Introduction

Since the industrial revolution, the world has been experiencing unprecedented large-scale and high-spirited urbanization (Dong et al. 2022; Wang et al. 2022a). According to the United Nations, by 2030, ~60% of the world’s population is projected to reside in urban settings. However, the rapid urbanization process has also triggered a series of pervasive urban living problems, plunging many countries and areas into unsustainable urban development modes (Chen et al. 2023; Cheng et al. 2022). Along with the fast-growing economy and rising living standards, meanwhile, people’s expectations and demands for a high-quality living environment are increasing continuously. Under this situation, the concept of livable city that emphasizes the comfort and happiness of residents’ living environment has provided a turnaround, which has gradually attracted the attention of urban managers and researchers around the world (Kang et al. 2022; Liu et al. 2023a; Shi et al. 2022). This above-sketched situation also holds for China, the world’s largest emerging economy and developing country. Based on the commonly held law of urban development, China’s urbanization is still in an unprecedented high-speed expansion stage, meanwhile, such non-ecological and unsustainable development model has brought about many risks. Numerous scholars have observed that most cities in China are facing the negative consequences of “urban diseases”, such as traffic congestion, housing shortages, employment difficulties, urban poverty, resource shortages, and environmental deterioration (Su et al. 2023; Wang and Zhou, 2023; Xiao et al. 2022). Referring to the China City Development Report, in 2023, only 16% of Chinese cities are in a relatively healthy development state. To this end, the Chinese government has proposed to “strive to build harmonious, livable, dynamic, and distinctive modern cities”, and a series of policies aiming at improving urban livability have been successively formulated, including the “National New Urbanization Plan” and livable city demonstration zones.

Fortunately, the rise of the smart city concept provides new insights into improving urban livability. The smart city is a novel paradigm of urbanization and urban sustainable development that is supported by modern information and communication technology, featuring energy-saving, environmentally friendly, convenient service, smart participation and people-orientation (Guo and Zhong, 2022; Kummitha and Crutzen, 2017). Worth noting that China is leading the world in the efforts of pilot duration and number of cities for smart city. Since its first implementation of smart city policy (SCP) in 2012, China has launched over 500 smart city projects. And over time, smart city construction offers strong potential for urban sustainability, which has been supported by some studies. For example, Liu et al. (2023b) confirmed that smart city policy can positively affect urban green economic growth in China. Studies conducted by Chen et al. (2024b), Jiang et al. (2021), and Wang (2023) proposed similar insights. Further, some scholars found that smart city construction has a significant promoting effect on green technological innovation (Yan et al. 2023), industrial restructuring (Lv and Gao, 2023), energy efficiency improvement (Wu et al. 2024), and pollution abatement (Guo et al. 2023; Shu et al. 2023), thus improving urban happiness (Chen, 2023) and citizens’ quality of life (Macke et al. 2018). It seems plausible that smart city construction may be a key to promoting urban livability. Hence, could SCP necessarily lead to improvement in urban livability? What is its transmission mechanism? Is there exist a spatial spillover effect in the impact of SCP on urban livability? However, little research has been conducted on these questions.

With the above understanding, this study pays special attention to clarifying the relationship between SCP and urban livability in China that drives urban competitiveness, human well-being, and sustainable development. Specifically, our research innovatively begins with the construction of a composite urban livability index framework that involves four dimensions, including economic affluence, environmental beauty, social security, and convenient living. Treating SCP as a quasi-natural experiment, we then use China’s prefecture-level city data from 2003 to 2019 and employ the time-varying Difference-in-Differences (time-varying DID) model to systematically assess its total effect and heterogeneity impact on urban livability promotion. With the mediating effect model, we further identify the potential effect paths and transmission mechanisms of technological innovation and government social governance. Finally, our spatial DID model confirms notable positive spillover effects of SCP on livability in surrounding cities.

The marginal novelty of our study mainly lies in three aspects: Firstly, we empirically verify the causal relationship between SCP and urban livability, which broadens the existing literature on the outcomes evaluation of macro policies on urban modernization construction and contributes to narrowing the research gap concerning the relationship between smart city and urban development. Secondly, this paper reveals technological innovation and government social governance as the underlying mechanisms, which aid in opening the mechanism “black box” about SCP and urban livability and offers theoretical logic analysis for the influencing factors of urban livability promotion. Thirdly, our research innovatively optimizes the research design of SCP and its impact on urban livability from the perspective of policy spillovers, which can offer new insights into the important role of SCP in urban livability and may help to improve related policies. In conclusion, we believe our research could offer important empirical data and case support for urban development, and may shed light on countries at the same stage or are experiencing this in the future. Figure 1 presents the research framework of this study.

Fig. 1: Research framework.
figure 1

This figure outlines the conceptual and methodological framework of this study.

Full size image

Literature review

SCP and its effect

“Smart City” is an extension of IBM’s “Smart Earth” strategy proposed in 2008 (Palmisano, 2008). Subsequently, smart cities have begun to be deployed in many regions globally, which has sparked increased scholarly interest in this topic. Presently, smart cities have been focused on a broad sense from multiple perspectives, mainly including conceptualization, measurement, and evaluation. While no universally accepted definition exists in academia, there are still commonalities and intersections (Ismagilova et al. 2019). Scholars all seem to highlight smart technology and people-oriented as key attributes (Neirotti et al. 2014; Silva et al. 2018; Lim et al. 2021). For example, Yan et al. (2023) argue that technological progress is especially related to the construction of the smart city. Chen (2023) believes that one major task of a smart city is to offer adequate conditions for urban life and improve citizens’ well-being. To summarize, smart city can be conceptualized as a novel developing pattern of future city, aiming at integrating resources through information and communication technology to improve city operation efficiency and elevate citizens’ life quality (Ahvenniemi et al. 2017; Wang et al. 2020; Yu et al. 2020).

