Introduction

Economic growth is fundamentally driven by knowledge production and the accumulation of human capital, with technological progress from research and development (R&D) being essential for national development (Acemoglu, 2012). Insufficient output in knowledge sectors and slow technological advancement contribute to sustained low economic growth in some countries. Urban innovation has become a vital aspect of innovation strategies, as cities aggregate diverse resources, making them significant centers for technological activities and knowledge application, thus acting as catalysts for economic and social growth (Freire-Gibb, 2012). Innovation often involves substantial upfront costs, risks, complexity, and long payback periods. In response, since the 1990s, governments have launched “innovative city” initiatives to promote private sector involvement in urban innovation (Nemet, 2009). A notable example is the “Tech City” initiative in London, initiated in 2010 by the UK government and local authorities, representing a milestone in innovation (Ferm and Jones, 2017). Similar initiatives have emerged in cities like Boston and Sydney, highlighting a global trend toward creating innovation hubs (Van der Voort and De Jong 2004; Esmaeilpoorarabi et al. 2020).

To expedite the implementation of an innovation-driven development strategy, the Chinese government launched the Innovative City Pilot Policy (ICP) in Shenzhen in 2008, subsequently expanding the policy to encompass 14 cities, including Dalian and Qingdao, in 2009. With the gradual expansion of the innovative city policy, additional cities in central and western China have been included in the policy pilot list. As of early 2023, China’s National Development and Reform Commission and the Ministry of Science and Technology had approved 103 innovation pilot cities. The innovative city pilot policy aims to promote urban innovation activities, enhance urban innovation capabilities, and implement a national innovation-driven development strategy. As a typical region-oriented policy, innovative city construction is not only an important measure for the transformation and upgrading of various cities in China but also a concentrated embodiment of China’s innovation-driven policy.

In today’s era of economic globalization, more open and fluid innovative resources are flowing rapidly worldwide, strengthening the economic and technological connections among countries. With the deepening of economic globalization and vertical specialization, the continuous decrease in trade costs and the rapid development of information technology, the international division of labor is increasingly reflected in the intraproduct division, and the scale of intermediate goods trade is rapidly increasing. Along with the process of trade liberalization, China actively integrated into the global intraproduct division system, and the import of intermediate goods has become a significant part and a major force of China’s import trade. The data show that in 2023, China’s total imports of intermediate goods amounted to 14.29 trillion yuan, accounting for 79.4% of the total import value, with sources spanning over 200 countries and regions across six continents.

How does the innovative city pilot policy supported by the Chinese government to promote indigenous innovation affect local enterprises’ import of intermediate goods? The existing research presents two conflicting views on the relationship between introducing foreign technology through intermediate goods imports and indigenous innovation: the complementarity hypothesis and the substitution hypothesis. The complementarity hypothesis posits that, according to the endogenous growth theory, high-quality imports of intermediate goods, particularly the upgrading of quality and the increase in the variety of intermediate goods (Aghion and Howitt, 1992), are a source of endogenous technological progress. These are important avenues for developing countries to absorb technological spillovers and achieve technological catch-up (Romer, 1986). Domestic enterprises can access embedded advanced technologies by acquiring a broader range of intermediate inputs (Shepherd and Stone, 2012). In contrast, scholars supporting the substitution hypothesis argue that introducing extensive foreign technology through intermediate goods imports can stimulate and reinforce domestic firms’ dependence on foreign advanced technology, weakening the demand and motivation for indigenous innovation (Zhang et al. 2014). Although the impact of intermediate goods imports on indigenous innovation is debatable, the close connection between these two factors is indisputable. In this context, it becomes imperative to determine how the Chinese government’s innovative city policy to promote enterprise innovation affects local enterprises’ imports of intermediate goods. Will the policy increase the import quantity and variety of intermediate goods to obtain more new products, new processes (Bloom et al. 2016), and core technologies for key components (Goldberg et al. 2010)? In the context of the real economy, China’s decentralized system and promotion incentive mechanism (Zhou, 2007) have led to the continuation of a ‘subsidy–preference’ policy mindset in the pilot project of innovative cities, which is accompanied by the ‘visible hand’ of various levels of government in promoting innovative development. The policy logic of stimulating ‘specific’ goals to achieve the assessment targets for constructing innovative cities in the shortest possible time may complicate the relationship between the innovative city policy pilot and the import of intermediate goods.

On the basis of the above discussion, this paper addresses the following questions. Does China’s innovative city policy significantly impact enterprises’ imports of intermediate goods in pilot cities? What is the impact mechanism? Does heterogeneity exist among different types of enterprises? To answer these questions comprehensively, this paper conducts multidimensional research on the influence and mechanism of implementing innovative city policies on the quantity, variety and quality of intermediate goods imported by manufacturing enterprises in these cities. This paper also explore the differences in the impact mechanism of the innovative city policy pilot on intermediate goods imports among heterogeneous enterprise samples on the basis of factors such as enterprise size, age, ownership and trade mode. The answers to these questions can provide policy input for comprehensively evaluating the implementation performance of innovative city policies and further improving these policies.

Compared with existing research, this paper makes three contributions. First, it contributes novel empirical findings on the efficacy of innovation policies and industrial policies in the Chinese context. There has been a longstanding and substantial discourse within the national economic community concerning the efficacy of industrial policies. This study examines the influence and mechanism of the innovative city policy on intermediate goods imports, which presents potential supplementary insights into the discourse on the effectiveness of innovation policies in China. Second, current research on China’s innovative city policy has focused mainly on innovation behavior and output, overlooking the impact of innovation policies on intermediate goods imports, neglecting the crucial role of intermediate goods imports in fostering enterprise innovation. Therefore, this study examines China’s innovative city policy from the perspective of intermediate goods imports, fills the research gap between innovation policies and intermediate goods imports, and provides new empirical evidence to enhance urban innovation levels in an open economy context. Third, this study explores three impact channels of the innovative city policy on enterprise imports of intermediate goods through the acquisition of government subsidies, the mitigation of financing constraints, and the reduction of institutional transaction costs. This study also conducts empirical analyses utilizing diverse econometric techniques to ensure robust and dependable conclusions. The rest of this paper is structured as follows. Section “Policy Background” introduces the background of the Innovative City Pilot Policy (ICP). Section “Literature Review and Research Hypotheses” reviews the relevant literature and proposes research hypotheses. Section “Model, Variable and Data” presents the econometric model, variable and data. Section “Empirical Test and Analysis” conducts an empirical analysis. Section “Heterogeneity and Robustness Test” provides further tests of heterogeneity and robustness. Section “Conclusions and Policy Recommendations” summarizes the entire paper and discusses policy recommendations.

