leveraging the quasi-experimental nature of the “Broadband China” pilot program, this study employs the multi-time point difference-in-differences (DID) approach to assess policy effects, addressing endogeneity concerns and providing robust causal evidence.

Theoretical framework and research hypotheses

The “Broadband China” strategy has significantly enhanced the quality of life for middle-aged and older adults by directly expanding digital infrastructure, playing a positive role in alleviating depressive symptoms. Specifically, the impact of digital infrastructure construction on the mental health of middle-aged and older adults, particularly depressive symptoms, manifests in the following ways:

First, digital infrastructure construction has accelerated internet accessibility, helping to reduce feelings of loneliness of middle-aged and older adults, a major contributing factor to depressive symptoms. Fortunately, the widespread adoption of digital technologies, including social media and video calls, facilitates more frequent interactions with family and friends, mitigating the loneliness and depressive emotions caused by social isolation [21, 22]. Second, digital infrastructure construction has expanded channels for information dissemination, promoting cognitive improvements of middle-aged and older adults. The proliferation of the internet enables this demographic to access a wealth of information, including educational, cultural, and recreational content. Such information not only enhances their knowledge but also stimulates curiosity and interest in the world, improving their cognitive function. Improved cognitive function has been shown to significantly reduce the risk of depressive symptoms. Studies have demonstrated that older adults who access health-related information online are better equipped to understand health issues and adopt healthier behaviors, thereby lowering their risk of depression [23]. Building on the above analysis, this study proposes the following hypothesis:

• H1: Digital infrastructure construction can alleviate depressive symptoms of middle-aged and older adults

The widespread deployment and adoption of new digital infrastructures create a solid network foundation that supports the growth of high-tech enterprises and may contribute to improvements in mental well-being among middle-aged and older adults, potentially through mechanisms such as increased internet penetration and enhanced urbanization. Besides, the widespread adoption of digital technology has enhanced the feasibility of smart elderly care, including intelligent monitoring, remote care, and personalized health management, which have been proven effective in alleviating mental health issues among older adults [24, 25]. Therefore, new digital infrastructures have enhanced the overall development potential of social old age care services, laying the groundwork for further alleviating depressive symptoms of middle-aged and older adults. Specifically, digital infrastructure construction may influence the mental health of middle-aged and older adults through two key mechanisms. Firstly, the increased internet penetration rate has lowered the barriers to accessing and utilizing digital devices and services, this has enhanced their access to and proficiency with social tools such as smartphones, video calling software, and social media. The improvement in digital access has expanded the scope of interaction between middle-aged and older adults and their relatives, neighbors, and interest groups, helping to rebuild their social networks. Frequent social activities help to give middle-aged and older adults positive emotions, thereby alleviating loneliness, a key factor in the risk of depression, and reducing their risk of developing depressive symptoms [26]. Secondly, the continuous development of digital infrastructure has provided strong support for the improvement of regional urbanization levels since efficient information networks and logistics systems have enhanced the efficiency of information and material circulation between cities, promoted industrial agglomeration and resource coordination and strengthened cities’ population carrying capacity and ability to attract factors of production, thereby accelerated the development of urbanization. As urbanization levels rise, city governments gain stronger fiscal and planning capabilities, enabling them to provide more abundant public cultural spaces, including urban parks, green spaces, and community activity centers [3, 27]. These spaces offer middle-aged and older adults’ opportunities for emotional regulation. Existing research indicates that well-designed public cultural spaces significantly help alleviate depression and anxiety among urban residents, thereby enhancing their mental health levels [28]. This provides a useful reference framework for the application of digital tools in managing mental health in this population. Based on the above analysis, this paper proposes the following hypothesis:

• H2: Digital infrastructure construction alleviates positively depressive symptoms of middle-aged and older adults by enhancing the development potential of social elderly care services, particularly through increased internet penetration and improved level of regional urbanization

The advancement of new digital infrastructure construction has not only accelerated economic development but also significantly enhanced social wellbeing across multiple dimensions. In the long term, the improvement of digital infrastructure has optimized income distribution, thereby refining the structure of social wealth allocation. Studies have shown a significant positive correlation between increased internet access and household income growth [29, 30]. The widespread adoption of broadband has driven the development of the digital economy, boosting overall societal productivity and wealth accumulation [31]. Higher societal income levels enable governments and markets to allocate more resources to elderly care initiatives, such as expanding pension coverage, improving healthcare services, and optimizing elderly welfare policies, thereby enhancing the economic security of middle-aged and older adults [32]. Besides, the internet has broken down temporal and spatial barriers to information flow, lowering the threshold for accessing healthcare services and enabling more people to benefit from remote medical care and e-government services [33], and internet use can effectively increase social engagement and reduce feelings of social isolation of older adults, thereby alleviate their psychological stress to some extent. These advancements have optimized public services and living environments, with smart communities, telemedicine, and on-demand services improving resource allocation efficiency and enhancing the quality of life for middle-aged and older adults. Specifically, the convenience of information access, the availability of healthcare services, and the expansion of social networks, all facilitated by digital infrastructure, contribute to higher subjective well-being. Based on the above analysis, this study proposes the following hypothesis:

• H3: Digital infrastructure construction alleviates positively depressive symptoms of middle-aged and older adults by enhancing social wellbeing, particularly through increased household income, diversity of social activities, and life satisfaction

The specific logic is analysed as shown in Fig. 1.

Fig. 1.

Mechanism analysis of the digital infrastructure construction to alleviate the depressive symptoms of middle-aged and older adults

Empirical model design

Data source

The data for this study are derived from five waves (2011, 2013, 2015, 2018, and 2020) of panel data from the China Health and Retirement Longitudinal Study (CHARLS)1, supplemented by relevant data from the China Urban Statistical Yearbook. The CHARLS, a flagship data collection project hosted by the National School of Development at Peking University and implemented by the Institute of Social Science Survey at Peking University, is designed to gather high-quality microdata representative of middle-aged and older adults households and individuals in China. The survey tracks approximately 20,000 respondents across 28 provinces (autonomous regions and municipalities) in China, providing comprehensive longitudinal insights into the aging population.

