How does industry-university-research collaborative innovation affect energy intensity in China: a novel explanation based on political turnover

Abstract

As a form of integrating knowledge resources and promoting technological innovation, industry-university-research (IUR) collaborative innovation is thought to influence energy intensity reduction. However, the boundary conditions of IUR collaborative innovation affecting energy intensity have yet to be discussed. To fill this gap, we explore the impact of IUR collaborative innovation on energy intensity and the role of political turnover in its influencing mechanism, using a panel dataset of 30 Chinese provinces from 2010 to 2018. IUR collaborative innovation inhibits energy intensity, but this effect is only significant in the eastern region. Interestingly, political turnover positively moderates the inhibitory effect of IUR collaborative innovation on energy intensity. However, this moderating effect is only significant in the central and western regions. Moreover, the robustness tests provide abundant evidence to support the above findings. Finally, some policy implications are suggested.

Introduction

As the largest carbon emitter, China is critical in reducing global carbon emissions and mitigating climate change. The Statistical Review of World Energy 2022 published by British Petroleum reveals that China’s carbon dioxide (CO₂) emissions reached 10,523 million tonnes in 2021, with a per annum growth rate of 5.8%, accounting for 31.1% of the global CO₂ emissions. As a core member of the Paris Agreement, China has announced a 60% to 65% reduction in CO₂ emissions per unit of GDP of 2005, and a peak in CO₂ emissions by around 2030. Energy intensity is the amount of energy consumed per unit of output value, and is considered a significant indicator characterizing energy use efficiency (Shen and Lin, 2020). Reducing energy intensity has been proven an effective method to reduce carbon emissions while maintaining economic expansion, and all countries have been appealed to reduce energy intensity to address climate change (Hille and Lambernd, 2020).

In response to climate change, the drivers of energy intensity have been widely discussed in the existing literature, such as economic growth (Gardner and Elkhafif, 1998), energy price (Cornillie and Fankhauser, 2004), trade liberalization (Rafiq et al., 2016), and government policy (Huang et al., 2023). Technological innovation can effectively improve energy efficiency, having a far higher impact on energy intensity reduction than other influencing factors (Mughal et al., 2022). However, Negassi (2004) found that technological innovation requires substantial research and development (R&D) capacity. Even when a firm has external intellectual assets, weak internal R&D skills are an obstacle to the firm’s innovative activities. Compared with developed countries, many firms in developing countries do not have independent R&D competencies (Motohashi and Yun, 2007). With restricted specialized knowledge and resources, enterprises increasingly struggle to explore novel technologies with complete autonomy. In response to the innovation dilemma, an innovation strategy that relies significantly on cooperation is well accepted in developing countries, which can supplement the limited R&D abilities of firms (Kafouros et al., 2015). For instance, under China’s innovation-driven development strategy, the collaborative innovation policy system is gradually improving through multiple legislation, rules, and outlines (Guan and Zhao, 2013).

Industry-university-research (IUR) collaborative innovation originates from synergies. It describes collaborative R&D technology activities by the industry, university, and research institutes with quality resources and extraordinary abilities (Bonaccorsi and Piccaluga, 1994). Although companies are the primary agents of technological innovation and have a significant role in transforming knowledge into products, the source of knowledge for technological innovation of enterprises depends on universities and research institutes (Bai et al., 2020). One of the features ascribed to universities and scientific institutions is the transmission of knowledge to enterprises (Eom and Lee, 2010). Therefore, IUR collaborative innovation is conducive to integrating three innovative agents. This helps create something creative and practical for addressing market failures in emerging countries such as China (Shi et al., 2020). China initiated IUR collaboration in the 1980s and encouraged IUR collaborative innovation through various policies such as establishing business-university alliances, inviting companies to reside in university laboratories, and building industrial parks. However, the existing literature still needs to analyze whether IUR collaborative innovation contributes to energy intensity reduction in China.