On the basis of the fundamental concept, a series of measurement indicators has emerged in academia to evaluate the performance of smart city construction. To define a smart city, Caragliu et al. (2011) proposed a six-dimensional evaluation framework, widely recognized by researchers, which includes intelligent individuals, advanced mobility, effective governance, a thriving economy, a sustainable environment, and enhanced life quality (Roman, 2018). Barrionuevo et al. (2012) stated that a smart city represents an advanced system in which economic, human, social, environmental, and institutional factors are integrated. Yigitcanlar et al. (2018) further established a comprehensive evaluation framework linking driving factors to expected outcomes. Despite progress in measuring smart cities, creating a universally accepted evaluation framework remains challenging.

To alleviate the bias in indicator measurement, more Chinese scholars tend to use China’s smart city policy since 2012 as a quasi-natural experiment, to evaluate the social–economic consequences, innovation motivation effects, and eco-environment benefits of smart city construction. Regarding the social–economic consequences, numerous studies have verified that smart city construction can exert positive policy effects in many fields, such as urban economic growth (Kim et al. 2016; Zhu et al. 2019), industrial structure upgrading (Lv and Gao, 2023), social governance optimization (Alizadeh and Sharifi, 2023; Cai and Zhang, 2023), reduction of income gap (Dashkevych and Portnov, 2023), increase in labor employment (Ling et al. 2023; Wang et al. 2024) and raise residents’ life quality (Wang and Zhou, 2023). In terms of innovation motivation effects, research conducted by Yang et al. (2024) showed that smart city construction can encourage green innovation at the firm level, and this promotion effect is accompanied by heterogeneity. Studies from Caragliu and Del Bo (2019) and Qiu (2022) also support these findings. Moreover, some researchers have found that smart city construction can significantly optimize resource allocation (Jiang et al. 2023), improve energy efficiency (Haarstad and Wathne, 2019), raise environment quality (Guo et al. 2023), reduce carbon emissions (Song et al. 2023), and achieve sustainable development with the advancement of green technologies (Yan et al. 2023).

Urban livability and its influencing factors

In 1961, the WHO first introduced the four essential elements of a livable environment, namely safety, health, convenience, and amenity (Aquilani et al. 2018). The Second United Nations Conference on Human Settlements held in 1996 proposed the concept of the livable city, becoming a new urban concept in the 21st century. Concerning the connotation of a livable city, scholars hold different views. Generally speaking, a livable city is a synthetic concept, which referred to the overall urban environment related to city living quality and residents’ welfare, including full-fledged infrastructure, convenient traffic conditions, pleasant natural landscape, and good social order (Ley, 1990; Wang et al. 2011; Sheikh and van Ameijde, 2022). As the concept evolves, extensive efforts have been devoted to assessing urban livability and its influencing factors.

Studies on the evaluation of urban livability are more concerned with subjective features and objective environment (Liu et al. 2023a). Subjective evaluations often capture residents’ perceptions of life satisfaction and are mainly collected via telephone interviews, questionnaire surveys, and direct interviews (Zhan et al. 2024; del Mar Martínez-Bravo et al. 2019). While objective evaluations put a higher emphasis on the living environment, the commonly used research method is a multi-index comprehensive evaluation represented by the exploratory factor analysis and entropy method (Najafi et al. 2024; Ran et al. 2024; Shi et al. 2022; Wang and Miao, 2022). Based on the principles of abundance, convenience, comfort, welfare, and safety, Liu et al. (2017) selected 17 indicators and used the exploratory factor analysis to calculate the livability of the city. Li et al. (2021) adopted the entropy method to construct an indicator system for urban livability involving economic prosperity, environmental quality, social security, and convenience of living. Considering the data variability and cognitive differences among residents, researchers are turning more to objective evaluation methods.

As for the influencing factors affecting its change, some studies have identified the important role of economic development in urban livability. For instance, Xiao et al. (2022) held that economic growth can significantly enhance urban livability in underdeveloped regions, using the Loess Plateau as a case study. However, most of the economically developed cities are densely populated (Chowdhury, 2020; Wang et al. 2021), which in turn bring about serious problems such as environmental pollution and climate change (Liang et al. 2020; Jun et al. 2022; Wang et al. 2022c), posing great challenges to urban livability. Other researchers have found that urban livability is associated with a series of factors, such as transportation planning (Miller et al. 2013) and tourism development (Liu et al. 2023a). Notably, Zhao et al. (2022) concluded that traffic convenience can produce knowledge spillover and technology dissemination, and ultimately improve urban livability. Yang et al. (2021) emphasized the importance of social security in the construction of livable cities, thereby strengthening social governance, is necessary.

Research gaps

Although abundant studies have been dedicated to the analysis of SCP and urban livability, providing valuable references for our study. However, the current knowledge is insufficient to explain the relationship between SCP and urban livability. Firstly, existing research has not considered the policy effects of smart city construction on urban livability, which is a topic that is worth focusing on and discussing. Besides, the available literature has confirmed the role of technology innovation and social security in promoting urban livability, while those influencing mechanisms between SCP and urban livability remain unclear. More critically, earlier studies have mostly overlooked the policy’s spatial spillover implications, which may lead to an incomplete understanding of SCP’s effects. To address this research gap, utilizing a dataset encompassing 284 cities across China between 2003 and 2019, this paper employs the time-varying DID, mediating effect and spatial DID models to empirically examine the SCP’s impact on urban livability and its potential mechanisms. The aim is to offer both theoretical and practical insights for China and other developing countries seeking to improve urban livability via smart city endeavors.