For this study, a balanced panel dataset was constructed using the complete data from the five waves of CHARLS during the period from 2011 to 2020. The final analysis includes a total sample of 16,880 individuals (84,400 observations). This study was conducted in accordance with ethical principles for research involving human participants. Approval was obtained from the appropriate Institutional Review Board prior to data collection. All participants provided informed consent prior to their involvement.

Variable selection

Assessment of the depressive symptoms of middle-aged and older adults

The dependent variable in this study is the level of depression of middle-aged and older adults. This variable is measured using the CESD-10 (10-item Center for Epidemiological Studies Depression Scale) included in the CHARLS questionnaire dataset. Depression is one of the core modules in the CHARLS questionnaire, comprising 10 questions: 8 negative statements and 2 positive statements. Each question is scored on a scale of 0 to 3, with a total possible score of 30. The CESD-10 scale aggregates the scores from these 10 questions, where a higher score indicates more severe depressive symptoms, while a lower score suggests milder symptoms. Similar scale has been widely used in large-scale social surveys conducted by international organizations, such as the Survey of Health, Ageing and Retirement in Europe (SHARE) and the Health and Retirement Study (HRS) in the United States.

Assessment of the digital infrastructure construction

The core explanatory variable in this study is the implementation of the “Broadband China” policy (hdid), which is treated as an exogenous shock representing the improvement and upgrading of digital infrastructure. This policy effect is evaluated using the multi-time-point difference-in-differences (DID) approach. The “Broadband China” pilot policy was initiated in 2014, with pilot cities added in batches in 2015 and 20162. The list of specific cities for each wave is shown in Table 1.

Table 1.

The list of specific cities for each wave

Year	“Broadband China” Pilot Cities (City Clusters)
2014	Direct-controlled municipalities and city clusters: Beijing, Tianjin, Shanghai, Chang-Zhu-Tan City Cluster.
	Other cities and provincially administered counties: Shijiazhuang, Dalian, Benxi, Yanbian Korean Autonomous Prefecture, Harbin, Daqing, Nanjing, Suzhou, Zhenjiang, Kunshan, Jinhua, Wuhu, Anqing, Fuzhou (including Pingtan), Xiamen, Quanzhou, Nanchang, Shangrao, Qingdao, Zibo, Weihai, Linyi, Zhengzhou, Luoyang, Wuhan, Guangzhou, Shenzhen, Zhongshan, Chengdu, Panzhihua, Aba Tibetan and Qiang Autonomous Prefecture, Guiyang, Yinchuan, Wuzhong, Aral.
2015	Taiyuan, Hohhot, Ordos, Anshan, Panjin, Baishan, Yangzhou, Jiaxing, Hefei, Tongling, Putian, Xinyu, Ganzhou, Dongying, Jining, Dezhou, Xinxiang, Yongcheng, Huangshi, Xiangyang, Yichang, Shiyan, Suizhou, Yueyang, Shantou, Meizhou, Dongguan, Jiangjin District (Chongqing), Rongchang District (Chongqing), Mianyang, Neijiang, Yibin, Dazhou, Yuxi, Lanzhou, Zhangye, Guyuan, Zhongwei, Karamay.
2016	Yangquan, Jinzhong, Wuhai, Baotou, Tongliao, Shenyang, Mudanjiang, Wuxi, Taizhou, Nantong, Hangzhou, Suzhou, Huangshan, Ma’anshan, Ji’an, Yantai, Zaozhuang, Shangqiu, Jiaozuo, Nanyang, Ezhou, Hengyang, Yiyang, Yulin, Haikou, Jiulongpo District, Beibei District, Ya’an, Luzhou, Nanchong, Zunyi, Wenshan Zhuang and Miao Autonomous Prefecture, Lhasa, Nyingchi, Weinan, Wuwei, Jiuquan, Tianshui, Xining.

The selection of pilot cities was based on specific criteria established by the Ministry of Industry and Information Technology and relevant provincial authorities. These criteria included factors such as regional distribution, urban-rural classification, local GDP levels, and the capacity of telecommunications enterprises, ensuring a representative and feasible rollout across diverse geographic and economic conditions. As such, while not purely random, the selection process was structured and policy-driven to achieve broad coverage and maximize strategic impact. This study focuses on the impact of digital infrastructure construction on depressive symptoms of middle-aged and older adults during its implementation period. Given the variations in implementation timelines and coverage across different cities, the construction of the dummy variable follows this strategy: a dummy variable is generated based on the timing and scope of the pilot policy. If a prefecture-level city where the sample resides implemented the “Broadband China” strategy, the sample is considered systematically covered. It is important to note that the CHARLS data used in this study are derived from national surveys conducted in 2011, 2013, 2015, 2018, and 2020. Therefore, this research specifically examines the effectiveness of the “Broadband China” strategy during this defined period. The final sample used in this study includes data from 112 cities, among which 44 are pilot cities.

Assessment of mediating variables

Based on the aforementioned hypotheses, digital infrastructure construction may alleviate depressive symptoms of middle-aged and older adults by stimulating the potential of social elderly care services and enhancing people’s wellbeing. To examine these pathways, this study investigates five specific mechanisms and constructs an indicator framework for the mechanism variables as follows:

Internet coverage, a critical measure of digital infrastructure penetration, is quantified at the prefecture-level city scale using the logarithm of internet users per 1,000 people (num-net), sourced from the National Bureau of Statistics of China3 panel dataset. This metric captures the spatial variation in internet facility deployment across cities, providing a robust indicator of connectivity. Urbanization levels are assessed by computing the urbanization rate (cityrate), defined as the ratio of urban resident population to the total population, with the non-agricultural resident population serving as a proxy where urban data are unavailable; these data are annually matched from National Bureau of Statistics of China. A higher cityrate reflects greater urban concentration, consistent with economic development patterns.