In the mechanism by which IUR collaborative innovation affects China’s energy intensity, the “visible hand” of the government remains crucial in allocating resources (Shen and Lin, 2020). The gradual decentralization of China’s administrative system has given the local government flexible discretionary powers in policy establishment and developing local economies (Li and Zhou, 2005). Evidence has shown that the behavior of local officials affects energy development strategies, regional innovation capacity building, and environmental protection (Kong et al., 2021; Tian et al., 2023). Under China’s official promotion tournament system, local officials have incentives to intervene in business decisions to achieve good performance (Cao et al., 2022). As the primary designers and implementers of environmental and economic policies, officials’ turnover can break firms’ original political resources (Pertuze et al., 2019). Therefore, political turnover, defined as personnel changes in top leadership positions, may lead to policy discontinuity and affect the impact of IUR collaborative innovation on energy intensity (Pertuze et al., 2019; Shen and Lin, 2020). Despite the established literature affirming the facilitating role of political turnover in technological innovation, whether political turnover also has a positive role in IUR collaborative innovation lacks empirical evidence.

Therefore, the goal of our study is to solve two problems. (1) How does IUR collaborative innovation affect energy intensity? (2) Does political turnover have a moderating role in the impact of IUR collaborative innovation on energy intensity? Against China’s peak carbon and carbon neutrality targets, our work enriches the existing explanation about the driving factors of energy intensity. The research findings will help policymakers to propose practical, targeted, and feasible solutions from the perspective of IUR collaborative innovation.

Literature review

Although studies on the factors influencing energy intensity are relatively abundant, few have analyzed the impact of IUR collaborative innovation and political turnover on energy intensity. The literature relevant to our research can be grouped into two categories: how IUR collaborative innovation affects energy intensity reduction; and the effect of political turnover on technological innovation. These two lines of research can help us understand the relationship between political turnover, IUR collaborative innovation, and energy intensity.

With carbon emissions continuing to rise, the role of technological innovation in energy intensity decline is increasingly being emphasized. Some studies concluded that higher levels of technological innovation are conducive to reducing energy intensity, such as Zhang et al., (2020) and Liu et al., (2022). Technological progress can lower the volume of energy consumption required per unit of GDP output (Santiago et al., 2020). Thus, it has been recognized as a critical force contributing to energy intensity reduction (Liu et al., 2022). Differently, Li et al., (2019) argued that technological innovation does not affect energy intensity directly. Overall, the role of technological innovation in reducing energy intensity has been recognized in existing research.

As a more complex innovation model between closed and open innovation, collaborative innovation can enhance innovation capacity with limited resource allocation and has advantages in promoting technological innovation (Shi et al., 2020). Nevertheless, more focus should be placed on the impact of collaborative innovation on energy intensity and its impact mechanisms. In addition, regarding measuring IUR collaborative innovation, the existing studies often used patents, publications, and new processes as the proxy variable for innovation output (Szücs, 2018; Chen et al., 2019). However, using these indicators to measure IUR collaborative innovation has two limitations. IUR collaborative innovation is a process from input to output, whereas patents, publications, and new processes only reflect the output of IUR collaborative innovation (Shin et al., 2019). The other is that the number of those indicators cannot remember the contributions of industries, universities, and research institutes in the cooperation (Guan and Zhao, 2013).

As part of the macro-environment, how government factors influence technological innovation has received attention. For example, Shen and Lin (2020) analyzed the role of policy preferences in R&D investment from the perspective of political incentives. Hille and Lambernd (2020) found that environmental regulation policy is associated with innovation in renewable energy technologies. Nevertheless, most available literature analyzed the effect of government factors on technological innovation from a policy or whole-of-government perspective. Few studies have addressed how local government officials themselves influence technical innovation. Existing discussions on the effects of political turnover have focused on environmental pollution (Deng et al., 2019), energy conservation decisions (Kong et al., 2021), and public services (Akhtari et al., 2022). With the increasing severity of climate change, Cao et al., (2022) innovatively analyzed how the local official promotion affects technological innovation in renewable energy. However, the available research does not cover whether political turnover affects IUR collaborative innovation affecting energy intensity.