Policy background and mechanism analysis

Smart city policy

In 2009, Dibik City collaborated with IBM to build the world’s first smart city. Since then, smart cities have been considered an effective solution to address urban governance difficulties and promote sustainable urban development. According to statistics, there are currently over 1000 smart city projects launched or under construction worldwide, such as Singapore’s “Smart Country 2015” plan, Japan’s “e-Japan”, “u-Japan”, “i-Japan2015” strategies, and South Korea’s “u-Korea” strategy. China’s government explains a smart city as an innovative model that merges advanced technology, harmonizes information resources, consolidates business applications, and improves urban planning, construction, and management. China’s exploration of smart cities began officially over 10 years ago. In 2012, the Ministry of Housing and Urban–Rural Development (MHURD) promulgated the “Notice on carrying out the national smart city pilot work”, launching the first batch of pilot projects in 37 cities, 50 districts (counties), and 3 towns. In 2013, a second batch was added, followed by a third in 2014, bringing the total to 290 approved projects across cities, districts, and towns (Zhu et al. 2019). From high-level planning to actual implementation, many recent documents have emphasized the government’s unwavering commitment to promoting smart city initiatives.

Figure 2 displays the spatial distribution of the SCP pilot cities in China.

Fig. 2: Spatial distribution of SCP pilot cities in China.
figure 2

This figure provides the scope of SCP (smart city policy) in China.

Full size image

Mechanism analysis

Direct effect of SCP on urban livability

The improvement of urban livability is closely related to various factors such as economic development, ecological environment, public services, and infrastructure construction. SCP relies on new-generation information technology and digital platforms, aiming to improve urban management efficiency and promote sustainable urban development, thus having a positive impact on urban livability. Firstly, the networked, digitized, and intelligent attributes of SCP can not only integrate with traditional industries but also bring new models of enterprise and industrial development, thereby helping cities establish a more efficient economic operation system and enhance their core competitiveness (Li et al. 2023). Secondly, SCP can achieve efficient allocation and intelligent management of urban resources through real-time monitoring and big data analysis, such as reducing traffic congestion and optimizing energy consumption structure, thereby improving environmental pollution control capabilities and creating a greener and more sustainable urban environment (Wang and Zhou, 2022). Thirdly, SCP can promote the precision and convenience of public services through intelligent platforms such as smart communities, remote healthcare, and online learning, thereby enhancing residents’ sense of belonging and happiness (Wang and Deng, 2022; Yao et al. 2020). Finally, SCP can promote the digital upgrading and transformation of infrastructure, helping cities build intelligent and efficient new urban infrastructure systems, thereby enhancing urban livability (Wang and Zhou, 2023).

Based on this analysis, we propose Hypothesis 1: SCP can improve urban livability.

Mediating effect of SCP on urban livability

We analyze two mechanisms through which SCP affects urban livability. First, technological innovation serves as the primary channel. Firstly, SCP can formulate a more innovative and friendly environment, which is reflected in the continuous advancements of information infrastructure and the widespread use of information technologies, creating a resource-sharing platform. This platform promotes knowledge dissemination and technological exchange among innovation entities, forming an efficient and open innovation network that can accelerate the transformation and application of scientific research results, ultimately promoting technological innovation (Caragliu and Del Bo, 2019; Zygiaris, 2013). Secondly, as part of SCP, the government supports the development of technology enterprises through innovative policies such as increasing technology fiscal expenditures, which significantly reduces the innovation costs of enterprises and research institutions and encourages the concentration of innovative resources such as talent and capital in pilot cities, thereby promoting the formation of new technologies, new business models and enhancing the technological innovation capabilities of cities(Wang et al. 2022a). Through technological innovation, SCP can not only improve the convenience of residents’ lives but also enhance the inclusiveness and development resilience of cities, enabling them to better cope with urban governance challenges, and ultimately improving urban livability (Zhu et al. 2019).

To sum up, we propose Hypothesis 2: SCP can enhance urban livability through technological innovation.

Government social governance represents a critical mechanism through which SCP enhances urban livability. Firstly, SCP can break down the information barriers between government, market, and society by building a diversified smart government platform, promoting changes in social governance systems, and realizing the transformation of government roles and governance methods. This transformation is mainly reflected in the online, intelligent, and integrated provision of government services, which simplifies the administrative approval process and improves the efficiency of government departments, thereby providing more convenient and efficient services for residents and enterprises (Hao et al. 2024). Secondly, the construction of smart cities contributes to social collaborative governance. For example, smart community management platforms can integrate resident autonomy, property management, and government services, significantly enhancing transparency and public participation in government governance, and thereby improving urban management efficiency (Zhu et al. 2022). Through the above channels, SCP can promote the modernization of social governance, make the operation of cities more efficient, fair, and transparent, and create a more livable living environment for residents (Chen et al. 2024a).

Therefore, we posit Hypothesis 3: SCP enhances urban livability by advancing government social governance.

Spatial spillover effect of SCP on urban livability

The first law of geography states that economic entities are interconnected based on their spatial proximity, which indicates that closer entities tend to have stronger connections (Tobler, 1970). We argue that advanced information technology has strengthened the economic and social connections between cities, providing a prerequisite for SCP to affect the livability of neighboring cities through spatial spillover effects. This can be explained for two main reasons. First, On the one hand, SCP can fully utilize data elements and technological means to narrow the temporal and spatial distance between cities, break the limitations of physical space on socio-economic activities, and more efficiently promote information transmission, cross-regional sharing, and spillover of knowledge and technology, driving surrounding non-pilot cities to develop towards smarter production and lifestyle, and improving their livability (Tan and Chen, 2022). On the other hand, SCP’s outstanding advantages in resource sharing, collaborative office, and integrated government service platform construction can help alleviate the phenomenon of “information silos” between regional administrative entities and provide impetus for cross-regional collaborative governance of urban issues (Ma and Zhu, 2022). Specifically, non-pilot cities can leverage the policy spillover dividends of SCP to meet the diverse needs of social entities for urban development, services, and governance capabilities on a larger scale, creating conditions for improving urban livability.