Per capita household income of middle-aged and older adults, derived from the “Household income and expenditure” section of the CHARLS questionnaire, is expressed as its natural logarithm (lninc) to normalize scale effects, offering a reliable measure of economic well-being. Life satisfaction is evaluated using responses to the CHARLS item “Overall, are you satisfied with your life?” with a five-point scale from 1 (“very dissatisfied”) to 5 ( “very satisfied”), yielding the indicator (satlife) to encapsulate subjective welfare among this demographic. Diversity of social activities, reflecting engagement in community and interpersonal interactions, is quantified as the average frequency of participation (soc) in diverse activities4, providing insight into the social capital and quality of life of middle-aged and older adults across sampled regions.

In summary, the “macro-level potential” is measured using two indicators: the logarithm of internet users per 1,000 people (num-net) and the urbanization rate (cityrate). The “micro-level welfare” is measured using three indicators: the logarithm of per capita household income (lninc), diversity of social activities (soc), and life satisfaction (satlife).

Assessment of covariates

To control for the influence of other variables, this study includes individual-level, household-level, and regional-level characteristics as control variables. These include age (age), gender (gender), education level (edu), marital status (marry), household size (size), the number of living children (hchild), the GDP of the prefecture-level city where the sample resides (gdp, in 10,000 yuan), and the year-end population of the prefecture-level city (pop, in 10,000 people).

Model setting

The implementation of the “Broadband China” strategy provides a quasi-natural experiment setting for this study, which aims to examine the impact of digital infrastructure construction on depressive symptoms of middle-aged and older adults. To more clearly observe the “net effect” of the policy on the treatment group, this study employs a time-varying difference-in-differences (DID) model for empirical analysis. In line with the requirements of the multi-time-point DID model, we construct two sets of dummy variables. The first set is the policy dummy variable (Treat), where 0 represents the control group (unaffected by the policy) and 1 represents the treatment group (affected by the policy). The second set is the time dummy variable (Time), where 0 represents the pre-policy period and 1 represents the post-policy period. By comparing the differences in outcomes between the treatment and control groups before and after the policy implementation, this study systematically evaluates the impact of digital infrastructure construction on depressive symptoms of middle-aged and older adults. To capture the intervention effect of the policy on the treatment group, this study introduces the interaction term between the policy dummy variable (Treat) and the time dummy variable (Time) into the model. To enhance the accuracy of the estimation results and mitigate potential biases caused by omitted variables and other endogeneity issues, the model is specified as follows:

1

In Eq. (1): is the dependent variable, representing depressive symptoms of middle-aged and older adults. is the core explanatory variable, capturing the interaction effect of the digital infrastructure construction. is the coefficient of primary interest, reflecting the policy’s effect. is a vector of control variables, including individual- and household-level characteristics, with representing their coefficients. and represent time fixed effects and individual fixed effects, respectively. is the error term.

Results and analysis

Descriptive statistical analysis

The descriptive statistics of the key variables are presented in Table 2. The dependent variable, depressive symptoms (cesd10), has a mean score of 8.228, suggesting that most respondents experience mild depressive symptoms, with a standard deviation of 6.293 indicating substantial variability across the sample. The independent variable, hdid, has a mean value of 0.210, reflecting that only a small proportion of the sampled cities implemented the “Broadband China” strategy.

Table 2.

Descriptive Statistics

Type	Variable	Definition	Mean	Std.Dev.	Min	Max
Dependent variable	cesd10	Depression symptom score, used to evaluate the mental health level of middle-aged and older adults	8.228	6.293	0.000	30.000
Independent variable	hdid	The implementation situation of the “Broadband China” policy	0.210	0.408	0.000	1.000
Mediating variables	num-net	The logarithm of internet users per 1,000 people	6.538	0.984	4.249	8.824
	cityrate	City rate, it is calculated based on the proportion of the urban population to the total population.	0.532	0.139	0.245	0.893
	lninc	The logarithmic form of household income, calculated based on the annual income of the respondents’ households	8.501	1.611	4.007	11.468
	soc	The diversity of the respondents’ social activities	0.789	0.980	0.000	4.000
	satlife	The respondents’ score for life satisfaction	3.226	0.781	1.000	5.000
Other variables	word frequency number	The frequency of terms related to new digital infrastructure in government work reports	12.200	10.414	0.000	78.000
	word frequency ratio	The proportion of terms related to new digital infrastructure in government work reports	0.002	0.001	0.000	0.006
Control variables	age	The age distribution of the respondents	61.135	10.048	45.000	87.000
	gender	Gender distribution of respondents	0.481	0.500	0.000	1.000
	edu	The educational background of the respondents	2.040	1.108	1.000	5.000
	marry	The marital status of the respondents	0.858	0.350	0.000	1.000
	size	The size of the respondents’ families, that is, the number of family members included	3.162	1.567	1.000	8.000
	hchild	The number of children in the respondents’ families	2.587	1.367	0.000	7.000
	gdp	The GDP level of the prefecture-level city where the sample is located (100 million yuan)	3280.000	3900.000	290.000	22900.000
	pop	The Population of the prefecture-level city where the sample is located (10000)	620.469	440.951	130.900	3371.840

For the macro-level mediating variables, the logarithm of internet users per 1,000 people (num-net) ranges up to 8.824, with a mean of 6.538, demonstrating relatively high internet penetration rates in most sampled cities and a standard deviation of 0.984 indicating generally high levels of informatization. The urbanization rate (cityrate) has a maximum of 0.893 and a mean of 0.532, suggesting that most cities are in the middle or later stages of urbanization, with a standard deviation of 0.139 highlighting moderate variation across cities. For the micro-level mediating variables, the logarithm of per capita household income (lninc) has a mean of 8.501 and a standard deviation of 1.611, reflecting relatively high income level and equitable wealth distribution among the sample. Diversity of social activities (soc) has a mean of 0.789 and a standard deviation of 0.980, suggesting that respondents engage in relatively limited social activities. Life satisfaction (satlife) has a mean of 3.226, indicating that most respondents report a generally positive attitude toward their lives.