Research hypotheses

The impact of IUR collaborative innovation on energy intensity

As a form of integrating knowledge resources and promoting technological innovation, the available literature shows that IUR collaborative innovation can influence energy intensity reduction in two ways. First, IUR collaborative innovation can drive technological progress, reducing energy intensity (Song et al., 2020). Evidence has shown that IUR collaborative innovation can promote technological advances; with the promotion of IUR collaborative innovation, the level of technological innovation will be enhanced. For example, Marques et al. (2006) found that information about product utilization spreads more quickly in a country where collaborative innovation is encouraged. Xu et al., (2018) found that IUR collaborative innovation promotes the commercialization of scientific achievements. Moreover, scholars have identified that technological innovation is a crucial path for energy intensity reduction, such as Cagno et al., (2015), Hille and Lambernd (2020), and Santiago et al., (2020). Second, evidence has shown that a stable cooperative relationship of IUR collaborative innovation helps decrease transaction costs and makes innovation activities more efficient, leading to more output from a given energy input (Tang and Tan, 2013). Sohag et al., (2015) also suggested that IUR collaborative innovation can improve energy efficiency by solving technical, financial, and human resource problems, reducing energy consumption at a particular output level. In summary, we consider that collaborative innovation can reduce energy intensity. Thus, the following research hypothesis is proposed:

Hypothesis 1: IUR collaborative innovation can reduce energy intensity.

The role of political turnover in IUR collaborative innovation affecting energy intensity

Evidence shows that political turnover may strengthen IUR collaborative innovation in two ways. One way is that political turnover changes local government behavior and the rearrangement of resources. Since the Chinese government has promulgated or revised many regulations on emission reduction, the newly appointed officials’ energy-saving performance predominantly affects their promotion (Deng et al., 2019). Thus, political turnover may strengthen the internal incentives for officials to invest in environmental technological innovation. To achieve distinguished performance within their tenure, newly appointed officials may focus their scarce resources on technological innovation to promote energy efficiency (Aalbers et al., 2013). For example, Shi et al., (2020) found that political turnover can promote corporate innovation by increasing corporate innovation subsidies. Moreover, Shen and Lin (2020) found that policy incentives provide an incentive to invest in R&D. Another way is that political turnover provides a sensitive period during which collusion between business and government may be reduced, thereby eliminating the adverse effects of corruption (Pertuze et al., 2019). Sidorkin and Vorobyev (2018) found that the more the incumbents are sure they will not be reappointed, the weaker the motivation for corruption. Eliminating corruption is conducive to enhancing the efficiency of public resource utilization and rationalizing investment structure, thereby establishing a reasonable link between political turnover and IUR collaborative innovation (Jain, 2001; Deng et al., 2019). In summary, political turnover guarantees IUR collaborative innovation, directly affecting energy intensity. Therefore, we assume that political turnover potentially moderates the impact of IUR collaborative innovation on energy intensity. The following research hypothesis is proposed:

Hypothesis 2: Political turnover moderates the impact of IUR collaborative innovation on energy intensity.

According to the theoretical analysis and research hypotheses, we propose the conceptual model of IUR collaborative innovation, energy intensity, and political turnover, as shown in Fig. 1.

Fig. 1: Conceptual model.

Conceptual model of IUR collaborative innovation, energy intensity, and political turnover.

Methodology

Data

China has 34 first-level administrative regions, including 23 provinces, five autonomous regions, four municipalities directly under the central government, and two special administrative regions. Due to the unavailability of data, Tibet, Hong Kong, Macao, and Taiwan are excluded. Therefore, our dataset covers 30 provincial-level administrative regions in China from 2010 to 2018.

The advantages of our data are as follows: first, the panel dataset covers more than 90% of China’s total population and economic output, making the study population highly representative. Second, our data cover different types of regions in China, exceptional types of regions such as old revolutionary areas, border areas, and ecologically degraded areas. Third, the data we use is authoritative, credible, and available. Data on energy consumption and IUR collaborative innovation are from the China Energy Statistical Yearbook and the China Science and Technology Statistical Yearbook, respectively. Data on R&D expenditure input intensity, population density, foreign direct investment (FDI), and industrial structure are from the China Statistical Yearbook. These yearbooks are authoritative and uniform-caliber sources of information. The disadvantage of our data is that there are missing data in the statistical yearbook for specific years in individual provinces. We fill in the missing values through other data sources, such as the National Bureau of Statistics and provincial government work reports.