Based on the above analysis, we provide Hypothesis 4: SCP’s impact on urban livability exhibits positive spatial spillover effects.

The theoretical analysis framework is depicted in Fig. 3.

Fig. 3: Theoretical analysis framework.
figure 3

This figure illustrates the transmission mechanisms of SCP (smart city policy)‘s impact on urban livability (UL).

Full size image

Model, variable, and data

Time-varying DID model

In this study, the implementation of SCP is considered a quasi-natural experiment. To quantify SCP’s impact on urban livability, the DID model is utilized. Given that smart city pilots were established in phases at various time intervals, the time-varying DID model is presented below:

$${{{ {UL}}}}_{{it}}={\alpha }_{0}+{\alpha }_{1}{{{ {SCP}}}}_{{it}}+{\alpha }_{2}{X}_{{it}}+{\lambda }_{i}+{\mu }_{t}+{{\rm{\varepsilon }}}_{{it}}$$
(1)

where ULit stands for the livability of city i in year t, SCPit represents the core independent variable, which takes a value of 1 if city i became an SCP pilot in year t. Xit indicates several control variables that influence urban livability, including per capita carbon emissions, population density, environmental regulations, government intervention, urbanization level, and foreign direct investment. α0, α1, and α2 are the estimated parameters. λi and μt refer to city-fixed and time-fixed effects, respectively. ɛit denotes the random error term.

Variables selection

The dependent variable

Urban livability (UL) serves as the dependent variable in our research. As mentioned earlier, the measurement of urban livability can be divided into two types: individual perception and objective conditions. While individual perception focuses more on residents’ feelings of life satisfaction, such as cultural needs and psychological feedback, which is hard to measure (Liu et al. 2023a). Therefore, the objective conditions evaluation method is more commonly used in practical research. Based on its connotation, urban livability refers to the capacity of a city to offer its residents an urban environment with social harmony, economic prosperity, life convenience, and eco-friendliness.

Based on the “Scientific Evaluation Standard for Livable Cities” issued by the Chinese government in 2007, we referred to the livable city index evaluation system of Li et al. (2021) and added natural and socio-economic characteristic variables of the study area, which can enhance the objectivity and accuracy of the measurement results. Given the data availability and comparability, we construct an urban livability evaluation indicator system comprising four primary indicators and 29 secondary indicators (see Fig. 4). These indicators address various aspects, including resources, environment, economy, society, safety, and living conditions.

Fig. 4: The evaluation indicator system urban livability.
figure 4

This figure presents the indicators involved in the evaluation of UL (urban livability). T the values in parentheses represent the weights.

Full size image

The four primary indicators are as follows:

  1. (1)

    Economic affluence: This reflects residents’ living quality and serves as the material foundation for building livable cities (Xiao et al. 2022).

  2. (2)

    Environmental beauty: This is a key aspect of urban livability, focusing on pollution prevention and control (Wang and Miao, 2022).

  3. (3)

    Social security: This is important for the well-being and safety of urban residents, offering a foundation for a livable environment (Tao et al. 2014).

  4. (4)

    Convenience of life: This encompasses infrastructure, cultural amenities, healthcare, and education levels, which are core elements of urban livability (Mohit et al. 2010).

To evaluate urban livability, we utilize the entropy approach for weighting separate indicators, facilitating a comprehensive measurement.

The core independent variable

The smart city policy (SCP) serves as the core independent variable in this analysis, denoted by a dummy variable. For city i in year t, SCPit is assigned a value of 1 if it is identified as an SCP pilot; otherwise, it is assigned a value of 0. Since SCP targets prefecture-level cities, we exclude county-level cities from our sample. In addition, following Guo and Zhong (2022), cities with only partial district-level SCP implementation have been incorporated into the pilot scope.

Control variables

To mitigate potential bias from omitted variables, we introduce several control variables that may affect UL, which include: (1) Per capita carbon emissions (carbon): Measured as the logarithm of total per capita carbon emissions in cities. (2) Population density (pop): Measured as the logarithm of population per square kilometer. (3) Environmental regulation (ers): Measured by the average rate of solid, liquid, and gaseous waste treatment (Wang et al. 2022b). (4) Government expenditure (gov): Measured as the ratio of government expenditure to GDP. (5) Urbanization (urban): denoted by the percentage of the non-agricultural population in the total population (Dong et al. 2022). (6) FDI (fdi): Measured as the percentage of FDI relative to GDP.

Mediating variables

On the basis of mechanism analysis, we adopt two mediating variables: technological innovation (inn) and government social governance (gsg). These are defined as follows: (1) Technological innovation (inn): As patents are a key indicator of innovation (Krammer, 2009), we measure inn using the number of urban patent applications. (2) Government social governance (gsg): Referring to Xu et al. (2024), gsg is measured by the ratio of the number of people employed in public administration, social security, and social organizations to total urban population.

Data sources

In this study, considering the availability of data and the accuracy of estimation results, we used prefecture-level cities as the samples, while county-level cities were excluded. Due to the measurement indicators of urban livability being closely related to socio-economic factors, the COVID-19 that broke out after 2019 has exerted a significant impact on the economy, society, and residents’ lives, which may lead to the deviation of the estimated value from the actual results. Besides, after 2019, some necessary indicators such as the comprehensive utilization rate of industrial solid waste have no longer been released by China’s National Bureau of Statistics. Therefore, the sample of this paper includes 284 cities in China from 2003 to 2019. Specifically, data on SCP is sourced from the MHURD website in China. Data for other variables are obtained from the China City Statistics Yearbook and various city statistical yearbooks (2004–2020). For variables with missing data, we supplement them by linear interpolation. Descriptive statistics for variables are listed in Table 1.