Relationship between digital infrastructure construction and depressive symptoms

The Broadband China Policy serves as the overarching framework for China’s digital infrastructure construction initiative. Therefore, this paper preliminarily explores the relationship between digital infrastructure development and depressive symptoms among the middle-aged and older adults. This paper uses the frequency and proportion of terms related to new digital infrastructure in government work reports5 as indicators to measure the level of infrastructure development in prefecture-level cities. Table 3 presents the results of the relationship between these two variables. Model (1) and Model (3) report the baseline regression results of the total word frequency (word frequency number) and the proportion of word frequency (word frequency ratio) related to new digital infrastructure on depressive symptoms. Model (2) and Model (4) present the regression results after including additional control variables. The estimation results show that, regardless of whether control variables are included, both the total word frequency and the proportion of word frequency related to new digital infrastructure significantly reduce the CESD-10 scores of middle-aged and older adults. These findings confirm that middle-aged and older adults are positively influenced by the externalities of the new digital economy. Under the high-quality development goals of China’s economy, prefecture-level governments have increasingly prioritized the construction of new digital infrastructure. These top-level designs optimize the allocation of modern technological resources, promoting the healthy development of new digital infrastructure. This progress meets the diverse needs of middle-aged and older adults for social interaction and cognitive engagement, mitigates issues such as cognitive decline and negative emotions, and ultimately reduces the incidence of depressive symptoms of this population. Based on this result, this article will further explore the net effect of the Broadband China policy on the depressive symptoms of the middle-aged and older adults.

Table 3.

Regression analysis of the relationship between digital infrastructure construction and depressive symptoms in the middle-aged and older adults

VARIABLES	Model (1)	Model (2)	Model (3)	Model (4)
	cesd10	cesd10	cesd10	cesd10
word frequency number	-0.0518***	-0.0218***
	(0.0034)	(0.0034)
word frequency ratio			-387.2108***	-173.6362***
			(28.1874)	(27.5895)
age		0.0057		0.0056
		(0.0041)		(0.0041)
gender		-1.5622***		-1.5609***
		(0.0698)		(0.0698)
edu		-0.8227***		-0.8246***
		(0.0322)		(0.0322)
marry		-1.3315***		-1.3329***
		(0.1116)		(0.1116)
size		-0.1000***		-0.1002***
		(0.0185)		(0.0185)
hchild		0.1396***		0.1406***
		(0.0305)		(0.0305)
gdp		-0.0001***		-0.0001***
		(0.0000)		(0.0000)
pop		0.0005**		0.0005**
		(0.0002)		(0.0002)
Constant	8.7774***	11.7040***	8.8212***	11.7679***
	(0.0562)	(0.3052)	(0.0620)	(0.3062)
R-squared	0.031	0.093	0.030	0.093
Adjust R-squared	0.032	0.092	0.031	0.092
Observations	84,400	84,400	84,400	84,400
IndividualFE	NO	YES	NO	YES
CityFE	YES	YES	YES	YES
YearFE	YES	YES	YES	YES

*, **, and *** represent 10%, 5%, and 1% significance levels, respectively; Robust standard error based on City clustering are in parentheses in the table

Benchmark regression analysis

The implementation of the “Broadband China” strategy represents a significant initiative in China’s efforts to accelerate the development of digital infrastructure. This study employs the multi-time point difference-in-differences (DID) model specified in Eq. (1) to estimate the impact of digital infrastructure construction on the mental health of middle-aged and older adults. Table 4 primarily reports the estimated coefficients of the core explanatory variable (Table 4). Model (5) in the table does not include individual- and household-level characteristic variables, while Model (6) to Model (8) control for these variables. Additionally, Model (6) and Model (7) incorporate individual fixed effects and time fixed effects, respectively. Model (8) presents the final estimation results after introducing a full set of control variables and controlling for both time and individual fixed effects.

Table 4.

Benchmark regression Results

VARIABLES	Model (5)	Model (6)	Model (7)	Model (8)
	cesd10	cesd10	cesd10	cesd10
hdid	-0.6693***	-0.2255***	-0.1589***	-0.2672***
	(0.0645)	(0.0715)	(0.0617)	(0.0660)
age		0.0158***	0.0021	0.0052*
		(0.0036)	(0.0028)	(0.0028)
gender		-1.7122***	-1.5416***	-1.5605***
		(0.0589)	(0.0460)	(0.0457)
edu		-0.8432***	-0.8422***	-0.8305***
		(0.0285)	(0.0216)	(0.0220)
marry		-1.2458***	-1.3696***	-1.3087***
		(0.0898)	(0.0705)	(0.0701)
size		-0.1143***	-0.0777***	-0.0991***
		(0.0214)	(0.0150)	(0.0152)
hchild		0.1259***	0.1735***	0.1461***
		(0.0271)	(0.0197)	(0.0203)
gdp		-0.0001***	-0.0002***	-0.0001***
		(0.0000)	(0.0000)	(0.0000)
pop		0.0006***	0.0010***	0.0005***
		(0.0001)	(0.0001)	(0.0001)
Constant	8.3056***	11.0052***	11.5424***	11.5458***
	(0.0261)	(0.2673)	(0.2028)	(0.2103)
R-squared	0.031	0.094	0.077	0.094
Adjust R-squared	0.032	0.094	0.078	0.094
Observations	84,400	84,400	84,400	84,400
IndividualFE	YES	YES	YES	YES
CityFE	YES	YES	NO	YES
YearFE	YES	NO	YES	YES

*, **, and *** represent 10%, 5%, and 1% significance levels, respectively; Robust standard error based on City clustering are in parentheses in the table

The results indicate that digital infrastructure construction has a statistically significant negative effect on depressive symptoms of middle-aged and older adults. The coefficients across Models (5) to (8) range from -0.1589 to -0.6693, demonstrating that the implementation of the “Broadband China” policy significantly reduces depressive symptoms, regardless of whether control variables are included. This finding underscores the robustness of the estimation results. Based on this analysis, H1 is supported.

