Can a new national system policy for science and technology promote industrial restructuring in China’s national innovation center cities?

Abstract

We establish two types of difference-in-difference (DID) models to study the effects of the policy shocks of the new national system of science and technology on the rationalization and advancedization of industrial structure of national innovation center (NIC) cities. Meanwhile, considering the differences in resource endowment between NIC cities and general cities, we use propensity matching analysis (PSM) to establish matching data on the basis of benchmark data to empirically demonstrate the effect of this policy shock on the industrial structure of NIC cities. The study shows that the policy shock has a promoting effect on the level of industrial structure rationalization in the NIC cities, but not significant effect on the industrial structure advancedization, the improvement of the financing environment can promote the development of industrial structure advancement, the growth of personal wealth has a promoting effect on the industrial structure rationalization, and the increase of the government investment is not conducive to the advancement of industrial structure, although it can promote the rationalization of the industrial structure in China’s NIC cities. Compared with the eastern NIC cities, this policy shock has a greater effect on the rationalization of industrial structure in the Central-western NIC cities of China. In addition, the financing environment and per capita income of NIC cities have a moderating effect on the rationalization of industrial structure under the policy shock. Therefore, NIC cities should actively promote the promotion effect of this policy on the rationalization of industrial structure.

1 Introduction

In the process of high-tech industry development, the use of policies to promote the agglomeration and optimization of regional innovation resources is the main way for the government to accelerate the transformation and upgrading of regional industries. Common policy impacts include financial and tax policy support under the dominance of market resource allocation, policy promotion under the dominance of industrial planning, and industry promotion under the synergy of government and market. Relying on the region’s educational and research resources, professional and technical talents, and favorable financing environment, Silicon Valley in the United States has created many multinational companies such as Hewlett-Packard, Intel, Cisco, and Apple with the support of tax and other policies (Michel and Granovetter 2009). The Bangalore Software Park in India, relying on the government’s "Software Technology Park Program", has brought the Indian software industry into the top three in the world by coordinating the supply of innovative resources through the government, combining preferential industrial policies and an efficient park management system (Saleman and Jordan 2015). Shanghai Specialty Industrial Park, which was established in 2020, is a unique industrial park in China. Industrial Park is guided by government planning and policies, relying on urban endowments to promote the participation of market players, forming a development model under the coordination of the government and the market (Wang et al. 2023).

From the experience of various countries, the development of science and technology industry is generally concentrated in economically developed cities, especially within the scope of innovation center cities where innovation resources are concentrated. One reason is that in innovation center cities, industrial resources and structure are more in line with the needs of science and technology industry development, and the second is the important supportive role of developed cities’ industries to regional development, making the government more incentivized to promote the matching of cities’ scientific and technological resources with industries with policies, so as to promote the upgrading of cities’ industries. Therefore, the focus of promoting the rapid development of the science and technology industry is to promote the rationalization and advancedization development of the industrial structure of innovation center cities. However, the focus of policy shocks may be different in different economies. In Europe and the United States, where the development of urban industries has had a longer history, the transformation and upgrading of urban industries has been accompanied by the advanced development of industrial resources, and the degree of matching of human, financial and material resources with industrial demand is higher in the cities, the policy shocks are mainly aimed at promoting the rationalization of resources within the science and technology industry and further promoting the advanced development of the industry (e.g., the integration of the manufacturing industry with the service industry). Taking the Silicon Valley of the United States as an example, the focus of its management is to promote the efficiency of the market circulation of innovation resources, so that innovation subjects can quickly obtain the resources needed for the development of science and technology industry through market behavior (such as purchasing and technology exchange, etc.). In contrast, developing economies have a relatively short period of industrial development, with the industrial resources at the middle and low-end levels of the city having a low degree of match with the technological characteristics of the science and technology industry, resulting in a high degree of dependence on foreign countries in terms of industrial technology. The main purpose of the policy is to use the government’s executive power to rapidly allocate industrial resources, enhance the rationalization of the city’s industrial structure, and further promote the advanced industrial structure. Just as the way of the construction of Bangalore in India and Shanghai Specialty Industrial Park in China, the government firstly chooses the cities with better economic and innovation conditions as the basis for developing industries, and further provides policy support and promotion plan for the city’s science and technology industry, and promotes the development of the high-tech industry in the form of the government allocating resources. Therefore, whether in developed or developing economies, innovation center cities are the main carriers of policy shocks in the process of the government’s efforts to promote the rational allocation of urban scientific and technological industrial resources and the rationalization and advancedization of industrial structure. In addition, since urban endowment resources such as manpower, capital and supporting facilities affect the supply of urban industrial resources on the one hand, and on the other hand, are subject to constant changes under the influence of industrial policies, urban endowment resources may play an important role in the process of policy influence on industrial structure.

In this paper, we study the effects of policy shocks to the new national system of science and technology on the rationalization and advancedization of industrial structure of cities. Due to the differences in endowment conditions among cities, there may be differences in the effects of the same policy shock on the industrial structure of the national innovation center (NIC) city and the general city, therefore, the specific effects of the policy shock on different cities need to be discussed. In addition, in order to discuss the role of the geographic location of NIC cities in the process of the effects of policy shocks on the industrial structure, the NIC cities are divided into two categories, namely, Eastern and Central-western cities, according to their geographic locations. Based on the benchmark data and matching data of the two types of cities, we aim to answer the following questions: (1) Whether the effect of this policy shock is significant in the role of rationalization and advancedization of industrial structure of NIC cities. (2) In the process of the effect of this policy shock on industrial structure, the direct and indirect effects of the city’s financing environment, personal wealth and government investment on the rationalization and advancedization of industrial structure, as well as whether the effect of this policy shock on the rationalization and advancedization of urban industrial structure is the same for NIC cities in different locations.

To investigate these issues, we use a difference-in-difference (DID) model to analyze the role of a policy shock on the rationalization and advancedization of industrial structure of a city, which involves a policy shock of a new national system of lifting the state in science and technology, the criteria for delineating a NIC city from a city in general, and three types of factors affecting urban financing, wealth, and government inputs. First, based on the city’s economic and demographic data, we construct and calculate the indicators of rationalization and advancedization of the city’s industrial structure as the dependent variable of the DID model. Based on the Chinese government’s documents on the construction of innovative cities, national science centers and national science & innovation centers, we screen out NIC cities and general cities, determine the time point of the policy shock, and form a benchmark dataset to test the relationship between the role of this policy shock on the rationalization and advancedization of the industrial structure. Second, using city financing, personal wealth and government input as covariates, we use propensity matching analysis (PSM) to find a control group of cities similar to the NIC cities to form a matching dataset, and test the role of the policy shock of the new national lifting system for science and technology on the rationalization and advancedization of the industrial structure with this data. Based on the comparison of the results of the two types of data, we derive the specific role of this policy shock on the rationalization and advancedization of industrial structure. We find that the policy shock has a significant contribution to the rationalization of industrial structure in NIC cities, but not to the advancedization of industrial structure. Meanwhile, the improvement of the financing environment promotes the industrial structure advancedization, but has no significant effect on the industrial structure rationalization; the growth of personal wealth promotes the industrial structure rationalization, but has no significant effect on the industrial structure advancedization; the growth of the government’s supporting inputs promotes the industrial structure rationalization, but the growth of the expenditures on science and technology industry is not conducive to the transformation of the city’s industries into advanced ones. In the extended discussion, there are geographical differences in the effect of this policy shock on the rationalization of industrial structure. In the extended discussion, there are geographical differences in the role of this policy shock on the rationalization of industrial structure. The policy shock has a stronger effect on the industrial structure of cities in central and western NIC cities, and the policy shock has a significant inhibiting effect on the rationalization of the industrial structure of cities in eastern NIC cities, suggesting that providing policy support for a particular industry may not always be optimal in terms of government allocation of resources, although it accelerates the agglomeration of resources in that industry. For Central-western cities, government allocation of industrial resources may be more efficient due to the low level of marketization in the region, while for Eastern cities, government allocation of resources may interfere with the original market allocation due to the high level of marketization in the region. We also find that financing environment, per capita wealth and government investment have moderating effects in the process of rationalization and advancedization of industrial structure of NIC cities affected by policy shocks, suggesting that NIC cities should actively act to bring out the positive roles of the above three types of factors in this process.