Table 1 Descriptive Statistics of included variables.
Full size table

Empirical strategy

Benchmark regression results

This study employs the DID model to quantify SCP’s effect on urban livability in China. The Hausman test result shows a p-value of less than 0.05, meaning that using a fixed effect model for regression is more applicable. To avoid multicollinearity, control variables are introduced to examine SCP’s effect on urban livability, with the estimated results listed in Table 2. Column (1) shows the estimated results without control variables, followed by columns (2)–(7) which successively add control variables such as carbon, pop, ers, gov, urban, fdi. It can be observed that in columns (1)–(7), the SCP coefficients are all significantly positive at the 1% confidence level. Hence, SCP is confirmed to significantly improve urban livability in China, and Hypothesis 1 is verified. Moreover, regression results based on column (7) show that SCP has increased China’s urban livability by 3.67%.

Table 2 Benchmark results.
Full size table

For the control variables, the coefficient of carbon is significantly positive, indicating a positive association between per capita carbon emissions and urban livability. This may be due to China’s current carbon-intensive economic development model, where higher per capita carbon emissions generally reflect higher economic development levels (Shi et al. 2022). The coefficient of pop is also significantly positive, which suggests that an increase in urban population can trigger agglomeration effects. This drives economic growth and public infrastructure development, which in turn improves living standards and promotes urban livability (Liu and Sweeney, 2012). The coefficient of ers is significantly positive, indicating that stricter environmental regulations can reduce pollutant emissions, improve the urban eco-environment, and enhance urban livability (Wang et al. 2019). The coefficient of urban is significantly positive, showing that urbanization accelerates the aggregation of resources and improves public services, thereby enhancing urban livability (Yao et al. 2021). By contrast, the coefficient of fdi is significantly negative, supporting the “pollution haven” hypothesis. This implies that FDI may contribute to the transfer of pollution, deteriorating the urban ecological environment and hindering urban livability (Zeng and Zhao, 2009). Finally, the coefficient of gov is not statistically significant, which implies that government spending does not have a measurable effect on urban livability in this study.

Robustness tests

Parallel trend test

In order to apply the DID model, it is crucial to satisfy the parallel trend assumption, which entails the presence of consistent trends in urban livability for both the treatment and control groups prior to the implementation of SCP. Drawing on Guo et al. (2022), we use the event analysis to test the assumption, with the model specified below:

$${{{ {UL}}}}_{{it}}={\beta }_{0}+\mathop{\sum }\limits_{-6}^{5}{\beta }_{1}{{{ {SCP}}}}_{{it}}^{k}+{\beta }_{2}{X}_{{it}}+{\lambda }_{i}+{\mu }_{t}+{{\rm{\varepsilon }}}_{{it}}$$
(2)

where \({{{\rm {SCP}}}}_{{it}}^{k}\) indicates whether a city is designated as a pilot city or not. The test results are illustrated in Fig. 5, indicating that prior to the implementation of SCP, β1 does not pass the 95% significance test. This suggests that there is no significant disparity in UL trends before policy enforcement, thereby satisfying the assumption of a parallel trend. Given this, the DID model can be adopted to evaluate SCP’s impact on urban livability.

Fig. 5: The result of the parallel trend test.
figure 5

This figure shows the differences in urban livability (UL) between the treatment and control groups prior to the implementation of smart city policy (SCP).

Full size image

Using staggered DID model

Due to the varying establishment years of SCP in China, we use the time-varying DID model to assess its effect on urban livability. Recent studies, however, suggest that applying the traditional fixed-effects model in time-varying DID may lead to significant biases caused by heterogeneous treatment effects (De Chaisemartin and Haultfoeuille, 2020; Sun and Abraham, 2021). Specifically, the bidirectional fixed-effects estimator may encounter the ‘bad control group’ problem, in which previously treated samples can act as a control for those treated later. This mismatch, arising from differences in policy implementation, results in biased policy evaluations (Goodman-Bacon, 2021). To address this, we employ heterogeneity-robust estimation methods to test the reliability of benchmark results (Cengiz et al. 2019; Gardner, 2021).

First, we apply the stacking regression method to obtain heterogeneity-robust estimators (Cengiz et al. 2019). This method pairs each treatment group individual with samples that have never received or have not yet received treatment. The data is then stacked by relative event time and regressed. Second, we use an imputation-based approach, following Gardner (2021), to estimate the counterfactual outcome for each treated individual at each period, using untreated samples. The average treatment effect of the policy is obtained by computing the weighted mean of the disparity between the observed and hypothetical outcomes. The alternative DID model, known as the staggered DID model, is seen in Fig. 6. The findings are in line with those in Fig. 5, validating the resilience of benchmark results against heterogeneous treatment effects.

Fig. 6: Estimated results of staggered DID model.
figure 6

This figure displays the heterogeneous treatment effects of the staggered DID model.

Full size image

Placebo test

To confirm the causal relationship between SCP and urban livability, we perform a placebo test on the benchmark results using random sampling, with reference to Li et al. (2016). Specifically, we generate a virtual dataset of pilot cities through random sampling and conduct a DID regression. This process is repeated 1000 times, respectively, and the estimated coefficients’ distribution is displayed in Fig. 7. The analysis reveals that the SCP coefficient is centered around 0 and closely obeys normal distribution, suggesting that the benchmark results are reliable and not influenced by random factors.

Fig. 7: Results of the Placebo test.
figure 7

This figure describes the Placebo test results of 1000 random samples.