Besides, the data on the turnover of governors and provincial secretaries are obtained from officials’ resumes on the websites of the people’s governments of each province. The resumes contain detailed information about governors’ and provincial secretaries’ promotion source, tenure, and departure. Price-related indicators, including GDP, internal R&D expenditures, and sales revenue of new products, are converted to constant prices, with 2010 as the base year. The price indices are obtained from each province’s relevant years of the Statistical Yearbook. The descriptive statistics of variables are presented in Table 1.

Table 1 Descriptive statistics of variables.

Model specification

Panel data help address endogeneity issues due to unobservable individual differences. It also reveals additional information about the individual’s dynamic performance. Compared with cross-sectional analysis, panel analysis can handle a series of econometric issues in regional data and improve the precision of estimates (Man, 2016). Therefore, a panel model is constructed for multiple linear regression analysis to investigate the role of IUR collaborative innovation on energy intensity, which is as follows:

$$EI_{it} = \beta _0 + \beta _1CI_{it} + \beta _2RD_{it} + \beta _3PD_{it} + \beta _4FDI_{it} + \beta _5IS_{it} + \varepsilon _{it}$$

(1)

where i and t denote the province and year, respectively; $EI_{it}$ represents energy intensity; $CI_{it}$ indicates collaborative innovation; $RD_{it}$ is R&D expenditure input intensity; $PD_{it}$ is population density; $FDI_{it}$ is FDI; $IS_{it}$ is industrial structure; and $\varepsilon _{it}$ represents a random disturbance term. The coefficient of interest is $\beta _1$, which is used to measure the effect of collaborative innovation on energy intensity.

Further, to test the effect of political turnover on IUR collaborative innovation and energy intensity, we introduce the variable of political turnover and its interaction term with co-innovation in the model (1). The specific model is as follows:

$$\begin{array}{l}EI_{it} = \lambda _0 + \lambda _1CI_{it} + \lambda _2PT_{it} + \lambda _3CI_{it} \times PT_{it} + \lambda _4RD_{it} + \lambda _5PD_{it}\\ \qquad \quad + \,\lambda _6FDI_{it} + \lambda _7IS_{it} + \varepsilon _{it}\end{array}$$

(2)

where $PT$ presents political turnover. In model (2), we are most interested in $\lambda _3$, which reveals whether political turnover has a moderating role in IUR collaborative innovation and energy intensity.

Dependent variable

In this study, energy intensity is the dependent variable expressed as the ratio of primary energy consumption (PEC) to gross domestic regional product (GRDP). This measurement is widely recognized and has been applied in various studies (Farajzadeh and Nematollahi, 2018; Lin and Wang, 2021). The equation is as follows:

$$EI_{it} = \frac{{PEC_{it}}}{{Y_{it}}}$$

(3)

where $PEC_{it}$ is the PEC of the province i in the year t, and $Y_{it}$ is the GRDP of the province i in the year t.

Independent variable

Collaborative innovation is the independent variable. We construct a synergy model of the composite system to measure the degree of synergy of IUR collaborative innovation using the following steps:

Step1: constructing IUR collaborative innovation index system

The selection of the inputs and outputs used to construct the index system is based on the study of Thursby and Kemp (2002) (see Table 2). The full-time equivalent of R&D personnel captures the intensity of input from technical staff. The internal expenditure of R&D reflects investments in the R&D process (Song et al., 2020). Therefore, we use the full-time equivalent of R&D personnel and internal expenditure of R&D as indicators to measure labor and capital input, respectively. Moreover, we add the R&D projects into the input indicators of the industry subsystem, as R&D projects reflect investments in innovation.

Table 2 Index system of IUR collaborative innovation.