The remainder of the paper is organized as follows. Section 2 reviews the relevant literature and formulates the hypotheses. Section 3 describes the modeling framework and Sect. 4 presents our findings. We conclude the paper and give policy implications in Sect. 5.

2 Literature review and research hypothesis

2.1 Literature review

Several aspects of directed innovation policies have been studied. Mowery et al. (2010) develops a neo-Schumpeterian framework emphasizing national innovation systems and analyzes the macroeconomic effects of national directed innovation policies. Acemoglu et al. (2012) incorporate directed technological innovation policies into an environmentally constrained growth model by analyzing temporary taxes or subsidies to promote the adoption of new technologies to achieve sustainable growth. Mazzucato (2018) further summarize the key features of mission-oriented innovation policies, pointing to the investment-pull effect of directed policies on economic growth. Deleidi et al. (2020) simulate mission-oriented innovation policies, carving out the fiscal multiplier effect of policy shocks, arguing that policies can have a crowding-out effect on private R &D but that policy shocks are effective in addressing the key technological problems and create market demand. Their studies show that governments can promote market and economic growth through directed innovation policies. Unlike the above literature, we take cities as the main target of the policy shock and study the effect of this policy shock on the industrial structure of cities.

Our study is related to the literature that considers the role of innovation systems for industrial development. Lundvall et al. (2002) construct a structured and institutionalized national system of innovation by analyzing the ways in which the state, industry and firms organize and manage resources, and establish an analytical framework for innovation and economic performance and industrial development. Walrave and Raven (2016) argue that technological innovations promoted by the state through policy should match local industrial needs, otherwise it is difficult to mobilize local participation in technological innovation. Pipkin and Fuentes (2017) use the “inducement-search” model to find that the motivation of firms in developing countries to initiate industrial upgrading mainly comes from the impact of national policies, and that the institutional environment determines the final industrial upgrading effect. On this basis, Granstrand and Holersson (2020) and Erzurumlu et al. (2022) argue that the regional innovation ecology jointly constructed by the state and the locality is the foundation for promoting the upgrading of the local industry. Donges et al. (2023), by analyzing institutional inclusiveness, argued that institutions are the primary determinants of innovation, and that the key mechanisms of regional industrial development result from the synergy of national and local innovation agents and the policy environment. In contrast to the above literature, we argue that innovation systems allocate regional industrial resources through innovation policies; however, the differences in the level of regional development make differences in the effectiveness of government allocation of innovation resources.

In our model, policy shocks accelerate the agglomeration of resources to a specific industry and promote the rationalization and advancedization of industrial structure. Gambardella and Mcgahan (2010) suggests that general technology enterprise in the industry chain, based on the advantage of intellectual capital, accelerate the agglomeration of industrial technology resources with licensing and transfer, which becomes the key to promote industrial upgrading. Through a discussion of the evolutionary framework of firms and the national innovation system, Lundvall and Rikap (2022) argue that policy support has enabled Chinese technology firms to gain access to more resources, and that the clustering of resources by technology firms has driven the structural upgrading of the technology industry. All these papers take enterprises as the research object and discuss the role of enterprise resource agglomeration on industrial upgrading supported by policies and institutional systems. In contrast, we consider the research object to be the NIC cities and consider the effect of policy shocks on urban industrial resource agglomeration. Although some people have studied the effect of policy shocks on industrial structure (Zheng et al. 2021), our question differs from theirs in that policy shocks affect resource agglomeration rather than discussing the role of policy compensation on industrial profits.

Our model includes the interaction of policy and local resources, and thus may be relevant to the economic literature on policy incentives, market openness controls, and some scholars have investigated the role of industrial policy and market openness on the efficiency of resource allocation (Andreoni and Chang 2019; Stiglitz 2021; Chen et al. 2022). Criscuolo et al. (2019) discuss the different impacts of investment subsidy policies on the promotion of employment in small and large firms under the role of industrial policies and market mechanisms. Unlike (Criscuolo et al. 2019), our model focuses on the role of national science and technology promotion strategies on the industrial structure of cities with different resource endowments, rather than the effect of a specific policy on local firms.

Finally, our paper is relevant to the neo-structuralist literature on economic catch-up in late-developing countries. Our model reveals the role of policy shocks on the industrial structure of cities at different levels of economic development, which is reminiscent of Justin (2011), who advocate that the formulation of industrial strategies should be based on the allocation of roles between the government and the market in order to accelerate the flow of resources from the inefficient sector to the efficient sector. Based on the neo-structuralist framework, Dutt (2019) explores the possibility of a state-led Latin American possibility of industrial structural transformation. In our question on policy shocks, the implementation of policy shocks should be weighed against the level of regional marketization, and policy shocks are more likely to promote the rationalization of local industries in China’s central-western regions, where the level of marketization is low, while policy interventions in industrial resource allocation should be reduced in China’s eastern regions, where the level of marketization is high.

2.2 Research hypothesis

From the point of view of duration and technical support, the policy of the new national system of science and technology to promote the development of local industries is mainly manifested at three levels: (1) short-term local transformation of national technological achievements to support the development of local industries; (2) medium-term national support for local major technological special projects, where the state and the localities jointly invest innovative resources, formulate industrial development programs and implement management; (3) long-term construction of NIC cities to provide technology supply and policy support for industrial upgrading. As the fulcrum of national regional innovation, NIC cities are the most powerful platform for the new national system of science and technology policy to promote local industrial development, and also the main target of science and technology policy impact. The policy impact of the new national system policy of science and technology on the NIC city accelerates the process of optimization of the city’s industrial structure and system (Taalbi 2020). The science and technology industry, with policy support, rapidly acquires and optimizes innovation resources, which makes labor productivity improve and leads to a decline in the relative cost of capital, labor and other factors. With the agglomeration of high-tech industries in the city, the urban industrial structure is optimized (Zhu et al. 2021), and industrial profits are improved, prompting city managers to focus on forward-looking industrial layout, promote the integration of the innovation chain, industrial chain and capital chain, improve the synergy and benefit-sharing mechanism among industrial subjects, and promote the adjustment and upgrading of the urban industrial structure. Based on the above analysis, hypothesis 1 is proposed.

H1: The policy of the new national system of science and technology can promote the upgrading of the industrial structure in the cities of NICs.

Industrial space theory suggests that the direction and speed of industrial upgrading are determined by the spatial location of an economy (Hausmann and Klinger 2007). Differences in the level of marketization between regions in China have resulted in unbalanced factor flows and economic development between regions. The Eastern region has opened up to the outside world and technology exchange earlier, and under the higher level of marketization and relatively mature industrial support, the industrial subjects are able to obtain resources such as capital, talents and technology from the market quickly, and the industries in the region are evolving from incremental expansion to stock optimization. On the other hand, although the Central-western regions are rich in natural resources, the relatively backward level of marketization hinders the flow of factors, coupled with the loss of talents and insufficient policy support, which makes the transformation rate of scientific and technological achievements in the region lower (Shahzad et al. 2022), and the development of high-tech industries is relatively slow. From the viewpoint of the policy purpose of the new national lifting system of science and technology, in the Eastern NIC cities, the policy mainly promotes the optimization of the city’s industrial structure and makes up for the market’s failure in resource allocation. While in the Central-western NIC cities with a lower level of marketization, the efficiency of allocating resources by the government may be higher, which is more conducive to the development of high-tech industries in the region, thus the policy impact may be more favorable to the industrial development of the Central-western NIC cities, based on the above analysis, hypothesis 2 is proposed.

H2: In comparison with the coastal national innovation cities, the science and technology policies under the new national lifting system have a stronger role in promoting industries in the inland areas, especially in the less market-oriented cities.