Model validity testing

Parallel trends test

To validly apply the Difference-in-Differences (DID) methodology for evaluating the policy effects, it is essential that the treatment group (subject to the policy intervention) and the control group (not subject to the intervention) exhibit parallel time trends prior to implementation, thereby fulfilling the parallel trends assumption. More specifically, the trajectories of depressive symptoms among middle-aged and older adults in both groups should not differ before the rollout of the “Broadband China” strategy. To verify this assumption, we employed an event study framework to estimate dynamic treatment effects and to assess the parallelism of pre-intervention trends. This approach involved interacting the policy implementation indicator with year-specific dummy variables, using the policy implementation year (2014) as the reference point. Figure 2 displays the estimated coefficients of these interaction terms before and after the policy’s rollout. Our findings show that, prior to the 2014 implementation (in the Pre-3 and Pre-1 periods), the coefficients are statistically indistinguishable from zero, confirming that the treatment and control cities followed similar trends in depressive symptoms, thereby satisfying the parallel trends requirement.

Fig. 2.

Parallel trends test for depressive symptoms

Following the policy implementation (Post-1, Post-4, and Post-6), depressive symptoms among middle-aged and older adults in treatment cities showed significant improvements, with these effects reaching statistical significance. Moreover, the effect of digital infrastructure construction on depressive symptoms is not immediate. Rather, its beneficial impact becomes more evident starting around the fourth year after implementation (2018), indicating a temporal lag in the policy’s influence. This lag is plausibly attributable to the considerable time required for infrastructure projects—from initial planning through construction to full operationalization—before benefits can materialize. Taken together, these results validate the use of our multi-period DID model and corroborate the baseline regression outcomes, confirming the robustness of our empirical strategy.

Placebo test

To examine whether improvements in depressive symptoms of middle-aged and older adults might be due to unobservable factors or random events, this study conducts a placebo test by randomly assigning cities to the treatment group. From the full sample of 16,880 observations, 1,658 were randomly selected as a “pseudo” control group and a dummy policy variable was constructed for re-estimation. Using Monte Carlo simulations, this randomization process was repeated 500 times, yielding 500 regression coefficients and their corresponding p-values for the pseudo core explanatory variable. The kernel density distribution of these coefficients is shown in Fig. 3.

Fig. 3.

Placebo test for depressive symptoms

The results show that the estimated coefficients of the pseudo core explanatory variable are concentrated around zero, with most being statistically insignificant. In contrast, the true core explanatory variable has a coefficient of -0.2672, which is significant at the 1% level. The estimated coefficients of the pseudo core explanatory variable differ significantly from the true estimates in the baseline regression. This suggests that the coefficient estimates in the baseline regression are unlikely to be due to chance, as they represent low probability events in the placebo test. These findings confirm that the estimated effects of the core explanatory variables are not driven by random factors, and demonstrate that the impact of the digital infrastructure construction on depressive symptoms of middle-aged and older adults is genuine and robust.

Endogenous treatment

Endogeneity concerns in this study primarily arise from omitted variables and potential reverse causality. Specifically, regions exhibiting higher levels of depressive symptoms among middle-aged and older adults tend to have lower economic development and less advanced digital infrastructure, making them more probable targets of the Broadband China strategy aligned with China’s high-quality economic development objectives. To mitigate endogeneity bias, we utilize a two-stage least squares (2SLS) approach. We select the average slope of the prefecture-level city as the instrumental variable, reflecting the natural geographic characteristic that directly influences the technical difficulty and cost of constructing digital infrastructure. An effective instrumental variable should meet three strict requirements: exogeneity, relevance, and exclusivity. The slope data reasonably satisfy these requirements because this variable is exogenous and is determined based on geographical location. Therefore, apart from having a correlation through its impact on policy implementation, it has no other relationship with the depressive symptoms themselves. It is also one of the more common instrumental variables used in research. For instance, studies by scholars such as Yun [34] and Qian [35] have all employed this variable for robustness analysis. The average slope therefore satisfies the relevancy, exogeneity, and exclusion restrictions necessary for a valid instrument.

Table 5 reports the results of the 2SLS estimation. Model (9) presents the results of the first stage regression using the instrumental variable. The results show that the Kleibergen-Paap Wald F-statistic is 449.054, which is significantly above the empirical threshold of 10, indicating the absence of weak instrument problems. In addition, the Kleibergen-Paap LM statistic rejects the null hypothesis at the 1% significance level, confirming that the instrumental variable is not unidentified. Model (10) reports the results of the second stage regression using the independent variable. The estimated coefficient of the policy dummy variable is -19.5996, which is statistically significant at the 1% level. The findings demonstrate that, subsequent to the resolution of endogeneity through the utilization of the 2SLS method, the baseline regression findings maintain their validity. This lends further support to the robustness of the empirical conclusions drawn in this study.

Table 5.

Results of instrumental variable approach (Two-stage ols)

VARIABLES	Model (9)	Model (10)
	first stage	second stage
	hdid	cesd10
hdid		-19.5996***
		(1.0656)
Average slope	-0.0406***
	(0.0021)
Control variables	YES	YES
Kleibergen-Paap LM	447.168***
Kleibergen-Paap Wald F	449.054***
Constant	0.2355***	12.2543***
	(0.0020)	(0.2565)
R-squared	0.060	0.060
Adjust R-squared	0.061	0.061
Observations	84,400	84,400
IndividualFE	YES	YES
CityFE	YES	YES
YearFE	YES	YES

*, **, and *** represent 10%, 5%, and 1% significance levels, respectively; Robust standard error based on City clustering are in parentheses in the table

Robustness check

Replacement of dependent variable

This study employs the alternative variable method to conduct robustness checks by replacing the dependent variable with the “self-rated health” score (srh) from the CHARLS project questionnaire. The question provides respondents with two negative options (“very unhealthy” and “somewhat unhealthy”), two positive options (“very healthy” and “somewhat healthy”), and one neutral option (“fair”) to measure self-rated health, based on the ordered categorical variable classification. The total score ranges from 1 to 5, with higher scores indicating better self-rated health. This paper draws on the existing research results and uses “self-rated health” score (srh) as an alternative indicator for depressive symptoms in middle-aged and older adults. The regression results after replacing the core explanatory variable are presented in Table 6. The findings demonstrate that, irrespective of the inclusion of control variables, the development of digital infrastructure significantly alleviates depressive symptoms among middle-aged and older adults, even when the core explanatory variable is replaced with “self-rated health (srh)”. This corroborates the robustness of the baseline regression results in this study.