Full size image

Using entropy balancing and PSM-DID method

To effectively evaluate the policy effect of adopting the DID model, it is preferable that the treatment and the control groups be randomly generated. However, in reality, the Chinese government may select smart city pilots according to factors including city size, development potential, infrastructure, and information technology level, which is not random and may lead to estimation bias. Given this, we employ the entropy balancing and the PSM methods to alleviate the sample selection bias, and then use the DID model for regression. The estimated results of the entropy balancing-DID and the PSM-DID are shown in columns (1) and (2) of Table 3, respectively. It is found that the estimated results do not differ considerably from the benchmark results.

Table 3 The estimated results of entropy balancing and PSM-DID method.
Full size table

Eliminating interferences from related policies

To confirm the reliability of benchmark results, we control for other relevant policies implemented during the sample period that could potentially affect urban livability. Specifically, we add dummy variables for carbon emissions trading policy (CETP), low-carbon city pilot policy (LCCP), Broadband China policy (BC), innovation-oriented city policy (ICP), and green finance experimental zone policy (GFPZ) as control variables in Eq. (1) for regression, as reported in Table 4. After controlling for the influence of these policies, it is evident that the SCP coefficient remains significantly positive, confirming the benchmark results’ robustness.

Table 4 Estimated results after eliminating related policies’ interference.
Full size table

IV test

To address endogeneity issues resulting from the omitted variables, we employ the IV approach. Following Wang et al. (2022a), we use the number of telephones in each city in 1984 as the IV. The rationale for this choice is twofold: (1) Cities with more telephones in 1984 are more likely to have a higher level of ICT and be designated as SCP pilot cities, satisfying the relevance assumption of the IV. (2) The number of telephones in 1984 is a historical variable that does not influence the current livability of cities, meeting the exogeneity assumption of the IV. Thus, the number of telephones in 1984 meets both the relevance and exogeneity conditions, making it a suitable instrument.

However, since the number of telephones in 1984 is a cross-sectional variable, it cannot be used directly in panel data regression. Given this, we use the interaction between the number of telephones in 1984 and time dummy variables as the IV. Table 5 presents the IV test results. The 1st-stage regression indicates that the number of telephones in 1984 is significantly and positively correlated with the establishment of SCP pilot cities. The F-statistic for the weak IV test is 3809.94, well above the threshold of 16.38, indicating that the chosen IV is valid. The 2nd-stage regression results show that SCP has a significant positive impact on urban livability. Therefore, after addressing endogeneity, our finding that SCP promotes urban livability remains robust.

Table 5 Regression results of IV test.
Full size table

Replace the dependent variable

Considering that the measurement methods for urban livability may interfere with the estimation results, we further re-measured the urban livability by machine learning and the Critic-Topsis weighted method and re-conducted the regression using model (1). Specifically, we first applied an unsupervised learning method in machine learning to reduce the dimensionality of the indicators and then obtained the composite index of urban livability by integrating the weighted data. Besides, we used the Critic method to determine the weights of the indicators for urban livability, and then employed the Topsis method to aggregate the results and finally derive the urban livability. The re-estimated results are presented separately in Table 6. It can be observed that the coefficient of SCP remains significantly positive, further confirming the robustness.

Table 6 Regression results of replacing the dependent variable.
Full size table

Heterogeneity analysis

Given the substantial variations in economic growth, resource allocation, and population size across cities, the SCP’s effect on urban livability may vary. Therefore, we categorize cities according to city size, resource endowment, and human capital to analyze the heterogeneity of SCP’s effects.

City size heterogeneity

Following the “Adjustment of Urban Size Division standards (2014),” cities are grouped into two categories: small and medium-sized cities, and large cities with populations exceeding 3 million. Separate regressions are performed for each group. As listed in columns (1) and (2) of Table 7 the SCP coefficient for small and medium-sized cities is not statistically significant, whereas the coefficient for large cities is significantly positive. This suggests that SCP has a more pronounced effect on livability in large cities. The likely reason is that large cities, compared to smaller ones, typically have larger economic scales and better resource bases, enabling higher investments in the infrastructure and operational costs required for smart city construction (Qian et al. 2021). As a result, large cities are better equipped to integrate advanced technologies, digital infrastructure, and innovative resources, which in turn lead to greater improvements in urban livability.

Table 7 Heterogeneity analysis results.
Full size table

Resource endowment heterogeneity

The SCP’s effect on livability may vary due to differences in urban resource endowment. According to the “National Sustainable Development Plan for Resource-based Cities (2013–2020),” we group cities into resource-based and non-resource-based categories and conduct separate regressions. As shown in columns (3) and (4) of Table 7, SCP improves livability in both resource-based and non-resource-based cities. However, the effect is more pronounced in non-resource-based cities. This may be because, to meet the requirements of sustainable development, these cities are typically dominated by technically advanced industries, and have a more diversified and complete industrial structure. These features enable them to better absorb the policy dividend brought by SCP, thereby effectively enhancing urban livability (Wang and Zhong, 2023). In contrast, resource-based cities, due to prolonged resource extraction and extensive development, often face more severe urban development issues, such as resource waste and environmental pollution, which to some extent weaken the implementation effect of SCP.

Human capital heterogeneity

We further classify cities according to their human capital level: cities with higher human capital (those with 985 or 211 universities) and cities with lower human capital. Separate regressions are performed for each group, with the results listed in columns (5) and (6) of Table 7. The findings show that SCP improves livability in both high and low-human capital cities. However, the effect is pronounced in cities with higher human capital. This is likely because highly skilled and educated individuals are more adept at utilizing new technologies brought by SCP, which fosters innovation, economic growth, and social progress (Jiang et al. 2021), all of which enhance urban livability. Additionally, a higher human capital improves the social and cultural environment of cities, helping to reduce social conflicts and improve public safety, further boosting livability. Thus, SCP exerts a more significant positive impact on livability in cities with higher human capital.