3 The model

We construct a comparative experiment of the impact of practicing a certain policy (e.g., the new national system policy of science and technology) on the evolution of the city’s industrial structure between the NIC cities (such as Beijing, Chengdu, Hefei, etc., with the total number of N1 cities) and other cities (such as Tianjin, Jinan, Zhengzhou, etc., with the total number of N2 cities), with the total number of cities in the two categories being $N=N1+N2$. As stated in the literature review, the promotion of urban industries is the main driving force behind the practice of this policy in both types of cities (Walrave and Raven 2016), and therefore, we evaluate the policy effects in terms of the evolutionary characteristics of urban industrial structure. Specifically, we portray the evolution of industrial structure in terms of two dimensions: rationalization of industrial structure (denoted as $\textrm{TL}$) and advancedization of industrial structure (denoted as $\textrm{IS}$). The industrial structure rationalization $\textrm{TL}$ reflects the degree of coordination and resource utilization among the city’s industries, as well as the coupling of the city’s factor input structure and output structure, so the Theil index is used to portray the rationalization of the industrial structure of city $i (i=1, 2,..., N)$,

$$\begin{aligned} \textrm{TL}_{i} =\sum _{p=1}^{3} \left( \frac{Y_{ip} }{Y_{i}} \right) \ln \left( {\frac{{Y_{ip} }/{Y_{i}} }{{L_{ip}}/{L_{i}} } }\right) , \end{aligned}$$

(1)

where the urban industry is divided according to three industries, that is, $p = 1, 2, 3$, denoting the primary (i.e., agriculture), secondary (i.e., manufacturing), and tertiary (i.e., services) industries, respectively. $Y_i$ represents the total output value of city i. $Y_{ip}$ represents the output value of industry p in city i. $L_{ip}$ represents the number of employees in industry p in city i, and $L_i$ represents the total number of employees in city i. All of these data can be obtained from the statistical reports published by the government. The Theil index reflects the degree of deviation of the industrial structure and output value of different cities, the closer to 1, the greater the degree of deviation of the industrial structure, the more unreasonable the industrial structure, the closer to 0 means the more balanced the industrial structure, the higher the degree of rationalization of the industrial structure. Industrial structure advancedization ($\textrm{IS}$) reflects the dynamic evolution of industries in a region from primary to secondary and from secondary to tertiary industries. In general, the industrial layout of cities in developing countries (e.g., China) is usually dominated by the secondary industry, and the process of industrial sophistication is mainly from the secondary industry to the tertiary industry (that is, the upgrading from the primary industry to the secondary industry has been completed). In order to portray this characteristic, we measure the industrial structure advancedization of a city only by the ratio of the output value of tertiary industry to that of secondary industry in the city i,

$$\begin{aligned} \textrm{IS}_{i} =\frac{Y_{i3} }{Y_{i2} }, \end{aligned}$$

(2)

the indicator $\textrm{IS}$ reflects the development trend of the city’s transition from manufacturing to services, and the larger the value of $\textrm{IS}$, the higher the level of industrial sophistication of the city.

The discussion of the net effect of this policy shock (in the context of policy implementation) on the rationalization and advancedization of industrial structure using a DID model requires the identification of a subgroup dummy (denoted as src) and a dummy for the time of policy implementation (denoted as time). In the discussion of sample groupings, the policy’s effect on industrial structure may vary across cities. For example, in the process of breaking through key technologies for industrial development, the main body of research and development is concentrated in national laboratories and innovation demonstration zones, which are concentrated in the cities of NICs, where the policy impact of the new national system of science and technology may have a stronger effect on such cities. The cities with general innovation endowment, limited by the lack of innovation resources and capacity, are difficult to effectively promote the policy to develop key technologies, and the policy impact on the industry of such cities has limited effect. Therefore, in the process of city grouping, the NIC cities are set as the treated group of the policy, and the grouping variable src is taken as 1, while the other cities are set as the control group, and term src is taken as 0. As for the recognition of the timing of policy shocks, the implementation of local government policies mainly comes from the guidance of the central government, and the time when the central government asks localities to promote the policy can generally be regarded as the starting point of the policy shock. For example, China proposed in 2015 to take advantage of the new national system in major science and technology projects, therefore, in the setting of the policy time node, the time dummy variable time before 2015 can be taken as 0 to indicate that cities were not subjected to the policy shock before 2015, while the time dummy variable time after 2015 can be taken as 1 to indicate that they were subjected to the policy shock from 2015 onwards.

As the calculation of industrial structure rationalization and advancedization indicators is based on the output value and employment of each industry, as described in Eqs. (1) and (2), and these factors are affected by policies. In order to judge the effect of policy shocks on industrial structure, we use DID models to explore the effects of cities in terms of the role of policy shocks on the rationalization (TL) and advancedization (IS) of the industrial structure of NIC cities and general cities. We first use a DID model without control variables to initially explore the effects of policy shocks on the industrial structure of cities, that is

$$\begin{aligned} \textrm{TL}_{it}= & {} \alpha _{0} +\alpha _{1}\textrm{src}_{i} +\alpha _{2}\textrm{time}_{it} \nonumber \\{} & {} +\alpha _{3}\textrm{src}_{i}\times \textrm{time}_{it} + \varepsilon _{it}, \end{aligned}$$

(3)

$$\begin{aligned} \textrm{IS}_{it}= & {} \beta _{0} +\beta _{1}\textrm{src}_{i} +\beta _{2}\textrm{time}_{it} \nonumber \\{} & {} +\beta _{3}\textrm{src}_{i}\times \textrm{time}_{it} + \delta _{it}, \end{aligned}$$

(4)

where the subscript i of each variable denotes each city ($i=1, 2,..., N$), and the subscript t denotes the year ($t=2007, 2008,..., 2020$). $\varepsilon _{it}$ and $\delta _{it}$ are the error terms, independent and obeying normal distribution. The intercept terms $\alpha _0$ and $\beta _0$ denote the level of rationalization and advancedization of industrial structure when the grouping variable $\mathrm{src_i}$ and the time dummy variable $\mathrm{time_{it}}$ are 0, respectively, that is, the level of rationalization and advancedization of the industrial structure in the control group of cities when they are not subject to policy shocks. The coefficients $\alpha _1$ and $\beta _1$ denote the role of the NIC cities as the treated group on the rationalization and advancedization of the city’s industrial structure, $\alpha _2$, $\beta _2$ denote the role of the city’s industrial structure rationalization and advancedization after the occurrence of the policy shocks, and the coefficients $\alpha _3$ and $\beta _3$ denote the role of the city’s industrial structure rationalization and advancedization after the policy shocks in the treated group. In fact, the coefficients $\alpha _3$ and $\beta _3$ describe the net effect of the policy shock on the rationalization and advancedization of the industrial structure of the treated group, and we use Eq. (3) as an example to explain this process. Before the policy shock ($\textrm{time}_{it}=0$), as a NIC city in the treated group ($\textrm{src}_i=1$), the policy’s effect on the industrial structure is mainly $\alpha _0+\alpha _1$, while the policy effect in the control group ($\textrm{src}_i=0$) is mainly $\alpha _0$. After the policy shock ($\textrm{time}_{it}=1$), the policy effect in the treated group ($\textrm{src}_i=1$) is mainly $\alpha _0+\alpha _1+\alpha _2+\alpha _3$, while the policy effect of the control group ($\textrm{src}_i=0$) is $\alpha _0+\alpha _2$. In terms of the difference scores of the policy effects between the treated group and the control group, the difference between the treated group and the control group before the policy shock is $\alpha _1$, while the difference of the policy effects between the treated group and the control group after the policy shock is $\alpha _1+\alpha _3$. The difference scores before and after the policy shock are further differenced again to obtain the net effect of the policy shock of the treated group on industrial structure, which is obtained as $\alpha _3$. The above process is summarized in Table 9 in appendix A. Similarly, the net effect of industrial structure rationalization can be obtained as $\beta _3$, which is also summarized in Table 10 in appendix A.

The industrial development of a city is affected by factors such as the financing environment, wealth status and public inputs of that city, which are usually used as control variables in empirical analysis. Some literatures use factors such as the level of urban development, urban human capital input and urban fiscal expenditure as control variables (Sun et al. 2016; Chen et al. 2017; Zhong et al. 2022), but in the calculation of the representative indicators of these literatures, there is a certain amount of overlap, with the mask effect between the variables, for example, urban human capital input is calculated as the urban education expenditure as a proportion of urban financial expenditure, this indicator has collinearity with urban financial expenditure, and the assumption of independence between variables is difficult to be satisfied. In this paper, on the basis of considering the independence between variables, we take the three types of variables, namely financial development level (denoted as $\textrm{fd}$), urban per capita income (denoted as $\textrm{pergdp}$) and urban fiscal expenditure (denoted as $\textrm{gov}$), as control variables, represented by the vector ${{\textbf {x}}}=(\textrm{fd}, \textrm{pergdp}, \textrm{gov})$, to measure the roles of urban financing, wealth and governmental inputs on industrial development. Specifically, the level of financial development is calculated as the ratio of total credit to the city’s total output, which reflects the city’s financing capacity. The development of financial markets can effectively promote industrial development. City per capita income is calculated by the average salary of city workers. According to the Petty-Clark Theorem and the Kuznets curve, the increase of city per capita income can promote the mobility of labor from the primary industry to the secondary and tertiary industries, which to a certain extent reflects the level of development of the city’s economy. City financial expenditure data can be obtained through government statistical reports, which to a certain extent reflects the government’s support for the industry.