Universities and research institutes are mainly responsible for knowledge creation regarding the selection of output indicators. Thus scientific papers and patent applications are output indicators (Bai et al., 2020). Regarding the industry subsystem, innovation activities do not only produce knowledge but also transform knowledge into new products (Fiaz, 2013). Therefore, we use the sales revenue of new products and patent applications as output indicators to reflect the market recognition degree of innovative products and knowledge achievements, respectively.

Step2: Calculating the degree of order

Assuming the synergy of IUR innovation systems is S, $S = f(S_1,S_2,S_3)$, where $S_1$ represents university subsystem; $S_2$ and $S_3$ represent the industry and research institute subsystems, respectively. The order parameter of each subsystem is $e_j$,$e_j = (e_{i1},e_{i2},...,e_{in}),\alpha _{ij} \le e_{ij} \le \beta _{ij},n \ge 1,j \in [1,n]$, where $\alpha _{ij}$ and $\beta _{ij}$ represent the upper and lower limits of the order parameter, respectively. Assuming $e_{1j},e_{2j},...,e_{kj}$ are positive indicators, the higher the value of the indicators, the higher the order of the system. $e_{kj + 1},e_{kj + 2},...,e_{jn}$ are negative indicators, the higher the value of the indicators, the lower the order of the system. Therefore, the degree of order of the subsystem order parameter component $e_j$ can be defined as follows:

$$u_i(e_{ij}) = \left\{ \begin{array}{l}\frac{{e_{ij} - \alpha _{ij}}}{{\beta _{ij} - \alpha _{ij}}},j \in [1,k]\\ \frac{{\beta _{ij} - e_{ij}}}{{\beta _{ij} - \alpha _{ij}}},j \in [k + 1,n]\end{array} \right.$$

(4)

where $u_i(e_{ij})$ represents the degree of order of the subsystem order parameter component $e_j$, $u_i(e_{ij}) \in [0,1]$. The closer the value of $u_i(e_{ij})$ to 1, the higher the contribution of the order parameter to the subsystem. The contribution of the order parameter component $e_j$ to the subsystem $S_i$ can be integrated through the coupling of $u_i(e_{ij})$. Therefore, the degree of order of the subsystem $S_i$ can be defined as follows:

$$u_i(e_i) = {\sum} {\lambda _ju_i(e_{ij})} ,\,\lambda _j \ge 0,{\sum} {\lambda _j = 1}$$

(5)

where $\lambda _j$ represents the weight of each indicator (see Table 1). The weight is determined using the entropy method. We add the time variable to the calculation of the traditional entropy method to compare different years, which makes the analysis result more reasonable. Assuming that there are r years, n provinces and cities, and m indicators, we use $\chi _{\theta ij}$ to represent the j indicator value of province i in θ year. The formula is as follows:

$$y_{\theta ij} = \chi _{\theta ij}/\mathop {\sum}\limits_\theta {\mathop {\sum}\limits_i {\chi _{\theta ij}} }$$

(6)

$$e_j = - k\mathop {\sum}\limits_e {\mathop {\sum}\limits_i {y_{ij}\ln (y_{\theta ij})} ,\quad k = \ln (rn) \,>\, 0}$$

(7)

$$\lambda _j = 1 - e_j/\mathop {\sum}\limits_j {1 - e_j}$$

(8)

Step3: Calculating the degree of synergy of innovation

Assuming that the initial time is $T_0$, the degree of order of each subsystem is $u_i^0(e_i)$; when the time is $T_1$, the degree of order of each subsystem is $u_i^1(e_i)$, and the degree of the synergy of innovation of the IUR system during this period of change is as follows:

$$S = \zeta \root {3} \of {{\left| {u_1^1(e_1) - u_1^0(e_1)} \right|\left| {u_2^1(e_2) - u_2^0(e_2)} \right|\left| {u_3^1(e_3) - u_3^0(e_3)} \right|}}$$

(9)

where $S \in [ - 1,1]$, the value of i is 1, 2, 3. When $u_i^1(e_i) - u_i^0(e_i) \,>\, 0$, $\zeta = 1$; otherwise, $\zeta = - 1$. The more consistent the increase or decrease among the industry, university, and research institute subsystems, the greater the value of the synergy of IUR collaborative innovation.