Mechanism verification

As discussed above, SCP can enhance urban livability by promoting technological innovation and social governance. To empirically test these mechanisms, we apply the mediating effect model as follows:

$${{{ {UL}}}}_{{it}}={\alpha }_{0}+{\alpha }_{1}{{{ {SCP}}}}_{{it}}+{\alpha }_{2}{X}_{{it}}+{\lambda }_{i}+{\mu }_{t}+{{\rm{\varepsilon }}}_{{it}}$$
(3)
$${M}_{{it}}={b}_{0}+{b}_{1}{{{ {SCP}}}}_{{it}}+{b}_{2}{X}_{{it}}+{\lambda }_{i}+{\mu }_{t}+{{\rm{\varepsilon }}}_{{it}}$$
(4)
$${{{ {UL}}}}_{{it}}={\gamma }_{0}+{\gamma }_{1}{{{ {SCP}}}}_{{it}}+{\gamma }_{2}{M}_{{it}}+{\gamma }_{3}{X}_{{it}}+{\lambda }_{i}+{\mu }_{t}+{{\rm{\varepsilon }}}_{{it}}$$
(5)

where Mit represents the mediating variable, which includes inn and gsg. Regression analysis is conducted based on Eqs. (3)–(5). If α1, b1, and γ2 are significant and γ1 is also significant but smaller than α1, it indicates a partial mediating effect of Mit. If α1, b1, and γ2 are significant, but γ1 is not significant, a full mediating effect of Mit is observed. Otherwise, no mediating effect is present.

Table 8 presents the mediating effect estimation results. Specifically, the results in column (1) show that SCP has a significantly positive impact on inn, indicating that SCP can significantly drive technological innovation in pilot cities. In column (2), the coefficients of SCP and inn are both significantly positive, implying that technological innovation acts as a mediating variable in the relationship between SCP and urban livability, confirming Hypothesis 2. As for column (3), the coefficient of SCP is significantly positive, indicating that SCP effectively improves the social governance capacity of pilot cities. In column (4), after the inclusion of gsg, both the coefficients of SCP and gsg are significantly positive, indicating that social governance plays a mediating role in the process through which SCP enhances urban livability. Hypothesis 3 is confirmed.

Table 8 The results of mediating effect analysis.
Full size table

Spatial spillover effect analysis

Due to the continuous transfer of resources and economic activities among cities, the impact of SCP may exist spatial spillover effect. This suggests that SCP’s influence extends beyond the pilot cities, affecting the livability of both these cities and their surrounding cities. Therefore, we examine the spatial spillover effects below.

Spatial correlation test

We conduct a Moran’s I test to assess the spatial correlation of urban livability. First, we use an inverse distance weight matrix W to create local Moran’s I scatterplots of urban livability for the years 2003, 2008, 2013, and 2019, as seen in Fig. 8. The scatterplots mainly cluster in the 1st and 3rd quadrants, indicating a positive spatial correlation of livability across China cities.

Fig. 8: Scatter plots of the local Moran’s I of urban livability in 2003, 2008, 2013 and 2019.
figure 8

This figure demonstrates the space correlation of urban livability (UL) for the selected years 2003, 2008, 2013 and 2019.

Full size image

Subsequently, we further examine the spatial correlation of urban livability adopting the global Moran’s I index, which can be expressed below:

$${{\rm {Moran}}}^{\prime} s\,I=\frac{n}{{\sum }_{i=1}^{n}{\sum }_{j=1}^{n}{W}_{{ij}}}\times \frac{{\sum }_{i=1}^{n}{\sum }_{j=1}^{n}{W}_{{ij}}\left({x}_{i}-\bar{x}\right)\left({x}_{j}-\bar{x}\right)}{{{\sum }_{i=1}^{n}\left({x}_{i}-\bar{x}\right)}^{2}}$$
(6)

where n denotes the sample cities’ number, and Wij refers to the inverse distance spatial weight matrix. xi and xj stand for the livability levels of cities i and j, respectively, and \(\bar{x}\) indicates the average livability level of all sampled cities. If the global Moran’s I index is larger (less) than 0, it shows that the variable has a positive (negative) spatial correlation.

As seen in Table 9, the Moran’s I indices for 2003–2019 are all significantly positive, suggesting a considerable positive spatial correlation of livability in China’s cities.

Table 9 The global Moran’s I Index of urban livability.
Full size table

The spatial spillover effect

Given the confirmation of spatial correlation in urban livability from both local and global Moran’s I indices, we adopt the spatial DID model to analyze SCP’s spatial effect on livability in China cities. To ensure the reliability of our estimates, we apply three spatial DID models: SDM-DID, SAR-DID, and SEM-DID. The LR and Wald tests in Table 10 provide significant evidence against the null hypothesis at the 1% level. This shows that the SDM model cannot be simplified to the SAR or SEM models. Therefore, we consider the SDM-DID model to be the most suitable for our research, and it is formulated as follows:

$$\begin{array}{ll}{{{{UL}}}}_{{it}}={\beta }_{0}+{\beta }_{1}{{{ {SCP}}}}_{{it}}+{\beta }_{2}{X}_{{it}}+{\beta }_{3}W* {{{ {SCP}}}}_{{it}}+{{\rho }}W* {{{{UL}}}}_{{it}}\\\qquad\\\qquad+{\beta }_{4}W* {X}_{{it}}+{\mu }_{i}+{\delta }_{t}+{\varepsilon }_{{it}}\end{array}$$
(7)

where ρ stands for the spatial autocorrelation coefficient, W denotes the inverse distance spatial weight matrix, W*SCPit refers to the spatial lag term of SCP, W*ULit denotes the spatial lag term of UL. Additionally, to validate the statistical reliability, the estimated results for SAR-DID and SEM-DID models are also reported in Table 10. The λ value of the SEM model and the coefficients of the spatial lag term ρ of the SAR-DID and SDM-DID models are found to be significantly positive at the 1% level, showing a clear spatial dependence. This suggests the presence of significant spatial correlation in urban livability.