In this way, we introduce the control variables (vector ${{\textbf {x}}}$) into the analysis of the policy shock effects, and thus Eqs. (3) and (4) are rewritten as Eqs. (5) and (6), which we refer to as the DID model containing the control variables, as a way to analyze the net effect of the policy shocks on the industrial structure rationalization and advancedization. The industrial structure rationalization ($\textrm{TL}_{it}$) and advancedization ($\textrm{IS}_{it}$) of city i at time t is modeled as:

$$\begin{aligned} \textrm{TL}_{it}= & {} \alpha _{0} +\alpha _{1}\textrm{src}_{i}+\alpha _{2}\textrm{time}_{it} \nonumber \\{} & {} +\alpha _{3}\textrm{src}_{i} \times \textrm{time}_{it} +\sum _{j=1}^{3}d_{j}{x_{jt}} + { \varepsilon _{it}}, \end{aligned}$$

(5)

$$\begin{aligned} \textrm{IS}_{it}= & {} \beta _{0} +\beta _{1}\textrm{src}_{i} +\beta _{2}\textrm{time}_{it} \nonumber \\{} & {} + \beta _{3}\textrm{src}_{i} \times \textrm{time}_{it} +\sum _{j=1}^{3}d_{j} x_{jt} +{ \delta _{it}}, \end{aligned}$$

(6)

where $x_{jt}$ denotes three types of control variables ($j = 1, 2, 3$).

In the process of using the above model to discuss the effects of policy shocks, we construct two types of data, namely benchmark data and matching data, for empirical testing. For the benchmark data, mainly based on the city statistics released by the government, we calculate the indicators of industrial structure rationalization ($\textrm{TL}_{it}$) and advancedization ($\textrm{IS}_{it}$) according to Eq. (1) and Eq. (2), meanwhile, we establish the dummy variables of the city grouping and the time of the policy shock, and collect the data related to the control variables to satisfy the needs of empirical analyses. The main reason for constructing matching data is that when analyzing the effects of policy interventions, the ideal is to conduct a randomized experiment, controlling for all other conditions and making these conditions the same, differing only by whether or not an intervention is taken, so that the effects of policy interventions can be understood by comparing the differences between the treated and control groups in a randomized experiment. However, when there is a large difference between the treated and control groups, the assumption of conditional independence between variables (CIA) in the regression analysis cannot be fully satisfied, and thus, it is not possible to accurately determine whether or not the difference between the two groups is due to the policy intervention. To overcome this problem and enhance the robustness of our conclusions, we follow the "propensity matching-difference in difference" method (PSM-DID) proposed by Heckman et al. (1997) to match the treated and control groups, constructing quasi-natural experiments in order to satisfy the CIA condition as much as possible, and forming the matching data of the present study by PSM. The specific process includes four steps: first, define the similarity between the treated group and the control group. Select the factors that can affect the intervention variables and outcome variables at the same time as covariates, and after controlling these covariates, reduce the differences between the treated group and the control group, and with the help of the logit model, establish the linear function of the log likelihood ratio and the three types of covariates, namely, the level of financial development, the per capita income of the city, and the city’s fiscal expenditure, and obtain the propensity score. Second, the treated and control groups are matched based on the similarity principle and the propensity score. One-to-many nearest neighbor matching is used to find multiple individual matches in the control group for each individual in the treated group to enhance the estimation effect. Third, diagnose the matching effect. Matching is to obtain more similar samples, making the treated group comparable to the control group, and after the potential random samples in the observed data are screened out by using the matching method, the balance test is used to determine whether it is close to a randomized trial (King et al. 2016), in order to prepare for the further analysis of the effect of the policy intervention. Fourth, causal estimates of policy intervention effects are obtained by DID model. After obtaining a more reliable sample for analysis, the causal effect of policy intervention is estimated again using the DID model given in Eqs. (5) and (6). In order to enhance the explanatory power of the model, the variables involved in the model are logarithmized and the descriptions of the relevant variables are summarized in Table 1.

Table 1 Description of model variables

Table 2 Descriptive statistical analysis of variables for cities

4 Research process and empirical findings

In this section, the role of this policy shock on the industrial structure of cities is empirically analyzed by taking China’s new national lifting system policy for science and technology as an example. The whole analysis process consists of three aspects: first, the source of sample data and the descriptive statistical characteristics of each indicator are clarified, and the rules for screening NIC cities are given. Second, the methodology given in Sect. 3 is utilized to construct the benchmark data, and regressions are conducted based on the DID models without control variables in Eqs. (3) and (4), and the DID models with control variables in Eqs. (5) and (6), respectively, to assess the policy shock effect. Third, following the matching data construction method in Sect. 3, matching data are created for the regression analysis of Eqs. (3) and (4), and Eqs. (5) and (6) to assess the effect of this policy shock again. Finally, the robustness of the conclusions is discussed by comparing the regression results of the benchmark data and the matching data. In addition, since the geographical nature of cities, as well as the characteristics of urban development, have an impact on the effect of the policy shock, we further discuss the specific effects of the policy shock in different regions, and analyze the moderating roles of urban financing, wealth, and government expenditures in the policy shock, in order to find a focus point for the government to promote the policy.

4.1 Source of data and basis of grouping

The initial data are based on the data released by the National Bureau of Statistics of China for cities at the prefecture level and above, totaling 343 cities, with 2007–2020 as the study interval, and the data are obtained from the Wind database, the CCER database, and China’s national and local statistical yearbooks, including information on the number of employed people in cities, the output value of the primary, secondary, and tertiary industries in cities, the average salary of urban workers, the financial expenditures of the urban government, the total amount of credit in cities, and the GDP of cities and other information. The samples with missing data from 2014 to 2016 are excluded, and interpolation is used to fill in the missing data at the tail end of the study period, finally obtaining data for 285 cities, whose descriptive statistical characteristics are summarized in Table 2.

As described in the model construction section, the DID model requires setting the exact point in time of the policy shock and creating treated and control groups. In this case, we choose 2015 as the point of time for the policy shock, which is a critical period when China’s government proposes to take advantage of the new national lifting system policy for science and technology. By setting 2015 as the time point of the policy shock, we can ensure that the results analyzed by the model are specific to the policy effects triggered by the new national lifting system policy for science and technology. In addition, considering that the long-term impact of the policy is more likely to promote the construction of cities as national innovation centers, we use the cities of national innovation centers as the basis for grouping the treated group and the control group, so that we can more effectively distinguish the specific effects of the policy on different shock targets. However, there is no official statement on the concept of the NIC city in the existing researches, and we establish this kind of city based on the following two principles: firstly, by the end of 2015, the city has completed the national innovation city construction, and second, the city is located within a national science center or national science and innovation center zone. Since national innovation cities have a three-year construction period from the time of approval, NIC cities hit by the 2015 policy should have been approved by the government to start building national innovation cities in 2012 at the latest, and should have completed the construction of national innovation cities by the end of 2015. A total of 41 cities were approved by the government to build national innovation cities before 2012, which are summarized in Table 11 in Appendix A. There are a total of 11 existing national science and technology innovation centers or major cities in national science and technology centers, and we summarize in Table 12 in Appendix A. According to the above screening principles, nine cities (including Beijing, Shanghai, Guangzhou, Shenzhen, Chengdu, Chongqing, Hefei, Wuhan, and Xi’an) are finally selected as the treated group, and other cities as the control group. Geographically the treated group includes two types of cities, one is the Eastern cities, including Beijing, Shanghai, Guangzhou, and Shenzhen; the other is the Central-western cities, including Chongqing, Chengdu, Wuhan, Hefei, and Xi’an.