Moderating variable

In this study, political turnover is the moderating variable. If the provincial leader is newly appointed, the dummy variable is set to 1; otherwise, it is set to 0 (Cao et al., 2019). China’s political system empowers the secretary of the provincial party committee to allocate resources, and their economic decisions have a critical influence on the economic performance of the provinces (Li and Zhou, 2005). Therefore, we employ the turnover of provincial secretaries as a moderating variable.

As the impact of the new policies implemented by new officials on collaborative innovation and energy consumption is difficult to have an effect in a short period, we follow the study of Deng et al., (2019) and use June as the dividing point. Political turnover takes one if the provincial secretary assumes office between January 1 and June 30 and 0 if the provincial secretary assumes office between July 1 and December 31.

Control variables

Several control variables are introduced to address the omitted variables bias. These variables are as follows:

R&D expenditure input intensity (RD). Earlier studies have confirmed that R&D investment is a critical factor in decreasing energy intensity; the more the former is invested, the more the latter is reduced (Fisher-Vanden et al., 2004; Shen and Lin, 2020). Therefore, we include RD as a control variable and employ the ratio of R&D expenditures to GRDP to express it.

Population density (PD). PD is a vital factor affecting energy consumption (Glaeser and Kahn, 2010). China’s high PD has placed tremendous pressure on the energy supply for household consumption (Song et al., 2020). Therefore, we use PD as one of the control variables and express it as the natural logarithm of the PD of each province.

Foreign direct investment (FDI). Knowledge diffusion and technology transfer of energy-efficient technologies among economies depend partly on FDI (Keller, 2004). Evidence has confirmed that FDI can reduce energy intensity (Herrerias et al., 2013). As a result, we employ FDI as one of the control variables and express it as the natural logarithm of the net inflows of FDI.

Industrial structure (IS). IS is an essential factor affecting energy consumption (Zhang et al., 2020). Therefore, we introduce IS as a control variable into the model and measure it as the proportion of secondary industry in the GRDP.

Robustness tests

Considering the possibility of endogenous bias in our model caused by reverse causality, we use the instrumental variables approach for robustness testing. The history of the opening of the commercial port is used as an instrumental variable. First, the earlier a region opens up to the outside world and trade, the longer the history of technology introduction and self-innovation. The variable satisfies the relevance condition of an instrumental variable. Second, the time of opening foreign ports cannot be altered and is rigorously exogenous. Moreover, we employ other robustness methods, such as variable substitution, grouped regression results, counterfactual simulation, and considering the effects of relevant policies.

Results

Evaluation results of IUR collaborative innovation

From the time trend of IUR and its subsystems from 2010 to 2018 (see Fig. 2), we find that (1) the degree of synergy of the subsystems has been increasing, and the degree of synergy of the university subsystem is higher than that of the industry and research institute subsystems, ranging from 0.0769 to 0.1557. (2) In the sample interval, the average degree of synergy of IUR collaborative innovation is 0.0055. (3) The degree of synergy in 2012 (0.0066) and 2018 (0.0067) is higher than that of other years. This is consistent with the degree of order in the three subsystems in 2012 and 2018. (4) The degree of synergy of IUR collaborative innovation in 2010 (0.0040) and 2015 (0.0049) is lower than that of other years. This is because, in 2010 and 2015, the degree of synergy of the university subsystem increased significantly, while the degree of synergy of the industry and research institute subsystems increased less rapidly.

Fig. 2

The time trend of IUR and its subsystems.