Policy background

The Innovative City Pilot Policy (ICP) is an important manifestation of China’s efforts to build an innovative country and implement an innovation-driven development strategy. In terms of its development process, it has gone through three main stages.

The first stage is the pilot exploration stage (2005–2009). In 2005, in response to the challenge of insufficient scientific and technological innovation capabilities since the reform and opening up, the State Council of China issued the “National Medium- and Long-term Science and Technology Development Program Outline (2006–2020)” (State Development [2005] No. 44), which proposed building an innovative country as a major strategic choice for the future. In 2008, the gradual practice of building innovative cities began with the approach of “piloting first, accumulating experience, and gradually expanding.” The approval of Shenzhen—a city with institutional advantages due to its reform and opening up—as the first national innovative city pilot marked the formal entry of the innovative city pilot into the national policy agenda.

The second stage is the systematic promotion stage (2010–2015). Building on the pilot experience in Shenzhen, the “Guiding Opinions on Further Promoting the Pilot Work of Innovative Cities” (National Science and Technology Development [2010] No. 155, referred to as the “Opinions”) issued by the Ministry of Science and Technology of China in 2010 noted that cities are centres of regional economic and social development, gathering places for various innovation factors and resources, and play a significant role in the overall development of regions and countries. It is necessary to encourage a group of cities to lead the way to becoming innovative cities, setting an example and guiding more cities to embark on innovation-driven development. In 2010, the number of national innovative city pilots approved by China reached the largest scale in the entire gradual reform process, with the National Development and Reform Commission approving 16 cities, including Dalian, and the Ministry of Science and Technology approving the first batch of 20 cities (districts), including Beijing’s Haidian District, and the second batch of 18 cities (districts), including Shijiazhuang in Hebei Province, as national innovation pilot cities (districts).

The third stage is the high-quality development stage (2016 to present). At the end of 2016, the Ministry of Science and Technology and the National Development and Reform Commission jointly issued the “Guidelines for Building Innovative Cities” (National Science and Technology Development [2016] No. 370, referred to as the “Guidelines”), integrating the innovative cities (districts) approved by the Ministry of Science and Technology and the National Development and Reform Commission in the early stage and forming a list of 61 innovative city pilots. In 2017, all 61 innovative city pilots passed onsite inspections. As of early 2023, the two ministries had approved a total of 103 innovation pilot cities. During this period, governments at all levels actively promoted constructing innovative cities and strengthening the implementation and monitoring of construction plans and tasks. Significant results have been achieved with the coordinated support of related fiscal and tax incentives, financial guidance, talent incentives, and other supporting policies.

The Innovative City Pilot Policy (ICP) aims to promote urban innovation activities, enhance urban innovation capabilities, and drive the implementation of a national innovation-driven development strategy. In line with the guidelines for innovative cities, each city is expected to increase investment in innovation, focus on developing innovation resources, and create an environment conducive to innovation. The specific development goals, key tasks and policy guarantees of the innovative city pilot policy are detailed in Table 1.

Table 1 Development goals, key tasks, and policy guarantees for innovative cities.
Full size table

Literature review and research hypotheses

Literature review

Innovation is the core of civilization’s progress and an inexhaustible driving force for a country’s prosperity and development. With the continuous advancement of urbanization, many cross-regional, cross-cultural and even global innovation resources and elements are being gathered in cities. Cities have become essential carriers and centres of innovation for a country or region (Lv et al. 2018). The ‘creative city’ emphasized by Jacobs (1984) focused on driving urban revival through creative cultural ideas, human capital, innovative class, open organizational culture and inclusiveness, diverse cultural industries, perfect infrastructure and social diversity. Additionally, the ‘innovative city’ developed by Simmie (2001) focused on the growth effects brought to cities and countries by transforming technological, knowledge and talent systems. Both consider the innovative city as a combination of agglomeration, diversity, instability and good reputation driven by innovative factors, such as technology, knowledge, human resources, culture and institutions, which can cultivate and develop a knowledge-based economy (Fang, 2013).

With the continuous deepening of policy implementation, many scholars have started researching China’s innovative cities and related topics. Many studies focus on verifying innovative city policy implementation performance through empirical analysis. Representative studies include Li and Yang (2021), who reported that implementing the innovative city policy has realized the flow and concentration of various production factors, especially innovation factors, in a particular geographical space. Such policies are conducive to improving urban innovation levels and have optimized the spatial layout of productivity (Wang et al. 2023), making them essential for transforming Chinese cities into high-quality development (Chen and Zhang, 2020). Innovative city policies can stimulate the innovative vitality of enterprises (Zhang and Wang 2022), significantly promoting both innovation input (Hu et al. 2021) and innovation output (Zhou and Li, 2021, Liu et al. 2019), particularly in terms of patent generation (Guo et al. 2021; Yan et al. 2021). Additionally, several studies have confirmed the positive impact of pilot policies regarding green ecological effects (Yang et al. 2023a), such as urban carbon performance (Li et al. 2016; Wei and Kong, 2022), urban green total factor productivity (Gao and Yuan, 2022), urban green innovation (Quatraro and Scandura, 2019), and efficiency in power usage (Wang et al. 2023). Moreover, some scholars have also conducted research on the quality effects of export products from innovative cities (Li and Zhong, 2021).