4.2 Analyzing the shock effects of the policy based on benchmark data

In applying the DID model to analyze policy shocks, the treated and control group industry structure variables need to satisfy the common support assumption prior to the policy shocks, and we examine this condition by using both the parallel trend test and the dynamic trend test. The parallel trend test requires that the treated and control groups have similar development trends before the policy shock. Taking urban industrial rationalization as an example, we use the mean values of industrial structure rationalization of the treated and control group cities in each year to see whether the development trends of the two types of cities satisfy the parallel trend test. Since 2007, China’s economic development has brought about a continuous updating of the urban interface, and one might expect the degree of rationalization of the city’s industrial structure to be increasing, but this conclusion does not always hold. As shown in Fig. 1, we plot the time trend of the mean value of the rationalization of the respective industrial structure (i.e., ln(Theil index)) of the treated and control group cities over time as a function of year. Prior to 2012, as time advances (i.e., the annual value increases), ln(Theil index) values for treatment group cities shows a rising trend, indicating a decrease in the degree of rationalization of the city’s industrial structure, which is mainly due to the fact that most of the treated group cities entered the period of the construction of the national innovative city in the period of 2010–2012, and the cities were in the process of industrial transformation and industrial turnover, as a result of the elimination of outdated industries and the introduction of new industries. During the process of industrial turnover, the mismatch between the original industrial resources and the new industrial structure led to a decrease in the level of rationalization of the industrial structure. In 2015, the treated group of cities all completed the construction of national innovative cities, and the matching of the city’s industrial resources with the structure improved, which contributed to the increase in the level of rationalization of the city’s industrial structure. In the same period, the cities in the control group showed a similar development trend at a much lower level of industrial rationalization, but with smaller fluctuations, without a larger problem of mismatch between industrial resources and structure. Compared with the similar development trend presented by the two groups of cities in the pre-2015 period, after 2015, the industrial rationalization indexes of the treated group no longer develop along the original path, and the level of industrial rationalization continues to increase, while the industrial structure rationalization indexes of the cities in the control group still develop along the original trend, and the two groups of cities no longer have a similar development trend.

Fig. 1

Time trend of rationalization of industrial structure

The dynamic effects test uses the event study approach (ESA) to observe whether there is a difference in industrial structure rationalization between the treated and control group cities from the pre- and post-shock phases of the policy shock, which not only answers whether there is a difference between the two groups of cities before the policy shock, but also describes whether there is a persistence of the policy after the policy shock. Generate a cross-multiplier term (e.g., $\mathrm{src \times year2007, src \times year2008,..., src \times year2020}$) with the treated group and a dummy variable for each year and establish a regression equation between these cross-multipliers and the rationalization of industrial structure, where the coefficients before the cross-multiplier term of the dummy variable and the treated group for each year reflect the difference in the industrial structure between treated group and the control group cities whether there is a significant difference in that year. The specific results are shown in Fig. 2, where we plot the relationship between the coefficients of the cross-multiplier terms before and after the policy shocks over the years, where the short line perpendicular to the horizontal axis is the 95% confidence interval of the regression coefficients of the cross-multiplier terms of the dummy variables and the dummy variables of the treated group in each year. It can be seen that before the policy shock (i.e., before 2015), except for the two annual coefficient confidence intervals of 2011 and 2012 which do not include 0, the confidence intervals of the coefficients of the rest of the annual coefficients include 0, indicating that there is no significant difference between the treated group and the control group in terms of the rationalization of the industry. After the policy shock (i.e., after 2015), the coefficient confidence intervals do not include 0, indicating that the industrial rationalization of the treated group and the control group is significantly different by the policy shock, which suggests that the policy shock has a certain degree of continuity (or a time lag), and the effect of the role of the effect has a tendency to increase. In addition, we also note that there is an upward trend in the last year’s data, and the difference between the treated group and the control group’s urban policy shocks is narrowing, which may come from the filling of missing data on the one hand, and the weakening of the policy effect on the other hand.

Fig. 2

Dynamic effects test

The parallel trend and dynamic effects tests described above suggest that the treated and control groups can use DID to discuss the impact of this policy shock on industrial structure rationalization. We use the benchmark data to conduct regression analyses for the DID models without control variables, i.e., Eqs. (3) and (4), and for the DID models with control variables, i.e., Eqs. (5) and (6). Table 3 summarizes the obtained regression results, where columns (1) and (2) are the effects of the policy shock on industrial structure rationalization obtained from regressions based on the models of Eqs. (3) and (5), while columns (3) and (4) are the effects of the policy shock on industrial structure rationalization obtained from regressions based on the models of Eqs. (4) and (6). Specifically, in terms of the effect of the policy shock (i.e., $\mathrm{src \times time}$), the policy shock has a significant effect on the rationalization of industrial structure ($\textrm{lnTL}$) in NIC cities, while it does not have a significant effect on the advancedization of industrial structure ($\textrm{lnIS}$). In the cross term between column (1) and row (1), the coefficient of term $\mathrm{src \times time}$ is $-1.013$ and is significant at $1\%$ level, which indicates that the policy shock ($\mathrm{src \times time}$) has significant negative correlation on $\textrm{lnTL}$, reflecting that the policy shock can significantly promote the rationalization of the industrial structure of NIC cities. After adding control variables and controlling the time effect and individual city effect, the sign of the coefficient of term $\mathrm{src \times time}$ in column (2) remains unchanged, and the two still show significant negative correlation, indicating that the impact of this policy shock on industrial structure rationalization is robust. In terms of the effect of the policy shock on industrial structure advancedization (the cross terms between column (3) and column (4) and row (1)), the correlation between term $\mathrm{src \times time}$ and term $\textrm{lnIS}$ is insignificant, that is, the effect of the policy shock on industrial structure advancedization of the cities in the NIC cities is insignificant in both the cases without and with the control variables. On one hand, the reason may be that the industries in Chinese cities are mainly concentrated in the secondary industry, while the overall development level of the tertiary industry is relatively low. Under the role of the inertia of the original industrial development, it is difficult for the NIC cities to change the ratio of the secondary and tertiary industries with a certain policy in a short period of time. On the other hand, the role of the new national system of science and technology policy is mainly to encourage the manufacturing industry to tackle the technology that hinders the development of industry, the direct effect on the development of the secondary industry is stronger, while the role of the tertiary industry is difficult to show in the short term.

Table 3 DID results with benchmark data

From the regression results of the model including control variables, the role of urban financial level, per capita income and government financial expenditure on the rationalization and advancedization of industrial structure is not the same. Specifically, in column (2), the financial development ($\textrm{lnfd}$) is not significant, that is, the role of financing on the rationalization of industrial structure is not significant, the reason may be related to the direction of the use of financing and the characteristics of the capital-intensive nature of the manufacturing innovation industry, as the financing may be used first to update the equipment and the introduction of new technologies, so that the industry in a certain period of time in a certain type of resources to increase rapidly, and the other resources of innovation because of the different adjustment speeds, leading to a decline in the matching of resources within the industry, and to a certain extent reducing the role of financing in the rationalization of industrial structure. The coefficients of per capita income ($\textrm{lnpergdp}$) and government financial expenditure ($\textrm{lngov}$) are both negative and these variables significantly correlated with the rationalization of industrial structure, indicating that the increase of per capita income and government financial expenditure has a positive role in promoting the rationalization of industrial structure, reflecting that with the improvement of per capita income in the city, the city’s attraction to innovative talents has been strengthened, which fits the talent demand of the innovative industry, while the expansion of government financial expenditure accelerates the construction of urban infrastructure and innovation supporting industries, which is conducive to the deployment of resources and access to services by innovation industries. With the role of innovation talent gathering and resource deployment efficiency enhancement, the city’s innovation resources are expanded and optimized, and thus the rationalization level of the city’s industrial structure is promoted. In Column (4), the coefficient of the level of financial development ($\textrm{lnfd}$) is positive and the variable significantly correlated with the advanced industrial structure, indicating a positive correlation between financing and the advancedization of industrial structure ($\textrm{lnIS}$), reflecting that the enhancement of the city’s level of finance promotes the development of the financial and innovation service industry, promotes the development of the city’s tertiary industry, and facilitates the advanced industrial structure of the city. The coefficient of government financial expenditure ($\textrm{lngov}$) is negative and the variable significantly correlated with the industrial structure advancedization, indicating a negative correlation between government expenditure and industrial structure advancedization ($\textrm{lnIS}$). Under the policy of the new national system of science and technology, the government financial expenditure is more invested in the secondary industry, which is dominated by the new manufacturing industry and high-tech enterprises, and less in the tertiary industry, which leads to the increase of the proportion of the secondary industry in the urban industry, while the proportion of the tertiary industry declines in relative terms, and then reduces the level of the advanced industrialization of the urban industry. It is noted that the relationship between urban per capita income ($\textrm{lnpergdp}$) and industrial structure advancedization ($\textrm{lnIS}$) is not significant, which may be due to the fact that the industrial structure of the national innovative cities has gradually shifted to capital-intensive enterprises, which have gathered a large number of skilled talents, and such talents need innovative service industries that are compatible with the innovative industries, whereas the service industries in the cities have not completed the transformation to innovative service industries. Therefore, the mismatch between the degree of development of the urban service industry and the demand for innovative talents has affected the role of talent aggregation in promoting the service industry.