Table 10 The spatial spillover effect analysis results.
Full size table

Since the policy effect measured by SDM-DID model includes mutual influences between regions, the coefficients cannot accurately reflect SCP’s net effect on the livability of both local pilot cities and surrounding non-pilot cities. Therefore, we decompose SCP’s impact on urban livability into direct and indirect effects. As reported in Table 10, the direct effect coefficient of SCP is 0.0382, which is significantly positive. This implies that, after accounting for spatial factors, SCP can still enhance the livability of local pilot cities. Besides, the indirect effect coefficient of SCP is 0.6195, and is also significantly positive, showing that SCP can also promote the livability of surrounding non-pilot cities, thereby verifying Hypothesis 4. Relying on advanced information technologies such as big data, artificial intelligence, and cloud computing, SCP can break through the limitations of time and space, accelerate knowledge and technology spillover, and resource sharing, thereby improving the production and lifestyle of surrounding non-pilot cities and promoting cross-regional collaborative governance of urban issues, ultimately improving their urban livability (Tan and Chen, 2022).

Conclusions and policy implications

Conclusions

Using panel data from China’s 284 prefecture-level cities covering the years 2003 to 2019 and treating SCP as a quasi-natural experiment, this study quantifies its impact on urban livability, the channels of its effect, and the spatial spillover effects by adopting the time-varying DID, the mediating effect, and the spatial DID models. The main findings are below: (1) SCP can promote urban livability in China’s cities, which remains consistent after various robustness tests. (2) SCP’s impact on urban livability is more pronounced in large cities, non-resource-based cities, and cities with higher human capital, compared to small and medium-sized cities, resource-based cities, and cities with lower human capital. (3) Technological innovation and government social governance are key pathways through which SCP enhances urban livability. (4) SCP can improve the livability of surrounding non-pilot cities, demonstrating positive spatial spillover effects.

Policy implications

Based on the findings above, the following policy implications can be derived:

  1. (1)

    Since SCP enhances urban livability, the Chinese government should actively increase the intensity and scope of pilot projects for smart cities, attracting more cities to participate in smart city construction. On the one hand, the government should establish special funds for smart cities, increase investment in information technology infrastructure such as hardware equipment, software systems, and data centers, and encourage more social capital to flow into the construction of smart cities. On the other hand, continue to broaden the application scenarios of information technology in smart construction and management, smart industries and economy, and smart services, and strengthen its deep application in fields such as public safety, urban management, healthcare, and education. Finally, excellent cases of pilot construction of smart cities should be created and their development experience should be promoted to drive the construction of smart cities in more cities.

  2. (2)

    Given that SCP’s impact varies by heterogeneity characteristics of cities, tailored policies for smart city planning to ensure urban livability can be further formulated. For cities with a higher foundation for livability, such as non-resourced-based cities and cities with larger population scales and higher human capital levels, policy makers should maintain their SCP advantage while promoting the further application of information technology in various subsectors of enterprise production, residents’ lives, and urban management, to effectively solve urban disease problems, and improve the intelligence, refinement, and efficiency of urban operations. Additionally, the government should be guided by digital policies and encourage all sectors of society to develop more advanced and complex technological tools, in order to provide innovative solutions for improving urban livability. Regard to smart city construction in the low-level cities, the key lies in the development of information technology infrastructure, which needs more financial support from the government. Meanwhile, these cities should pay attention to local issues and needs, and choose smart projects with strong adaptability based on the city’s endowment characteristics and development stage.

  3. (3)

    Taking advantage of the SCP, the government should continuously encourage technological innovation, and optimize the social governance system to enhance the livability of cities. Specifically, reallocating more financial funds and preferential policies into the research and application of advanced or emerging technologies, such as big data, artificial intelligence, and cloud computing cannot be overemphasized during the construction of smart cities in China. Furthermore, the government should also actively guide the transformation of the economic development mode, taking the emerging industries featured by technology, research and development, and information generated by SCP as future economic growth points. Finally, smart infrastructure should be integrated into various aspects of urban management, such as building platforms for smart transportation, smart energy, smart healthcare, smart government, etc., to promote the modernization of urban governance system and governance capacity.

  4. (4)

    The positive spatial spillover effect of SCP highlights the importance of optimizing smart city policy arrangements, which requires various entities such as governments, enterprises, and citizens to actively establish cross-regional cooperative governance mechanisms and jointly participate in smart city construction. Local governments at all levels should advocate the establishment of cooperation mechanisms and supporting systems for collaborative governance of urban issues, and promote cross-regional government service integration, so as to provide convenience for enterprises and citizens’ production and life. Enterprises should use technological means to break down barriers to factor flow, knowledge and technology dissemination between cities, and strive to seek cross-regional economic cooperation to enhance their competitiveness. For citizens, it is necessary to enhance their participation in urban governance. They can proactively provide feedback on urban issues through an intelligent complaint platform, thereby helping the government to accurately respond to social needs and improve the pertinence and scientificness of public policies.

There are still some limitations to this research. Firstly, the parallel trend test indicates that there is a significant lag in the impact of SCP on urban livability, which may require a longer sample period and more comprehensive data to verify this time lag effect. Besides, this paper preliminarily verifies the mediating roles of technological innovation and social governance. In the future, we can investigate the influencing channels of SCP in empowering urban livability from more dimensions, thus forming a more comprehensive understanding of SCP.