In recent years, the import of intermediate goods has prompted extensive debate within academic circles. Predominantly, scholars have focused their attention on the economic performance of micro-enterprises as a consequence of intermediate goods imports. The increasing variety of intermediate goods has significantly enhanced the total factor productivity of enterprises (Qian et al. 2011). However, Zhang et al. (2015) constructed a theoretical model from the perspectives of intermediate goods’ quantity, variety, and price affecting total factor productivity. Empirical tests revealed that the impact of the quantity and variety of intermediate goods on total factor productivity was insubstantial. Through micro-enterprises data analysis, Zhang et al. (2010) found that increasing the scale of intermediate goods imports can significantly increase the total factor productivity of enterprises. With respect to the impact of intermediate goods imports on micro-enterprises innovation, current academic views focus mainly on “promotion” and “inhibition.” From a promotion standpoint, importing intermediate goods helps enterprises acquire technology, promote innovation, and increase product value (Xu and Mao, 2019). Conversely, from an inhibition perspective, as the cost of intermediate goods imports continues to decrease, enterprises’ innovation cost far exceeds the purchasing cost, thereby hindering innovation and research and development (R&D)(Liu and Qiu, 2016). Huang et al. (2016) argued that, owing to low-quality intermediate goods imports resulting from low-end value chain lock-in, enterprises find it challenging to benefit from knowledge spillovers, leading to a low-value-added trap. Xu et al. (2017) suggested that intermediate goods imports significantly enhance the quality of export products through product quality, product variety, and spillover effects. Intermediate goods trade is gradually becoming multilateral, with institutional and policy impacts between trading entities becoming one of the factors influencing intermediate goods imports. Tian et al. (2023) reported that technical trade barriers significantly increase the quality of intermediate goods imports. Huang and Wu (2020) discovered that government efficiency, foreign direct investment, and trade complementarity have varied impacts on different types of intermediate goods. Scholars also use intermediate goods imports as an intermediate variable for mechanism testing. On the basis of Chinese micro-enterprises data research, Song et al. (2021) discovered that improvements in the institutional environment have a direct positive effect on product quality and strengthen the role of intermediate goods imports. Cai et al. (2023) did empirical analysis using Chinese enterprise data and reported that international market exposure through intermediate goods imports enhances innovation at the enterprise level in developing countries. Research conducted against the backdrop of the “Belt and Road Initiative” has shown that regional intermediate goods imports have a positive effect on total factor productivity growth over time.

Research hypothesis

As a city development model with innovation as the core driving force, the innovative city policy pilot is an important support and driver of China’s current construction of an innovative country. Some related supporting policies of innovative city policy pilots indicate that government subsidies and various preferential policies are still the primary means for the government to promote the construction of innovative cities. From this perspective, the innovative city policy is a production cost intervention policy. Cost is a key variable determining enterprise production and operation activities, strongly affecting enterprises’ innovation and trade behavior. Enterprises’ import behavior of intermediate goods is similar to their export behavior, with fixed costs and variable costs that intermediate goods imports enterprises cannot avoid. For example, when enterprises enter the import market, they often need to invest certain fixed costs to enter new markets to find potential trading partners and conduct comprehensive evaluations of their creditworthiness, reputation, financial capacity and business conditions. The subsequent production processes also require sufficient operational costs to support customs duties, insurance and potential losses in long-distance transportation, packaging, transportation, etc. Furthermore, enterprises also need significant financial strength when facing uncertain market risks in international trade. Therefore, enterprises’ various costs when importing intermediate goods increase their financial pressure. Scholars have suggested that obtaining more trade credit (Wu et al. 2020) and production subsidies (Xu et al. 2017) can alleviate enterprise financial pressure and result in importing more intermediate goods. Using the same logic, the pilot project for the innovative city policy will produce similar effects. In the actual implementation process of the pilot policy, financial subsidies from all levels of government supported by the innovative city policy and the policy effect of reducing enterprise costs affect enterprises’ innovation and research and development (R&D) and their overall production and operation. Thus, enterprises’ import behavior of intermediate goods will be impacted.

The underlying rationale behind the influence of innovative city policies on intermediate goods imports lies in the acquisition and absorption of advanced technology embedded in intermediate input imports. Therefore, in addition to focusing on the growth in the quantity of imported intermediate goods, attention should also be given to the increase in the variety and quality of imported intermediate goods. Considering the incomplete substitutability between foreign and domestic inputs, an increase in the variety of imported intermediate goods can be used in the production of final products by enterprises, and an increase in the variety of imported intermediate goods implies an increase in the types of final products and even new products produced by enterprises (Colantone and Crinò, 2014). Increasing the variety of imported intermediate goods helps enterprises replace outdated intermediate goods with more advanced ones, thereby increasing overall productivity through technology spillover effects (Yu and Li, 2014; Halpern et al. 2015). Moreover, if the increase in the variety of imported intermediate goods tends to increase the degree of differentiation of final products between enterprises, then improving the quality of imported intermediate goods can further promote technology spillover. Typically, the higher the quality of imported intermediate goods, the higher the technical level and manufacturing process of the embedded source country and the stronger the technology spillover effect (Blalock and Veloso, 2007). Improving the quality of imported intermediate goods provides domestic enterprises with more opportunities to access high-quality products or advanced technologies from abroad, enabling them to imitate and learn advanced foreign technologies or production processes. On the basis of the above analysis, this paper proposes the first hypothesis.

Hypothesis 1: The innovative city policy promotes the import of intermediate goods for enterprises in terms of quantity and, potentially, in terms of variety and quality improvement.