Table 4 Parallel trend hypothesis test

Table 5 Co-support hypothesis testing

4.3 Analyzing the shock effects of the policy based on matching data

Due to the differences in innovation resource endowment between NIC cities and general cities, directly using the benchmark data regression of these two types of cities may have a sample selectivity bias, in order to improve the robustness of the results, we construct matching data according to the method mentioned in Sect. 3, by selecting or simulating the matching data from the control group that is similar to the treated group. First, the unmatched cities and observations are excluded, at the same time, the observations with unmatched years in the matched cities are excluded. The 343 cities in the benchmark data are screened for matching, and 161 observations from 54 cities are retained, including 9 treated group cities totaling 67 observations and 45 control group cities totaling 94 observations. Second, the observations that do not satisfy the common support are eliminated, which means that only the values taken in the overlapping area of the propensity scores of the treated group and the control group are retained, and on the basis of the 343 cities in the benchmark data, 1237 observations from 243 cities are retained, of which 9 cities in the treated group total 70 observations and 234 cities in the control group total 1,167 observations. For ease of presentation, we refer to the first type of matching samples as matched city samples and the second type of matched samples as co-supported samples into the DID analysis in the following, and the above samples are summarized in Table 13 in appendix A.

The samples before and after matching are tested for parallel trends and common support, and we summarize the results of these tests in Tables 4 and 5. As can be seen in Table 4, before matching, there is a significant difference between the treated and control group cities, while after matching, there is no significant difference between the treated and control groups. For example, the level of financial development ($\textrm{lnfd}$) before matching, the t-value and p-value of the t-test are 13.63 and 0.000, respectively, indicating that there is a significant difference between the treated group and the control group in the characteristic of the level of financial development, while after matching, the t-value and p-value are $-0.81$ and 0.416, respectively, indicating that the treated group and the control group no longer have a significant difference in the characteristic of the level of financial development. Similar results can be obtained for the other two characterization variables, per capita income ($\textrm{lnpergdp}$) and government financial expenditure ($\textrm{lngov}$). In Table 5, from the common support hypothesis test, the treated and control group cities as a whole show a significant difference before matching (p-value of 0.000), while this difference is no longer significant after matching (p-value of 0.490).

According to the results of the parallel trend assumption and common support assumption test, the matching data satisfy the similar sample requirement of the quasi-natural experiment and are capable of PSM-DID analysis. Therefore, we regress the DID model with control variables, i.e., Eqs. (5) and (6), respectively, based on the two types of matching data, taking into account both the time effect and the city effect, and the results are summarized in Table 6. Columns (1) and (2) are the effects of the policy shock on industrial structure rationalization obtained from regressions based on Eq. (5), while columns (3) and (4) are the effects of the policy shock on industrial structure advancedization obtained from regressions based on Eq. (6), controlling for time effects and city effects for all models. Specifically, in terms of the coefficient of the policy shock effect ($\mathrm{src \times time}$), in the two types of matched samples, the effect of this policy shock is similar to that of the benchmark data, which is still significant for the industrial structure rationalization of the cities in the NIC cities ($\textrm{lnTL}$), and still insignificant for the industrial structure advancedization ($\textrm{lnIS}$). The coefficients obtained from the two types of matching data on the cross terms of the rows where columns (1) and (2) and term ($\mathrm{src \times time}$) are located are − 0.967 and − 0.697, and are significant at the 1% level, indicating that the policy shock term ($\mathrm{src \times time}$) has a significant negative correlation on the industrial rationalization of $\textrm{lnTL}$, which reflects that the policy shock can significantly promote the rationalization of industrial structure of NIC cities, and that the government can play an important role in the process of rationalization of industrial structure by the policy shock. In terms of the role of the policy shock on the industrial structure advancedization (the cross terms between columns (3) and (4) and the rows where term ($\mathrm{src \times time}$) is located), the correlation between term ($\mathrm{src \times time}$) and the industrial structure advancement $\textrm{lnIS}$ is not significant, that is, the role of this policy shock on the industrial structure advancedization of the NIC cities is not significant under the two types of matching data, and the role of the government with a presence is not significant under the process of the policy shock on the industrial structure advancedization of the NIC cities. The role of the government in this policy shock on the process of industrial structure advancedization is limited.

Table 6 Results of PSM-DID

The combined results under the benchmark data and matching data show that some of the results in hypothesis 1 are verified, that is, the policy shock of the new national system of science and technology can significantly promote the industrial structure rationalization of the NIC city, but the effect on the industrial structure advancedization of the city is not significant. Therefore, in the process of promoting the new national system of science and technology to overcome the core technology of industry, the NIC city needs to pay attention to the impact of the policy shock on the development of the city’s original industry. In the case of limited resources, the active government needs to consider the current development path of urban industries in the process of promoting the new nationalization system policy. For the NIC city which is developing with secondary industry as the leading industry, promoting this policy can accelerate the concentration of the city’s industrial resources in the secondary industry, which is conducive to enhancing the rationalization level of the city’s industrial resource allocation, and the positive effect of the government with a presence on the local industry and economy will be more obvious. For NIC cities developing towards service-oriented industries, under the constraint of limited resources, the government’s promotion of the policy may take up resources originally used for industrial upgrading, and may interfere with or even hinder the original industrial development path.

4.4 Extension discussions

4.4.1 Effects of policy shocks under subregional conditions

In both the benchmark and matching data DID models, we assume that the effect of this policy shock is indifferent across regions of the country. However, the order of China’s economic development is along the eastern coastal region, the middle and lower reaches of the Yangtze River and the western region gradually, the industry and market maturity is highest in the eastern region, followed by the central region, and the lowest in the western region, but the difference between the central and western regions is relatively small compared to the eastern region, so there is a more obvious difference between the eastern NIC cities and the central-western NIC cities in the level of industrial development and marketization, and there are also different needs for governmental allocation and market allocation in resource allocation and industrial collaboration between the two types of regions. Eastern NIC cities, such as Beijing and Shanghai, have a high degree of marketization, a good industrial base and technological innovation level, a strong industrial base, and a more mature development of high-tech industries, so that the allocation of industrial resources and collaboration among enterprises are more based on market regulation and less dependent on policies, while Central-western NIC cities, such as Wuhan and Xi’an, are less market-oriented, have relatively weak industrial foundations, and are not closely connected to each other, so that they need government policy support in attracting resources and building up industrial agglomeration. In order to accommodate this more realistic situation, where differences in industrial development and marketization within different regions result in different policy needs for cities in the industrial upgrading process, we develop a policy shock model that includes regional effects, which is used to discuss the specific effects of this policy shock in the Eastern and Midwestern regions.