Figure 3 depicts the average degree of synergy of IUR collaborative innovation and its subsystems in 30 Chinese provinces from (a) 2010 to 2012, (b) 2013 to 2015, (c) 2016 to 2018, and (d) 2010 to 2018. Jiangsu, Guangdong, and Beijing are the top three provinces with average degrees of synergy of IUR collaborative innovation, which are 0.0212, 0.0185, and 0.0154, respectively. In these provinces, the driving forces are from two aspects. One is that the local government strongly supports IUR collaborative innovation, and the policies are enacted to build an innovation system. For example, the government of Jiangsu province has issued the “construction of a technological innovation system combining industry, university, and research,” which provided favorable conditions for IUR collaborative innovation. The other driving force is that several high-tech industries, universities, and research institutes in Jiangsu, Guangdong, and Beijing have formed a mature innovation system and created favorable internal conditions for IUR collaborative innovation. However, Qinghai, Ningxia, Hainan, and Xinjiang exhibit a relatively low synergy of IUR collaborative innovation. This is due to the lower level of economic development, limited R&D investment, and fewer universities and research institutions in these provinces.

Fig. 3: The average of the synergy degree of IUR collaborative innovation in 30 Chinese provinces.

From 2010 to 2012 (a), 2013 to 2015 (b), 2016 to 2018 (c), and 2010 to 2018 (d).

In terms of vertical trends, the degree of synergy of IUR collaborative innovation of Xinjiang, Qinghai, Gansu, Inner Mongolia, Ningxia, Jilin, Yunnan, Guizhou, Guangxi, and Hainan is relatively low and remained unchanged from 2010 to 2018. The degree of synergy of IUR collaborative innovation of Beijing, Guangdong, Jiangsu, and Shanghai is relatively high without fluctuation from 2010 to 2018. The degree of synergy of Hebei and Zhejiang shows a downward trend from 2010 to 2015. Moreover, after years of construction of a collaborative innovation system, synergy has improved dramatically since 2016. In Sichuan, Shaanxi, Chongqing, Hunan, and Tianjin, the degree of synergy fluctuated significantly, increasing and decreasing. The synergy of Liaoning, Henan, Jiangxi, Fujian, Shanxi, Heilongjiang, and Hubei has increased from 2010 to 2018, whereas that of Shandong and Anhui has declined. Regional differences exist in IUR collaborative innovation, and many provinces are below the average value of IUR collaborative innovation. In addition, there are no noticeable changes between the three subsystems from 2010 to 2018. The contribution of industry and university subsystems in Guangdong, Jiangsu, Zhejiang, and Shandong is higher than that of other provinces, and the subsystem of the research institute in Beijing has the highest proportion.

Impact of IUR collaborative innovation on energy intensity

We perform the regression analysis of Model (1) in Table 3. We first run the regression using ordinary least squares (OLS), and the results in Column (1) show that the regression coefficient of collaborative innovation is −0.0582 and is significant at the 1% level. Second, we employ the fixed effects (FE) method and random effects (RE) method for the regression in Columns (2) and (3). We used the statistics constructed by the Hausman test to determine whether the FE or RE models should be adopted. The null hypothesis of the Hausman test is that unobservable individual effects are unrelated to the explanatory variables. In Table 3, the Hausman test results show that FE should be selected. In Column (2), the coefficient between IUR collaborative innovation and energy intensity is negatively significant at 10%. Our results mean that improving IUR collaborative innovation can help reduce energy intensity.

Table 3 Estimation results of collaborative innovation on energy intensity.

Considering that the current energy intensity may be affected by the previous energy intensity, we add a one-period lag of energy intensity (L.EI) as instrumental variables to model (1) and use the system generalized method of moments (SYS-GMM) method for estimation (Arellano and Bond, 1991; Blundell et al., 2000). Column (4) results show that IUR collaborative innovation still negatively affects energy intensity. This result supports Hypothesis 1.

Our findings support the results in the study of Hille and Lambernd (2020) on the role of innovation in energy intensity reduction. Furthermore, additional evidence affirms that IUR collaborative innovation is conducive to energy intensity reduction. For instance, Samargandi (2019) stated that technological innovation reduces energy intensity in the short and long run. Kafouros et al., (2015) found that enterprises that work with colleges and research institutes can enhance technological innovation performance by promoting specialized knowledge.