Table 6.

Robustness test results (Replacement of dependent variable)

VARIABLES	Model (11)	Model (12)
	srh	srh
hdid	0.0521***	0.0128***
	(0.0127)	(0.0013)
Constant	3.0363***	3.5471***
	(0.0040)	(0.0448)
Control variables	No	YES
R-squared	0.002	0.037
Adjust R-squared	0.002	0.038
Observations	84,400	84,400
IndividualFE	YES	YES
CityFE	YES	YES
YearFE	YES	YES

*, **, and *** represent 10%, 5%, and 1% significance levels, respectively; Robust standard error based on City clustering are in parentheses in the table

Exclusion of other policy interferences

The study period covers 2011 to 2020, during which several socioeconomic development policies related to the “Broadband China” strategy were implemented concurrently. These include the Atmospheric Key Control Zone Pilot Policy (AP), aimed at reducing extreme weather events and enhancing climate adaptability for middle-aged and older adults [36]; the Low-Carbon City Pilot Policy (LP), focused on regulating carbon emissions, improving air quality, and mitigating urban heat effects [37]; and the Gigabit City Pilot Policy (GP), which accelerates the construction of digital infrastructure and industrial innovation [38]. This study accounts for the potential confounding effects of these related policies on the baseline regression results6. To address these potential interferences, dummy variables representing each policy were incorporated into the model based on Eq. (1), capturing their possible impacts. The regression results after including these control variables are shown in Table 7. These results indicate that, after controlling for the effects of these policies, the estimated coefficients of the broadband policy remain consistent with the baseline findings, and the core explanatory variable’s impact on the dependent variable remains statistically significant. This reinforces the robustness of the baseline regression conclusions.

Table 7.

Robustness test results (Exclusion of other policy interferences)

VARIABLES	Model (13)	Model (14)	Model (15)	Model (16)
	cesd10	cesd10	cesd10	cesd10
hdid	-0.2613***	-0.2596***	-0.2623***	-0.2467***
	(0.0828)	(0.0822)	(0.0821)	(0.0829)
AP	0.0724			0.0873
	(0.1459)			(0.1455)
LP		0.1277		0.1437
		(0.0927)		(0.0923)
GP			-0.3374***	-0.3426***
			(0.0701)	(0.0697)
Control variables	YES	YES	YES	YES
Constant	11.5413***	11.5313***	11.5829***	11.5617***
	(0.3035)	(0.3035)	(0.3035)	(0.3039)
R-squared	0.094	0.082	0.095	0.104
Adjust R-squared	0.095	0.081	0.097	0.104
Observations	84,400	84,400	84,400	84,400
IndividualFE	YES	YES	YES	YES
CityFE	YES	YES	YES	YES
YearFE	YES	YES	YES	YES

*, **, and *** represent 10%, 5%, and 1% significance levels, respectively; Robust standard error based on City clustering are in parentheses in the table

Replacement of estimation method

Another concern regarding robustness stems from potential errors in the functional form of the model. The Dual Machine Learning model is employed to estimate the impact of digital infrastructure construction on depressive symptoms of middle-aged and older adults, and the Random Forest algorithm is used to re-estimate and solve the key regression steps [39]. The regression results following alteration of the algorithm are presented in Table 8. The initial column of the table reports the regression results following the control of county fixed effects, time fixed effects, and the primary-order terms of the control variables across the entire sample period. The subsequent column demonstrates that, following the additional control of the squared terms of the covariates, the estimated coefficients exhibit no substantial changes and maintain their statistical significance. These findings further demonstrate the robustness of the empirical results in this study.

Table 8.

Robustness test results (Replacement of estimation method)

VARIABLES	Model (17)	Model (18)
	cesd10	cesd10
hdid	-0.2526**	-0.2616**
	(0.0574)	(0.0578)
Control for FE	YES	YES
Control variables	YES	YES
Square term of control variables	NO	YES
Observations	84,400	84,400
Constant	0.0000	0.0004
	(0.0220)	(0.0219)

*, **, and *** represent 10%, 5%, and 1% significance levels, respectively; Robust standard error based on City clustering are in parentheses in the table

Mechanism analysis

Mechanism of action analysis

Analysis of “Increase the development potential of old-age services” mechanism

To investigate the mechanisms through which digital infrastructure construction affects depressive symptoms of middle-aged and older adults, this study employs a two-step approach for mediation effect analysis. The two-step method is preferred over the traditional three-step approach because it replaces the third step with theoretical analysis, effectively mitigating the impact of endogeneity on model estimation [40].

First of all, the mechanism of digital infrastructure construction enhancing the potential of social elderly care services is examined. Table 9 reports the results of the “potential enhancement” effect. The first column indicates that the digital infrastructure construction has a positive and statistically significant effect (at the 1% level) on the number of internet access points in prefecture-level cities. This suggests that digital infrastructure construction increases regional internet penetration, providing middle-aged and older adults with more platforms for information access and social interaction. This reduces their feelings of loneliness and information isolation, thereby alleviating depressive symptoms. For example, under the “Broadband China” strategy, Suzhou City in Jiangsu Province significantly improved internet infrastructure, leading to higher internet penetration rates in urban and rural areas. This enabled middle-aged and older adults to participate in social activities and access health information online, reducing depressive symptoms. The second column shows that the digital infrastructure construction has a positive and statistically significant effect (at the 1% level) on urbanization. This indicates that digital infrastructure construction significantly enhances regional urbanization levels, positively impacting depressive symptoms of middle-aged and older adults. Urbanization not only provides a more robust material foundation but also improves public services, particularly in social security and healthcare, alleviating the pressures of daily life and elderly care, and reducing the risk of depression.