In this subsection, for ease of interpretation, we rename the $\mathrm{src \times time}$ term reflecting the policy shock as did and divide the treated group of cities into Central-western and Eastern cities in the same way as in Sect. 4.1, and establish the Central-western and Eastern city dummy variables w and e to be added to the matching data. When the city is a Central-western city, w takes 1 and the rest take 0; when the city is an Eastern city, e takes 1 and the rest take 0. Using the cross-multiplication terms of the two types of dummy variables and the policy shock variable did, we construct the policy shock term of the Central-western cities ($\mathrm{w \times did}$, shortened to wdid) and the policy shock term of the eastern cities ($\mathrm{e\times did}$, shortened to edid), which are used to reflect the policy shocks of the Central-western cities and the Eastern cities, respectively. Two types of city policy shock models are established, and the results are interpreted with the policy shock model for the Central-western cities and the robustness of the conclusions is verified with the policy shock model for the Eastern cities. Where the policy shock model for the Central-western cities is,

$$\begin{aligned} \textrm{TL}_{it}= & {} \alpha _{0} +\alpha _{1}\textrm{src}_{i}+\alpha _{2}\textrm{time}_{it} +\alpha _{3}\textrm{did}_{it} \nonumber \\{} & {} +\alpha _{4}\textrm{wdid}_{it} +\sum _{j=1}^{3}d_{j}x_{jt}+\varepsilon _{it}, \end{aligned}$$

(7)

$$\begin{aligned} \textrm{IS}_{it}= & {} \beta _{0} +\beta _{1}\textrm{src}_{i} +\beta _{2}\textrm{time}_{it}+\beta _{3}\textrm{did}_{it}\nonumber \\{} & {} +\beta _{4}\textrm{wdid}_{it}+\sum _{j=1}^{3}d_{j}x_{jt} +\delta _{it}. \end{aligned}$$

(8)

The model is used to test the effect of policy shocks in Central-western cities, and the results obtained are summarized in Table 7, where columns (1) and (2) are the results of the policy shocks on the rationalization of the industrial structure based on the regression of equation (7), and columns (3) and (4) are the results of the policy shock on industrial structure advancedization based on the regression of equation (8), where the policy shock is decomposed into two items did and wdid. From the results, the policy shock has a significant effect on the rationalization of urban industrial structure, and the policy shock has a stronger effect on the rationalization of urban industrial structure in Central-western cities relative to Eastern cities, while the policy shock has a non-significant effect on the industrial structure advancedization all regions of China. Compared with the model results in Eqs. (5) and (6), in columns (1) and (2), the policy shock term wdid has a significant negative correlation with industrial structure rationalization term (lnTL) in the Central-western regions, indicating that the policy shock has a stronger role in promoting the industrial structure rationalization of the Central-western cities of the NIC and the policy works better in the Central-western regions. Notice that the did term is not significant, implicitly suggesting that the policy shock may have a limited effect on the rationalization of industrial structure in Eastern cities. And in columns (3) and (4), the Central-western policy shock term wdid is not significantly correlated with the industrial structure advancedization term (lnIS), reflecting that the policy shock has no effect on the industrial structure advancedization in the cities of NIC in different regions. The reason for the above results may be that the degree of marketization is relatively low in the Central-western cities, the market’s effect on industrial resource allocation is weaker than the government’s allocation, and the government’s allocation through policy is more efficient in industrial structure rationalization. Therefore, in the process of industrial development of NIC cities in Central-western China, the market should not completely allocate resources, and the government should actively play a positive role in mobilizing resources to promote the development of innovative industries.

Table 7 PSM-DID results for the Central and Western region

A similar approach is used to analyze the effects of policy shocks in Eastern cities, where the wdid term in Eqs. (7) and (8) is replaced with the eastern policy shock term, edid, to verify the robustness of the results of the Central-Western policy shock model. Table 14 in appendix summarizes the regression results of the eastern policy shock model, indicating that the policy shock has a significant effect on the rationalization of industrial structure, but the direction of the effect is affected by the region where the NIC city is located. Relative to Central-western cities, Eastern cities have an inhibitory effect on the rationalization of the city’s industrial structure after the policy shock, while Central-western cities have a promotional effect on the rationalization of industrial structure under the influence of the policy shock. Similarly, the policy shock has no significant effect on the advancedization of industrial structure. Specifically, the policy shocks of the model are decomposed into two terms, $\textrm{did}$ and $\textrm{edid}$, with the $\textrm{did}$ term reflecting the effect of Central-western cities on industrial structure under this policy shock, while the $\textrm{edid}$ term reflects the difference between the policy effects of Eastern cities and Central-western cities under this policy shock. The $\textrm{edid}$ term in Column (1) and Column (2) has a significant positive correlation with the industrial structure rationalization $\textrm{lnTL}$, indicating that relative to the central-western cities, the Eastern cities will reduce the rationalization of industrial structure under this policy shock, while the $\textrm{did}$ term has a significant negative correlation with the industrial structure rationalization ($\textrm{lnTL}$), indicating that the Central-western cities’ rationalization of industrial structure can be be improved. The reason for this result may be related to the degree of marketization in different regions, with the higher level of marketization in the Eastern cities, where industrial resources are more predominantly allocated by the market. The policy impact of the new national system of science and technology has promoted the upgrading of specific industries in Eastern cities, and although it can promote the agglomeration of high-tech industrial resources and industrial transformation, it has interfered with the original market resource allocation process and reduced the allocation efficiency of industrial resources in Eastern cities. Therefore, in the process of promoting industrial development by policies in Eastern cities, the government should pay attention to the impact of resource allocation by the government on market allocation, avoid the conflict between the government and the market in resource allocation, and play a positive role in the development of industries by policies. In terms of the role of the policy on the advancedization of industrial structure, the results in Column (3) and Column (4) show that the policy shock has no significant effect on the advancedization of industrial structure and there is no regional difference.

From the above analysis, it can be seen that some of the conclusions of hypothesis 2 are verified, that is, in the different regions of China, there are differences in the effect of the role of this policy shock on the industrial structure, and the policy shock of the Central-western inland NIC cities has a stronger effect on the rationalization of the industrial structure, but in the advancedization of industrial structure, this policy shock does not have a significant role in all the places. Therefore, the central government should consider the regional differences of the policy shock in the process of promoting the new national lifting system policy for science and technology. For the lower marketization degree of the Central-western NIC cities, the government allocation of industrial resources is more effective than the market allocation, and the strong government is conducive to promoting the policy implementation and accelerating the deployment of industrial resources. The higher marketization level of Eastern cities, the market allocation of industrial resources is more effective than government allocation, strong government allocation of resources interferes with the market resource allocation mechanism, and the vigorous promotion of policies will have an inhibitory effect on the development of industry, and it is necessary to explore the effective scope of the government and the market in resource allocation, and to reduce the negative impact of the allocation of resources by the government of the Eastern cities on the development of the industry.

4.4.2 The moderating role of urban finance, wealth and government inputs in policy shocks

In the DID model, we assume that the relationship between policy shocks and urban endowments (such as human, financial, and material resources) of NIC cities is independent of each other, and do not consider the interaction between urban endowments and policy shocks. In cities with a concentration of high-level talents, abundant capital and well-developed industrial support, the promotion of science and technology policies can better utilize urban resources, and the effective use of urban resources can better promote the practice of policies, so there may be an interaction effect between urban endowments and policy shocks. Based on the above considerations, we develop a model for the interaction between urban resources and policy shocks to discuss the role of urban resources in policy shocks. Similar to the regional discussion in the previous section, the policy shock term is named did, and the control variables are used to create the cross-multiplier terms of urban financing, wealth, and government inputs with the policy shock term to analyze the effect of the role of urban endowment in policy shocks,

$$\begin{aligned} \textrm{TL}_{it}= & {} \alpha _{0} +\alpha _{1}\textrm{src}_{i}+\alpha _{2}\textrm{time}_{it} +\alpha _{3}\textrm{did}_{it} \nonumber \\{} & {} +\sum _{j=1}^{3}d_{j}x_{jt} + \sum _{j=1}^{3}f_{j}\textrm{did}_{it}x_{jt}+\varepsilon _{it}, \end{aligned}$$

(9)

$$\begin{aligned} \textrm{IS}_{it}= & {} \beta _{0} +\beta _{1}\textrm{src}_{i} +\beta _{2}\textrm{time}_{it}+\beta _{3}\textrm{did}_{it} \nonumber \\{} & {} +\sum _{j=1}^{3}d_{j}x_{jt} + \sum _{j=1}^{3}f_{j}\textrm{did}_{it}x_{jt} +\delta _{it}. \end{aligned}$$

(10)