There are many determinants of intermediate goods imports, with cost undoubtedly being one of the most important, especially for micro-enterprises. According to the “Guidelines for Building Innovative Cities,” specific aspects of policy implementation and guarantees include strengthening organizational leadership, increasing policy support, improving government innovation governance, fostering a fair and orderly market environment, and enhancing promotional leadership. The subsidy benefits formed by policy support and the institutional costs reduced by government governance directly impact enterprises’ imports of intermediate goods. This study does not negate the influence of other factors, such as transportation distance and costs, or enterprises’ innovation capabilities, as the effects of these factors can be effectively controlled within the multi-period DID and PSM-DID framework. Therefore, this paper proposes that the pilot project for the innovative city policy will impact the import of intermediate goods for local enterprises through the following three channels: government subsidies, financing constraints, and institutional transaction costs.

The first channel is the government subsidies acquisition effect of the innovative city policy pilot. In building innovative cities, preferential policies and assessment methods are the central government’s primary tools to implement pilot projects for the innovative city policy. Against the backdrop of China’s decentralized system and promotion incentives, local governments at all levels will also introduce relevant similar policies to provide support. The most representative policy uses financial subsidies and rewards as the primary means to support innovation (Yan, et al. 2021). Enterprises in innovative national pilot cities can enjoy more financial subsidies from the government, which can directly increase the current capital stock of enterprises, increase their disposable funds and overcome the greater risks of international trade. This approach increases the import quantity of intermediate goods and positively influences the quality and variety of intermediate goods imports (Xu, et al. 2017).

The second channel is the financing constraints alleviation effect of the innovative city policy pilot. The availability of credit and financing constraints are also significant factors restricting enterprise imports of intermediate goods (Li and Peng, 2014). Xu and Mao (2019) reported that alleviating financing constraints increases the possibility of enterprises entering the import market and significantly improves the variety, quota and quality of imported products (Wei, et al. 2019). In essence, China’s innovative city policy can be seen as a ‘demonstration-type’ place-based policy that targets advanced regions, meaning that the government provides preferential and special policies to promote the economic development of specific areas. Compared with regions where the innovative city policy has not been implemented, pilot cities under ‘demonstration-type’ place-based policies provide more favourable policies at the micro-regional level to local enterprises and create specific resource allocation effects. Additionally, they can improve the market-oriented system and increase credit support from commercial banks to enterprises, thereby alleviating enterprises’ financing constraints. Therefore, the resource allocation effect triggered by the innovative city policy pilot can help alleviate financing constraints for enterprises and expand the import of intermediate goods.

The third and final channel is the institutional transaction cost reduction effect of the innovative city policy pilot. Institutional transaction costs refer to the costs that enterprises must bear when complying with a series of rules and regulations set by the government, which significantly impact enterprises’ imports of intermediate goods. Reducing institutional transaction costs can reduce the import costs of enterprises (Liu, 2016) and improve business efficiency and social benefits through improving the quality and efficiency of intermediary services and the external market and business environments. Reducing institutional transaction costs can effectively alleviate the pressure of import costs for enterprises, promote smoother import processes, save time and energy and enable more companies to enter the trading market quickly and conveniently, significantly alleviating the burden on small and medium-sized enterprises. Based on the above analysis, this paper proposes the second hypothesis.

Hypothesis 2: The innovative city policy promotes the import of intermediate goods for enterprises through obtaining government subsidies, alleviating financing constraints and reducing institutional transaction costs.

The innovative city policy pilot is essentially a production cost intervention policy, which, to some extent, is a continuation of government intervention or a government incentive policy to promote enterprise innovation. The government’s ability to identify enterprise innovation levels and effective regulatory capabilities are prerequisites for successfully implementing innovation incentive policies. However, due to information asymmetry and imperfect regulation, the Chinese government faces the dual dilemmas of moral hazard and adverse selection in implementing innovation incentives and subsidy policies (An et al. 2009). Moreover, owing to China’s decentralized system and promotion incentive mechanisms (Zhou, 2007), all levels of government hope that the established stimulus inputs can maximize output and added value. The government tends to allocate more social resources to enterprises that ‘appear to have innovative capabilities’, meaning that state-owned enterprises, large scale enterprises and more mature enterprises that can demonstrate innovative capabilities are more likely to be favoured by local governments. Moreover, owing to the ‘paternalism’ of various levels of local governments (Huang and Li, 2016), state-owned enterprises and large local enterprises (usually enterprises with a longer history) have closer relationships with the government, to some extent, helping the government shoulder a large number of policy and social tasks (Lin and Li, 2004). These enterprises will be allocated additional social resources and assist local governments in fulfilling their assessment tasks. Therefore, there may be a resource bias towards ‘specific’ enterprises in the implementation process of the innovative city policy and its supporting related policies. This bias leads these enterprises to obtain more resources and generate heterogeneous effects on the import of intermediate goods. Furthermore, China’s unique role in processing trade should not be overlooked. On the one hand, enterprises that engage primarily in processing trade within the global production network and vertical specialization production pattern face greater control and restrictions from developed countries. On the other hand, owing to its unique trading mode, processing trade differs significantly from general trade in regard to obtaining preferential government policies. On the basis of the above analysis, this paper proposes the third hypothesis.

Hypothesis 3: The impact of the innovative city policy on intermediate goods imports varies among enterprises of different size, age, ownership and trade mode.

Model, variable and data

Econometric model

This paper adopts the multi-period differences-in-differences (DID) model proposed by Beck et al. (2010), effectively handling cases where policies are gradually implemented. Like traditional DID methods, the multi-period DID model also requires the construction of three policy evaluation elements: the policy treatment (Treat), the outcome (Outcome) and the control group (Control Group). Using the reference system constructed from the control group, the effect of the ‘treatment’ on the ‘outcome’ can be evaluated. The city, year, industry, and firm in Eq. (1) are represented by i, t, k, and f, respectively. In this paper, the variable ‘the innovative city policy’ (Treatment) \({T}_{{it}}\left({{Treat}}_{{it}}\right)\) represents whether city i implemented the innovative city policy pilot in year t. If city i is designated an innovative city by the country in year t, then from year t \({T}_{{it}}=1\); otherwise, \({T}_{{it}}=0\). \({{IIG}}_{{fit}}\) denotes Outcome, which represents the behavior of manufacturing firms in importing intermediate goods (IIGs). The following econometric model tests the impact of the innovative city policy pilot on the quantity, variety and quality of intermediate goods imported by manufacturing firms.

$${{IIG}}_{{fikt}}=\alpha +\beta {T}_{{it}}+\delta {X}^{* }+{firm}{{\_}}{FE}+{year}{{\_}}{FE}+{\varepsilon }_{{fikt}}$$
(1)

In the above equation, \({X}^{* }\) is the control variable, \({firm\_FE}\) and \({year}{\rm{\_}}{FE}\) represent firm and year fixed effects, respectively. The coefficient β measures the difference in intermediate goods imports before and after implementing the innovative city policy pilot.