Table 9.

Analysis of “enhancing development potential of old-age services” mechanism results

VARIABLES	Model (19)	Model (20)
	num-net	cityrate
hdid	0.0692***	0.0299***
	(0.0081)	(0.0009)
Control variables	YES	YES
Constant	6.3261***	0.5133***
	(0.0258)	(0.0028)
R-squared	0.348	0.618
Adjust R-squared	0.351	0.610
Observations	84,400	84,400
IndividualFE	YES	YES
CityFE	YES	YES
YearFE	YES	YES

*, **, and *** represent 10%, 5%, and 1% significance levels, respectively; Robust standard error based on City clustering are in parentheses in the table

Based on the above analysis, this study concludes that digital infrastructure construction significantly enhances the potential of social elderly care services by increasing regional internet penetration, urbanization levels, thereby alleviating depressive symptoms of middle-aged and older adults. Thus, H2 is supported.

Analysis of “Enhance social well-being” mechanism

To examine whether the digital infrastructure construction affects depressive symptoms of middle-aged and older adults by enhancing social well-being, this study selects three key variables: household average income, diversity of social activities, and life satisfaction, to test this mediation mechanism.

Table 10 reports the estimation results of the mechanism test. The first column shows that the digital infrastructure construction has a positive and statistically significant effect on the increase in household average income of middle-aged and older adults, indicating that the policy can elevate their household income to a certain extent. This suggests that digital infrastructure construction may further promote internet penetration, improve regional economic structures, and optimize income distribution patterns, indirectly driving the growth of household income for middle-aged and older adults. This enhances their quality of life and, consequently, their mental health. The second column reveals that the digital infrastructure construction has a significantly positive effect on the diversity of social activities of middle-aged and older adults, implying that digital infrastructure construction can significantly enrich their social interactions. The third column indicates that the digital infrastructure construction has a positive and statistically significant effect (at the 1% level) on the life satisfaction of middle-aged and older adults. This demonstrates that digital infrastructure construction can significantly improve their life satisfaction.

Table 10.

Analysis of “Enhancing people’s wellbeing” mechanism results

VARIABLES	Model (21)	Model (22)	Model (23)
	lninc	soc	satlife
hdid	0.2425***	0.0236***	0.0310***
	(0.0195)	(0.0091)	(0.0077)
Control variables	YES	YES	YES
Constant	8.4559***	0.7920***	2.7077***
	(0.0082)	(0.0039)	(0.0258)
R-squared	0.207	0.002	0.029
Adjust R-squared	0.210	0.003	0.030
Observations	84,400	84,400	84,400
IndividualFE	YES	YES	YES
CityFE	YES	YES	YES
YearFE	YES	YES	YES

*, **, and *** represent 10%, 5%, and 1% significance levels, respectively; Robust standard error based on City clustering are in parentheses in the table

Based on the above analysis, this study concludes that digital infrastructure construction can alleviate depressive symptoms of middle-aged and older adults by increasing their household average income, enriching their diversity of social activities, and enhancing their life satisfaction. Thus, H3 is supported.

Heterogeneity analysis

Cognitive heterogeneity

Cognitive ability has been identified as a critical factor in determining the effectiveness of middle-aged and older adults in leveraging the social benefits offered by the “Broadband China” strategy [41–43]. This study utilises data from the CHARLS to assess the cognitive ability of middle-aged and older adults. The assessment encompasses a range of cognitive domains, including date recognition, calculation, drawing, and episodic memory. The date recognition test has a maximum score of 5, the calculation test has a maximum score of 5, and the drawing test scores 1 point, with a total maximum score of 11 for these three tests. The episodic memory test utilises the average score of immediate and delayed recall (maximum score of 10), resulting in a combined total score of 21. Consequently, samples falling below the median score are designated as the low cognitive ability group, while those above are classified as the high cognitive ability group.

As illustrated in Table 11, the analysis is based on cognitive ability group differences. The initial column demonstrates the estimated effect of digital infrastructure construction on the high cognitive ability group. The results indicate that the “Broadband China” strategy exerts a significant alleviating effect on depressive symptoms in this group. The second column reports the estimated effect of the policy on the low cognitive ability group, which is not statistically significant. This suggests that individuals with higher cognitive ability are more likely to access health information and increase social interactions through the internet, actively regulating their psychological state and alleviating depressive symptoms.

Table 11.

Analysis of cognitive heterogeneity results

VARIABLES	Model (24)	Model (25)
	cesd10 of high cognitive ability group	cesd10 of low cognitive ability group
hdid	-0.3247***	-0.1344
	(0.0960)	(0.1234)
Control variables	YES	YES
Constant	10.9596***	12.5545***
	(0.3443)	(0.4308)
Fisher’s Permutation test	0.1903***
R-squared	0.110	0.067
Adjust R-squared	0.110	0.068
IndividualFE	YES	YES
CityFE	YES	YES
YearFE	YES	YES

*, **, and *** represent 10%, 5%, and 1% significance levels, respectively; Robust standard error based on City clustering are in parentheses in the table

Urban-rural heterogeneity

From the perspective of funding and technology research and application environment, whereas rural areas lag in technology and management, limiting the widespread adoption and application of digital technologies [44, 45]. These disparities influence depressive symptoms of middle-aged and older adults through a series of mediating mechanisms. Therefore, this study divides the sample into urban and rural residents to separately estimate the differential effects of the digital infrastructure construction across regions. This approach helps policymakers identify which measures are more effective in specific areas, thereby optimizing resource allocation and enhancing the targeted effectiveness of policies. Table 12 reports the estimation results based on urban-rural heterogeneity. The results show that the implementation of the “Broadband China” strategy has a more pronounced alleviating effect on depressive symptoms of middle-aged and older adults in urban areas, with a coefficient of -0.2365, significant at the 5% level. However, its effect on depressive symptoms in rural areas is not statistically significant. A plausible explanation is that digital infrastructure in urban areas not only provides basic internet connectivity but also supports high-level, multidimensional applications in fields such as healthcare and social security, thereby more significantly alleviating depressive symptoms of middle-aged and older adults.