Based on the matching data, regressions are performed for equations (9) and (10) and the results obtained are summarized in Table 8. In terms of the role of urban resource endowments, urban endowments have a significant moderating role in the process of policy shocks on industrial rationalization. Specifically, on the cross term between column (2) and the $\mathrm{did \times lnpergdp}$ row, the $\mathrm{did \times lnpergdp}$ term is positively correlated with industrial structure rationalization ($\textrm{lnTL}$) at the 10% significant level, indicating that personal income ($\textrm{lnpergdp}$) attenuates the effect of this current policy shock, which means that the effect of this policy shock on the rationalization of industrial structure is weakened as personal income rises. The reason for this result may be that the policy shock of the new national system of science and technology promotes the city’s resource investment in the high-tech industry, crowding out the development resources of other industries, and in the case that the number of high-tech talents is difficult to grow rapidly, the scarcity of high-tech industry workers causes the rise of wages in this industry. At the same time, the rise in wages of workers in high-tech industry guides workers in other industries to move to high-tech industry, causing the scarcity of human resources in other industries, which ultimately pushes up the average wages of the whole region. Although the scarcity of urban human resources can be alleviated by attracting talents from other regions, the situation of scarcity of industrial human resources is difficult to be solved quickly under the stronger policy impact, which is not conducive to the rationalization of industrial structure. Therefore, in promoting the policy, the government needs to weigh individual income against the level of input of specific industrial resources to avoid the situation of industrial resource tension caused by the promotion of the policy. On the cross terms of column (1), column (2) and $\mathrm{did \times lnfd}$ rows, the $\mathrm{did \times lnfd}$ term has a significant negative correlation with industrial structure rationalization ($\textrm{lnTL}$). It indicates that the level of financial development ($\textrm{lnfd}$) has the effect of strengthening the effect of this policy shock, in other words, with the improvement of the financing environment, the effect of this policy shock on the rationalization of industrial structure is strengthened. Therefore, the policy shock of the new national system of science and technology needs the cooperation of an efficient financing environment to satisfy the demand for funds by the innovation main body and accelerate the industrial technological upgrading and industrial structure adjustment. On the cross terms of column (1) and column (2) with $\mathrm{did \times lngov}$ rows, the $\mathrm{did \times lngov}$ term does not have a significant correlation with industrial structure rationalization ($\textrm{lnTL}$), which may be related to the current situation of China’s urbanization development, where the central cities in each region have invested and constructed a large amount of infrastructure and industrial supporting facilities, and homogenized urban infrastructures have weakened the effect of the government’s investment on science and technology innovation industry. In the discussion of Column (3) and Column (4) on industrial advancedization, the three types of cross-multiplier terms (namely, the $\mathrm{did \times lnpergdp}$ term, the $\mathrm{did \times lnfd}$ term, and the $\mathrm{did \times lngov}$ term) are insignificant, suggesting that urban endowment does not have any significant moderating effect in the process of policy shocks on industrial advancedization. From the above analysis, it can be seen that urban endowment is a regulator affecting the effect of the policy in the process of promoting the new lifting system policy of science and technology, and the comprehensive effect of different roles of endowment in the process of policy shock needs to be further explored to decide the resource input in the process of policy promotion.

Table 8 Results of moderation effects model

5 Conclusion

We explore the mechanism of the role of science and technology policy shocks on the industrial structure adjustment of NIC cities, and take the policy of the new national system of science and technology as an example to establish the hypothesis of the effect of the policy shocks on the role of the NIC cities, collate a total of 14 years of economic panel data of the cities above the prefectural level, construct the benchmark data and the matching data for the empirical analysis, and the DID model is used to discuss the effect of this policy shock on the industrial structure of the NIC cities. In addition, the influence of geographical and urban endowment on this policy shock is discussed from the perspectives of regional differences and urban endowment. The results of the study show that: (1) the policy shock of the new national lifting system of science and technology has a significant promotion effect on the rationalization of the industrial structure of the NIC cities, but the effect on the advancedization of the industrial structure of this kind of cities is not significant. This result shows that the policy shock of the new national system of science and technology has promoted the industrial transfer and related industry synergy in the NIC cities, improved the matching degree between innovation resources and industries in the cities, and promoted the rationalization of the industrial structure of these cities. However, the large amount of resources invested in the secondary industry has resulted in the relative increase in the ratio of the secondary industry and the relative decrease in the ratio of the tertiary industry in the cities, which has made the policy shock not significant for the industrial advancement. (2) The policy impact of the new national system of science and technology on the adjustment of industrial structure shows more obvious regional differences, and has a greater effect on the rationalization of industries in the Central-western NICs than in the Eastern NICs. It reflects that in the case of regional marketization differences, the policy shock has a stronger policy guidance effect on the Central-western NIC cities with lower marketization level, and the government allocation of resources may be more efficient, whereas in the Eastern cities, with better industrial foundation and market-oriented resource allocation, the policy intervention may interfere with the market resource allocation efficiency and reduce the level of rationalization of the city’s industrial structure. (3) In terms of the role of urban endowment in shocks, the level of financial development of NIC cities has a positive moderating effect in the process of policy shocks on industrial structure rationalization, while per capita income has a negative moderating effect on the effect of this policy. For the NIC city, the improvement of financial level reduces the cost of innovation cooperation, creates conditions and support for enterprises, scientific research institutions and other micro-innovation main bodies to unite, and innovation cooperation promotes the efficiency of industrial division of labor and optimal allocation of resources, thus promoting the rationalization of the city’s industrial structure. Although the city’s personal wealth can attract more talents and improve the knowledge and technology structure of the workforce in the region, there may be a mismatch between human resources and industrial resources in the short term, which reduces the rationalization level of the city’s industrial structure.

Based on the above conclusions, in order to promote the positive effects of the policy shock of the new national system of science and technology on the industrial upgrading of NIC cities, the government can consider the following three aspects:

First, NIC cities should seize the opportunity of the national implementation of the new national system of science and technology, play the role of the government’s resource guidance and the market’s resource allocation, and promote the transformation and upgrading of the local industrial structure. On one hand, NIC cities should actively undertake national key technology research tasks based on innovation resource endowment, promote the development of local advantageous industries and emerging industries, and play a guiding role in policy for industrial transformation. On the other hand, while playing the role of government allocation, it should push forward the local market-oriented reform, expand the openness of the NIC cities to the outside world, reduce the technological barriers between the regions, and promote regional scientific and technological cooperation in key technology research, accelerate the industrialization of technologies, and promote the transformation and upgrading of local industrial structure.

Second, NIC cities should decide on the main ways of allocating industrial resources based on differences in geographic economic development. As the impact effect of the new national system of science and technology policy is related to the level of economic development in different regions, for the cities in Central-western of China, the government’s role in resource allocation should be strengthened, the policy support for science and technology parks and high-tech industrial parks in Central-western cities should be strengthened to accelerate the flow of innovation resources and to promote the transfer of sunset industries, to introduce and cultivate the emerging and high-technology industries, and to enhance the development of related industries, at the same time, it is important to note that the market’s role in the allocation of resources should be actively utilized. For Eastern cities, the role of the market in the allocation of innovation resources should be strengthened to promote the participation of innovation subjects in key technological breakthroughs, and to promote the optimization of local innovation resources and industrial structure, while the government’s role in the allocation of resources should be indirectly guided.

Third, in promoting science and technology policies, the government should pay attention to the specific effects of urban endowments ( such as human resources and other endowments) in the process of policy shocks. For example, in terms of urban human resources, the government should weigh the effect of innovation resource inputs on the matching of average wages and industrial resources in the city, so as to avoid the problem of increasing the scarcity and decreasing the matching degree of human resources in urban industries caused by resource inputs. In terms of urban financing capacity, the government should promote urban financial reform and cross-regional capital financing, so as to strengthen the positive role of the financing environment in the impact of the new national system of science and technology policies. Finally, since the policy shock of the new national system of science and technology has slowed down the city’s investment in the tertiary industry, which in the long run will result in mismatch between industries, the government should guide the development of innovative service industry as much as possible in the process of implementing the policy, give full play to the positive role of the service industry in scientific and technological innovation, and push forward the synergistic development of the advanced manufacturing industry and the productive service industry, so as to provide a further upgrading for the industries of the NIC city.

Table 9 The industrial rationalization process of the DID model

Table 10 The industrial advancedization process of the DID model

Table 11 List of regions qualified for national innovative city construction before 2012

Table 12 National Science and Innovation Center and National Science Center

Table 13 Two types of matched data

Table 14 PSM-DID results for the Eastern Region

Appendix A: Related tables

See Tables 9, 10, 11, 12, 13 and 14.