Indicator construction and variable selection

This paper measures the import quantity, variety and quality of intermediate goods to comprehensively examine the impact of the innovative city policy pilot on enterprises’ import behavior of intermediate goods. The import quantity of intermediate goods (Y1) is measured by the natural logarithm of the import quantity of intermediate goods by enterprise i in year t. The import variety of intermediate goods (Y2) is obtained with the method used by Bas and Strauss-Kahn (2014). The import quality of intermediate goods (Y3) is measured via a standardized indicator for intermediate goods quality, as used by Shi and Zeng (2015) and Zhao et al. (2017).

Other control variables (\({X}^{* }\)) that influence enterprises’ import of intermediate goods include enterprise size (Size), measured as the natural logarithm of total assets; enterprise age (Age), calculated as ln (year-year of establishment+1); capital intensity (Capital), calculated as ln (firm’s net fixed assets at the end of the year/number of employees+1); the leverage ratio (Level), the ratio of total liabilities to total assets; ownership type (Ownership), a binary variable indicating whether the firm is state-owned (1) or not (0); export status (Export), measured by the value of exported goods in the current year (1 if the value is greater than 0, otherwise 0); and total factor productivity (TFP), estimated via the OP method (the specific estimation process follows Kang et al. (2018)). The industry competition factor (HHI) that companies are facing is defined as the industry Herfindahl–Hirschman Index, which is calculated on the basis of the sales of companies in the industry’s top half. Table 2 presents detailed descriptive statistics.

Table 2 Descriptive Statistics.
Full size table

Data processing

The micro-enterprises data used in this study were obtained from the China Industrial Enterprise Database and the China Customs Import and Export Database on the epsdata platformFootnote 1. China Industrial Enterprise data provide detailed information on the names, legal codes, establishment years, and financial indicators of China’s industrial enterprises above a certain scale. Following the approach use by Brandt et al. (2012), the data were processed as follows. Firstly, samples with missing key indicators such as total output value, total sales, and legal codes were removed. Secondly, according to accounting principles, samples with total assets less than current assets, fixed assets, and net assets were eliminated. Finally, mining-related industries relying on natural resources (industries 6–12) were excluded from the database, retaining data from 33 manufacturing industries (industries 13–46). The China Customs Import and Export database records detailed information on the import and export goods under the HS8 code in various regions, including enterprise customs codes, export amounts, and quantities. For these data, samples with missing information were first removed, and then, following the BECFootnote 2 classification, only data related to intermediate goods imports were extracted. This study referred to international standard codes and classified the data according to the BEC classification. After both databases were cleaned, enterprise names and years are matched following the approach of Tian and Yu (2013) to obtain the merged data used in this study. The policy data used in this study related to the Innovative City Pilot Policy (ICP) mainly come from the “Guidelines for Building Innovative Cities” released by the National Development and Reform Commission and the Ministry of Science and Technology of China in 2016Footnote 3.

Empirical test and analysis

Benchmark regression

Columns (1) and (2) in Table 3 show the results for the import quantity of intermediate goods as the dependent variable. Columns (3) and (4) show the results for the import variety of intermediate goods as the dependent variable. And columns (5) and (6) show the results for the import quality of intermediate goods as the dependent variable. Taking the benchmark regression results for the import quantity of intermediate goods under the innovative city policy pilot as an example, Column (1) in Table 3 controls only firm and year fixed effects, providing a basis for comparison. The results indicate that the estimated coefficient \(\beta\) is 0.35, which passes the significance test at the 1% level. Column (2) in Table 3 introduces control variables, and the estimated coefficient \(\beta\) is 0.421, which also passes the significance test at the 1% level. The results in Columns (1) and (2) indicate that implementing the innovative city pilot policy increased the number of intermediate goods imported by the manufacturing enterprises in the respective cities. The estimated coefficients of the other control variables generally align with our expectations. Regardless of whether control variables are included, the estimated coefficients for the number of intermediate goods imported by local enterprises under the innovative city pilot policy are positive and pass the significance test at the 1% level. Combined with the regression results for the import variety and quality of intermediate goods, it can be concluded that implementing the innovative city pilot policy has significantly increased the quantity, variety and quality of intermediate goods imported by local manufacturing enterprises. Thus, Hypothesis 1 is supported.

Table 3 Benchmark regression results.
Full size table

Hypothesis test of common trend and estimation of dynamic effects

The reliable premise for testing via the multi-period DID method is to satisfy the parallel trends assumption. Enterprises’ development trends in regions that implemented the innovative city policy early and those that implemented it later should not exhibit systematic differences before policy implementation. Alternatively, even if differences exist, they should be corrected, meaning that the experimental and control groups have common trends before policy implementation. As a result, using the methodology employed by Beck et al. (2010), the formula for the parallel trends test is as follows.

$${{IIG}}_{{fit}}={\beta }_{0}+{\beta }_{1}{T}_{{it}}^{-3}+{\beta }_{2}{T}_{{it}}^{-2}+\cdots {\beta }_{7}{T}_{{it}}^{4}+\delta {X}^{* }+{A}_{i}+{B}_{t}+{\varepsilon }_{{fikt}}$$
(2)

Where \({{IIG}}_{{fit}}\) represents the import behavior of intermediate goods for enterprise f in year t. T represents the dummy variable for the implementation time of the innovative city policy and the number of periods in which the policy is ahead of or behind. To test this, we replace the policy shock time dummy variable in the baseline DID Model (1) with yearly time dummy variables and estimate it again. Because the data sample period used in this study is from 2005 to 2013 and Shenzhen firstly implemented the innovative city policy pilot in 2008, test is conducted with a 3-period lead and a 5-period lag.