Table 12.

Analysis of urban-rural heterogeneity results

VARIABLES	Model (26)	Model (27)
	cesd10 of the group that lives a city	cesd10 of the group that lives in rural
hdid	-0.2286**	-0.0793
	(0.1111)	(0.1224)
Control variables	YES	YES
Constant	10.9243***	10.5334***
	(0.4111)	(0.4600)
Fisher’s Permutation test	0.1493***
R-squared	0.095	0.081
Adjust R-squared	0.097	0.082
IndividualFE	YES	YES
CityFE	YES	YES
YearFE	YES	YES

*, **, and *** represent 10%, 5%, and 1% significance levels, respectively; Robust standard error based on City clustering are in parentheses in the table

Age heterogeneity

To reveal the specific impact of digital infrastructure construction on depressive symptoms among different age groups of middle-aged and older adults, this study further divides the sample into two subgroups: middle-aged adults (45–60 years) and older adults (over 60 years), and conducts detailed analyses for each subgroup. This classification helps policymakers identify which measures are more effective for specific age groups, thereby optimizing resource allocation and enhancing the targeted effectiveness of policies. Table 13 reports the results of the analysis based on age group differences. The findings indicate that the digital infrastructure construction significantly alleviates depressive symptoms in both middle-aged and older adult groups. For the older adult group, the policy’s effect coefficient is -0.1911, significant at the 10% level, while for the middle-aged group, the coefficient is -0.3461, significant at the 1% level. This suggests that digital infrastructure construction has a more pronounced positive effect on depressive symptoms of middle-aged adults, likely because they possess greater energy and motivation to adapt to the social benefits brought by digital infrastructure construction.

Table 13.

Analysis of age heterogeneity results

VARIABLES	Model (28)	Model (29)
	cesd10 of	cesd10
	older adults	middle-aged adults
hdid	-0.1911*	-0.3461***
	(0.1129)	(0.1140)
Control variables	YES	YES
Constant	14.7223***	7.4040***
	(0.5497)	(0.5391)
Fisher’s Permutation test	-0.1550***
R-squared	0.095	0.100
Adjust R-squared	0.096	0.102
IndividualFE	YES	YES
CityFE	YES	YES
YearFE	YES	YES

*, **, and *** represent 10%, 5%, and 1% significance levels, respectively; Robust standard error based on City clustering are in parentheses in the table

Limitations

This study has several limitations that merit consideration. First, although the CHARLS dataset covers the majority of provinces and regions across China, it omits certain remote and ethnically diverse areas, potentially limiting the representativeness of the findings. Moreover, its scope is restricted to a single country, underscoring the need for future research to incorporate multi-national cohort data to capture broader cross-national heterogeneity. Second, the exclusive reliance on the CES-D-10 as the primary measure of depressive symptoms among middle-aged and older adults may introduce subjective bias. To strengthen the validity of future investigations, integration of objective clinical or neuropsychological assessments is recommended. Third, despite employing a difference-in-differences analytical framework, residual confounding from unobserved variables cannot be entirely ruled out. Subsequent studies should consider adopting more rigorous causal inference techniques to address endogeneity concerns and enhance the robustness and unbiasedness of results.

Conclusions

This study employs a multi-period staggered difference-in-differences model with 2011–2020 CHARLS data to evaluate the “Broadband China” strategy’s effect on depressive symptoms in middle-aged and older adults. The results show the strategy reduced depressive symptoms by 0.267, with greater impact on adults aged 45–60, those with higher cognitive ability, and urban residents. The paper offers these policy recommendations: Firstly, further increase investment in broadband infrastructure, particularly in rural, remote, and urban fringe areas. Current research shows that broadband access significantly improves older adults’ access to information and social participation, but a digital divide between urban and rural areas still exists. Policymakers can adopt the United Nations’ “digital public goods” approach to ensure equitable coverage, enhance older adults’ access to digital services, and alleviate intergenerational exclusion. Secondly, we should actively explore the design and construction of online medical and mental health service platforms. By drawing on the experience of the Organization for Economic Cooperation and Development (OECD), governments, institutions, and technology companies can collaborate to combine internet technology with healthcare and provide customized mental health interventions for developing countries. Thirdly, special attention should be given to depression symptoms among middle-aged individuals aged 45–60 and older adults living in urban areas. On the one hand, measures such as tax incentives and loan interest subsidies should be provided to assist middle-aged individuals in coping with the pressures of career transitions, family responsibilities, and changes in social roles. On the other hand, in economically developed urban areas, digital service content should be further optimized, while in rural areas, efforts should focus on addressing shortcomings in digital infrastructure development. Additionally, for older adults with lower cognitive abilities, products and services with lower usage barriers should be provided to enhance the accessibility and convenience of connected products and devices, as well as online information services. Policy makers can draw inspiration from the EU’s “active aging” tools, such as tax incentives, to support this group, while optimizing digital services in developed regions and addressing infrastructure gaps in other areas.

Authors’ contributions

Conceptualization: L.Z., C.Z. and S.W.; methodology: C.Z. and S.W.; software: S.W.; validation: S.W.; formal analysis: Y.L.; resources: C.Z.; writing—original draft prepration: Y.L., C.Z. and S.W.; writing—review and editing: Y.L., C.Z., Q.Y and S.W. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Young Researcher Research Fund Project of the Future Humanity Joint Research Institute of Renmin University of China and Xihu University(2025RD0304).

Data availability

Data will be made available on request.

Declarations

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.

Ethics approval and consent to participate

The CHARLS study protocol was approved by the Ethics Review Board of Peking University (IRB00001052-11015) and all participants signed a written informed consent form.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.