The estimated results are plotted in Fig. 1 to present the test results more intuitively. The horizontal axis represents the number of years since the implementation of the innovative city policy, ‘t-3’ represents the third year before the implementation of the innovative city policy, and ‘t + 5’ represents the fifth year after the implementation of such a policy. The solid line depicts the marginal effects of the innovative city policy pilot on the import quantity of intermediate goods by enterprises, whereas the dotted line represents the 95% confidence interval. Figure 1 shows that during the three years before 2008 (from ‘t-1’ to ‘t-3’), the marginal effects line of the innovative city policy was relatively flat, and the β coefficient did not pass the significance test. In other words, the difference in the number of intermediate goods imported by enterprises between the treatment and control groups could not reject the null hypothesis of zero. The results in Fig. 1 indicate that the variables are statistically insignificant before implementing the innovative city policy. Thus, there is no significant difference in the import behavior of intermediate goods between the treatment and control groups in the sample before implementing the innovative city policy, which meets the common trend hypothesisFootnote 4.

Fig. 1: Results of common trend hypothesis test and estimation of dynamic effects.
figure 1

The solid line depicts the dynamic impact of the innovative city policy pilot on the import quantity of intermediate goods by enterprises, whereas the dotted line represents the 95% confidence interval.

Full size image

Further test using instrumental variable method

The process of determining an innovative city involves steps such as application, discussion, approval, and announcement, with various levels of government from the central to the provincial and municipal levels. The uncertainty of whether and when a city can be selected has an exogenous effect on enterprises. Therefore, granting a city this status is considered an exogenous shock to enterprise operations. Furthermore, the conclusions from the previous section also passed the balance test of the multi-period DID model. To further exclude potential nonrandom effects on the conclusion of the existence of innovative cities, this study further adopts instrumental variable method to test the robustness of the previous conclusion. Referring to Tsoutsoura (2015), in the use of instrumental variables in quasi-natural experimental studies, the quantity of higher education institutions in each city in 1985 is selected as the instrumental variable for the grouping variable (treat) in this study. In terms of relevance, higher education institutions are not only important for cultivating high-level innovative talent, but are also important sources of basic research and innovations in high-tech fields and inevitably are connected with the selection of innovative city pilots. In terms of exogeneity, the historical quantity of higher education institutions in a city in 1985 is unlikely to directly influence intermediate goods imports during the sample period. Thus, the historical variable of the quantity of higher education institutions in 1985 is unrelated to the random disturbance term, meeting the exogeneity requirement of instrumental variables. The test results in Table 4 show that the instrumental variable does not suffer from the weak instrument problem, satisfying both the relevance and exogeneity assumptions and indicating the effectiveness of instrumental variable selected in this study. Therefore, the estimation results in Table 4 indicate that the coefficients of the core explanatory variables are all positive and pass the significance test at the 1% level, showing no significant difference from the estimates in the benchmark regression.

Table 4 Results of instrumental variable estimation.
Full size table

Impact mechanism test

This section constructs an empirical test model based on the interaction term between the innovative city pilot and variables of impact mechanism (i.e., the triple interaction terms \({D}_{{fit}}\) = \({T}_{{it}}\) × \({M}_{{fit}}\)). This approach allows us to examine how the innovative city pilot policy affects enterprises’ import behavior of intermediate goods. The previous analysis indicated that the innovative city pilot policy may affect the import of intermediate goods for manufacturing enterprises through three channels: government subsidies, financing constraints and institutional transaction costs. Therefore, this paper constructs the following three variables of impact mechanism (\({M}_{{fit}}\)). (1) Government subsidies (SUB) are calculated via the formula ln (1 + subsidy income). (2) Financing constraints (FC) are calculated via the SA method proposed by Hadlock and Pierce (2010). According to the KZ index, the variables of enterprise age and size with strong exogeneity are selected to construct FC. Referring to the method used by Lu and Chen (2017), the absolute value of the enterprise SA index is taken and then logarithmically transformed. The larger the FC, the greater the degree of financing constraints for the enterpriseFootnote 5. (3) Institutional transaction costs (PC) are measured by (sales expenses+management expenses+financial expenses)/total profit, referring to the method used by Wang and Feng (2018) and Yang and Li (2020). Then, the following empirical model is constructed.

$${M}_{{fit}}={\alpha }_{1}+{\beta }_{1}{T}_{{it}}+\delta {X}^{* }+{A}_{i}+{B}_{t}+{\varepsilon }_{{fikt}}$$
(3)
$${{IIG}}_{{fit}}={\alpha }_{2}+{\beta }_{2}{D}_{{fit}}+{\beta }_{3}{T}_{{it}}+{\beta }_{4}{M}_{{fit}}+\delta {X}^{* }+{A}_{i}+{B}_{t}+{\varepsilon }_{{fikt}}$$
(4)

Tables 5–7 report the impact mechanism test results of government subsidies, financing constraints, and institutional transaction costs, respectively. The three tables show that the \({D}_{{fit}}\) coefficient is significantly positive. Taking Table 6 as an example, the coefficient of financing constraints is significantly negative, but the \({D}_{{fit}}\) coefficient is significantly positive. This finding indicates that after implementing the innovative city policy, the sample of enterprises facing severe financing constraints imported more and better intermediate goods in terms of quantity, variety, and quality than did the sample with less severe financing constraints. The innovative city policy pilot weakens the negative impact of financing constraints. Importing enterprises also face various fixed or sunk costs and various variable costs incurred in collecting market information. Insufficient funds limit the quantity, variety, and quality of intermediate goods imported by enterprises. The innovative city policy pilot, through three channels of government subsidies, financing constraints and institutional transaction costs, not only increase the quantity of intermediate goods imported by enterprises but also significantly promote the variety and quality of imported intermediate goods.

Table 5 Impact mechanism test results of government subsidies.
Full size table
Table 6 Impact mechanism test results of financing constraints.
Full size table
Table 7 Impact mechanism test results of institutional transaction cost.
Full size table

Heterogeneity and robustness test

Heterogeneity test

This section considers the subsidy preferences of government at all levels in the implementation process of the innovative city policy pilot, as well as the differences in the innovation and import capabilities and willingness of enterprises affected by the policy. As mentioned in the previous section, we divide the total sample of enterprises according to four criteria: enterprise size, age, ownership, and trade mode. The test will be retaken via formula (1) to verify whether Hypothesis 2 holds.

Differences in enterprise size

The samples are divided into large-sized and small- and medium-sized enterprises on the basis of the enterprises’ total assetsFootnote 6. Formula (1) is used to estimate the regression separately for these two types of enterprises. The test results in Table 8 indicate that the innovative city pilot policy only increases the quantity of intermediate goods imported by small- and medium-sized enterprises but has an insignificant effect on variety and quality. It is widely regarded that larger enterprises have more sufficient, well-established capital chains and more stronger technological capabilities and reserves. More importantly, their employees’ human capital is higher, making them more capable of importing a wide variety of high-quality intermediate goods and fully utilizing the ‘scale effect’ and ‘quality effect’ of intermediate goods imports to rapidly improve their technological levels and productivity.

Table 8 Heterogeneity test results of enterprise size.
Full size table

Differences in enterprise age

This paper divides all enterprise samples into start-up, growing, and mature enterprises on the basis of whether the enterprise age is less than or equal to 6 years, greater than 6 years and less than or equal to 12 years, and greater than 12 years, respectively. Formula (1) is used to conduct sample regression estimation on these three types of enterprises. The results in Table 9 show that, compared with start-up enterprises, the innovative city pilot policy significantly impacts the quantity, variety and quality of imported intermediate goods in growing and mature enterprises. For start-up enterprises, only the estimated coefficient of import quantity of intermediate goods passes the 1% significance test. Relatively mature enterprises have sufficient funds, technological capabilities and human capital. They are more capable and willing to rapidly improve their technological level and productivity through importing a variety of high-quality intermediate goods, utilizing the ‘scale effect’ and ‘quality effect’ of intermediate goods imports.

Table 9 Heterogeneity test results of enterprise age.
Full size table

Differences in enterprise ownership

As analyzed above, a close relationship exists between various levels of government and state-owned enterprises. This closeness leads to a biased policy known as ‘paternalism’, making state-owned enterprises more likely to be the focus and target of policy implementation. On the basis of the ownership nature of enterprises, they are divided into state-owned and non state-owned enterprises. Formula (1) is used to estimate the regression separately for these two types of enterprises. The results in Table 10 show that, for the sample of state-owned enterprises, only the import quantity of intermediate goods passed the significance test. In contrast, all variables for the sample of non state-owned enterprises passed the significance test at the 1% level. The following reasons may explain these results.

Table 10 Heterogeneity test results of of enterprise ownership.
Full size table

First, non state-owned enterprises, especially private enterprises, depend more on the global value chain. Owing to their technological level and productivity, many private Chinese enterprises join the global value chain as subcontract manufacturers, relying on many imported intermediate goods to enter the low-value-added production process in the division of labor system. These enterprises are often in a situation where they are ‘captured’ by large international buyers and multinational companies. Subcontract manufacturing enterprises have almost no ability to achieve high-end indigenous innovation, such as product innovation, functional innovation and industrial chain upgrading. However, this can be achieved if they rely on increasing and improving imported intermediate goods, especially the process innovation achieved by meeting the import parameter requirements of contract-issuing enterprises. This situation leads to significant demand for intermediate goods by private enterprises. The above sample tests based on general trade and processing trade also validate these conclusions. Second, state-owned enterprises face soft budget constraints, redundant employees, policy burdens, complex and inefficient principal‒agent dilemmas and insufficient incentives for innovation. These issues cause state-owned enterprises to lack some flexibility and the ability to make quick decisions when stimulated by the innovative city policy (Liang et al. 2011). This results in a relatively solidified behavior of intermediate goods imports, indicating an insignificant impact on the import variety and quality of intermediate goods.

Differences in enterprise trade mode

On the basis of the trade mode classification from the Chinese enterprise customs statistics database, this study categorizes the sample enterprises into general, processing and mixed trade (i.e., enterprises engaged in both processing trade and general trade). This study retains only the samples of general and processing trade and removes the samples of mixed trade for the analysis to highlight the differences between general trade and processing trade, as shown in Table 11. For processing trade enterprises, quantity, variety and quality all passed the 1% significance test. In comparison, general trade enterprises only passed the significance test regarding quantity. In reality, processing trade, compared with general trade, has a greater dependence on the global value chain and a greater likelihood of being captured, leading to a greater reliance on the import of intermediate goods to meet the need to improve export competitiveness.

Table 11 Heterogeneity test results of of enterprise trade mode.
Full size table

Robustness test

Propensity score matching test

This paper uses the propensity score matching (PSM) method to reconstruct the control group, overcome possible sample selection bias issues and further test the impact of local enterprises in innovative and non innovative pilot cities on the import performance of intermediate goods. Referring to the practice of Wang and Kong (2019), k-nearest neighbour matching (k = 4, radius = 0.05) in the calliper is conducted by selecting covariates, calculating tendency scores and matching tendency scoresFootnote 7. A multi-period DID test is conducted via PSM, and the results are shown in Columns (1) to (3) of Table 12. The consistency between the PSM matched results and the benchmark regression results in the previous section confirms the robustness and reliability of the regression results.

Table 12 Robustness test results I.
Full size table

Winsorize treatment

This paper applies a 1% winsorize treatment to the continuous variables of import quantity, variety and quality of intermediate goods for local enterprises to examine whether outliers influence the imports of intermediate goods for enterprises in the pilot policy area. The estimation results in Columns (4) to (6) of Table 12 indicate that the regression results are not affected by outliers, further validating the robustness of the results.

Exclusion of specific samples

Implementing the innovative city pilot policy may be influenced by factors such as city location and city characteristics. Owing to their uniqueness, municipalities directly under the central government, provincial capital cities and separately planned cities may have systemic differences from other cities. Therefore, this study excludes enterprise samples from innovative pilot cities containing these three types of cities. The results are presented in the first three columns of Table 13. The estimated coefficients are all significantly positive, indicating the robustness of the results.

Table 13 Robustness test results II.
Full size table

Substitution of the dependent variable

Different measurement methods for intermediate goods imports may lead to different results. Therefore, this paper replaces the measurement of quantity, variety and quality of the dependent variable with the import intensity of intermediate goods (i.e., the ratio of intermediate goods imports to total imports), as shown in the last column of Table 13. The results remain significantly positive.

Placebo test

Considering that other factors or policies may influence enterprises’ import behavior of intermediate goods, leading to biased estimates, and given the variations in the timing of innovation-driven policy impacts, Bai et al. (2022) conducted a placebo test with random policy timing and experimental group selection. This involved randomly selecting the same number of samples as the original experimental group from all prefecture-level city samples and randomly generating the timing of policy implementation, thereby creating a treatment group with randomized enterprise samples and policy timing. On the basis of these new samples, 500 random experiments were subsequently conducted to re-estimate the benchmark regression model.

With STATA software, pseudo innovation-driven policy shocks were applied to 135,846 enterprise samples 500 times. Each time, 80,226 enterprises were randomly selected as the experimental group, and regression analysis was performed with randomly generated policy timing. The kernel density of coefficients for the core explanatory variables and the distribution of P values for the import quantity, variety, and quality of intermediate goods were analyzed via 500 regressions, as shown in Fig. 2. The horizontal dashed line in the figures represents a P value of 0.1. If the scatter plot points are below this horizontal line, the coefficient is significant at least at the 10% level; otherwise, it is not significant. The vertical dashed lines represent the true benchmark regression coefficients of 0.421, 0.0217, and 0.285. The placebo test charts in Fig. 2 reveal that most estimated coefficients are concentrated at approximately 0, with many estimates differing significantly from the true coefficients of the benchmark regression, and most P values are above 0.1, indicating insignificance. Therefore, other unobserved factors do not significantly influence the innovation-driven policy effect, demonstrating robust results.

Fig. 2: Placebo test results for the import quantity, variety and quality of intermediate goods.
figure 2

It shows the kernel density of coefficients for the core explanatory variables and the distribution of P values for the import quantity, variety, and quality of intermediate goods, respectively.

Full size image

Conclusions and Policy Recommendations

China’s innovative city policy pilot is a urban development mode that aims to drive innovation as its core objective. It is designed to promote coordinated planning, optimized allocation and effective interaction of various innovation resources within a specific region through innovative city policies and develop regionally distinctive and vibrant innovation systems. The policy focuses on innovation input, enterprise innovation, technology transfer, high-tech industries, and technology benefiting the public and the innovation environment. However, the government and academia have not paid enough attention to the potential impact of the innovative city policy on enterprises’ imports of intermediate goods. In reality, the increase in the import quantity, variety and quality of intermediate goods is the source of endogenous technological progress (Romer, 1986), and it is an essential means for developing countries to absorb technology spillovers and achieve technological catch-up (Aghion and Howitt, 1992). According to the above logic, this study uses the 2004–2013 China Industrial Enterprise Database and China Customs Database and takes the innovative city policy as a quasi-natural experiment. The multi-period DID method and other approaches are employed to evaluate the impact of the innovative city policy on local enterprises’ import behavior of intermediate goods, and its impact mechanism and heterogeneity are also discussed. The main research findings are as follows. (1) Implementing the innovative city policy significantly increases the import quantity, variety and quality of intermediate goods by pilot city enterprises. A series of robustness tests show that this finding is still valid. (2) Implementing the innovative city policy significantly promotes local manufacturing enterprises’ import behavior of intermediate goods through three channels: obtaining government subsidies, alleviating financing constraints and reducing institutional transaction costs. (3) The impact of the innovative city policy on enterprises’ import behavior of intermediate goods is significantly heterogeneous and varies with enterprise size, age, ownership and trade mode.

The policy recommendations for this paper are clear given the aforementioned conclusions. With respect to the impact of the innovative city policy on enterprises’ import of intermediate goods, the government should guide and support the import of intermediate goods when formulating and implementing innovative city policies. It should focus more on improving the variety and quality of intermediate goods to fully utilize the embedded advanced technologies and leverage the “learning through import” effect of intermediate goods. The government can develop differentiated support policies for enterprises of different size, age, ownership and trade mode to stimulate various enterprises to fully utilize the “economies of scale” and “quality effects” of intermediate goods imports, thereby enhancing their own research and development (R&D) innovation levels through better technology spillovers and learning absorption. The government should also evaluate and adjust the Innovative City Pilot Policy (ICP) to ensure that the policy implementation effects align with the expected goals. For the innovative city policy, it does indeed have an impact on the quantity, variety, and quality of intermediate goods, and the import of intermediate goods is extremely essential for improving the technological level in developing countries. However, whether the monitoring and evaluation indicators of innovative city construction should include intermediate goods imports or related indicators needs further research, especially the relationship between intermediate goods imports and indigenous innovation by enterprises under the innovative city policy pilot. This limitation indicates a direction for subsequent research.