Moderating effect of political turnover

Table 4 shows the estimation results of the moderating effect of political turnover. The Hausman test show that FE is appropriate. The results of FE show that the regression coefficient of the interaction term between collaborative innovation and political turnover is −0.0131, and it passes the 10% significance level. This suggests that political turnover exacerbates the negative effect of collaborative innovation on energy intensity. When political turnover occurs, the inhibitory effect of IUR collaborative innovation on energy intensity becomes stronger. Also, we use the SYS-GMM method for regression in Column (3) and the results indicate that the coefficient of the interaction term is negatively significant at the 5% level.

Table 4 Estimation results of the moderating effect of political turnover.

Our result supports Hypothesis 2. It can be explained using the Chinese official evaluation system. In the past four decades, one of the driving factors of China’s sustained and rapid GDP growth is the “promotion tournament mechanism”. This argues that the promotion incentives of officials affect technological progress (Li and Zhou, 2005). Political turnover has intensified promotion tournaments among officials (Deng et al., 2019), and the energy-saving performance of newly appointed officials positively affects their promotion (Chen et al., 2016). To gain a promotion advantage by exceeding emission reduction targets, those incumbents are committed to technological innovation (Pu and Fu, 2018), thereby reducing energy intensity.

To ensure that the benchmark regression results are robust, we conduct a series of robustness tests, including instrumental variable methods, variable substitution, group regressions, counterfactual simulations, and consideration of the impact of other policies. The regression results of the robustness tests are presented in the Appendix. All the results support our research hypotheses.

Regional heterogeneity

Given some differences between regions, such as GDP growth rate, human resources, and geographical conditions, the inhibitory effect of IUR collaborative innovation on energy intensity may also be heterogeneous. To compare the internal differences between different regions, 30 Chinese provinces are divided into three according to economic and social development levels (Pu et al., 2020). These divisions are the eastern, central, and western regions¹.

The results of Columns (1), (3), and (5) of Table 5 show that IUR collaborative innovation in the eastern region negatively affects energy intensity. However, this negative implication is insignificant in the central and western regions. The sizeable regional disparity may be because the degree of synergy of IUR collaborative innovation in the eastern region is higher than in other regions. As an essential way to accelerate the transformation of scientific achievements, IUR collaborative innovation provides internal R&D capabilities for reducing energy intensity through technological progress. Second, the provinces in the central and western regions need to be developed; thus, economic growth is the primary task of the regions. Facing economic growth pressure, these provinces are prone to increase energy input rather than IUR collaborative innovation, which requires a lot of monetary and labor input. Therefore, the impact of IUR collaborative innovation on energy intensity reduction needs to be stronger in the central and western regions. Third, due to the energy resource endowment, China’s large-scale energy bases are distributed primarily in the western region. Western provinces have much higher energy resources than other parts. As a result, on the production side, the focus of the western provinces is on producing energy rather than improving energy efficiency on the consumption side. IUR collaborative innovation also received little attention in the western region as a driver of energy efficiency.

Table 5 Estimation results of models of different regions.

The results of Columns (2), (4), and (6) of Table 5 show that in the central and western regions, political turnover significantly reinforces the restraining effect of IUR collaborative innovation on energy intensity. However, political turnover has no moderating effect in the eastern region. This may be because provinces in the eastern region, such as Beijing, Shanghai, and Jiangsu, have high marketization. Fierce market competition has given birth to a complete market mechanism and property rights system, dismantling local officials’ discretionary control over resources to a certain extent, thereby weakening the role of local political turnover.

Discussion

Our research helps enrich the studies on how political turnover affects IUR collaborative innovation and energy intensity. In the existing literature, few scholars put political turnover, IUR collaborative innovation, and energy intensity in the same analytical framework. By revealing the pathways of energy intensity reduction from the perspectives of IUR synergistic innovation and political turnover, our research contributes to realizing China’s dual carbon goals.

By adopting a synergetic composite system model, we estimate the degree of synergy of IUR collaborative innovation from 2010 to 2018 for 30 Chinese provinces. The synergetic model of a hybrid system can reflect the collaborative innovation process from inputs to outputs and reveal the contribution of IUR institutions in the cooperation. The evaluation results of IUR collaborative innovation suggest that the degree of synergy of the subsystems has been increasing, and the degree of synergy of the university subsystem is higher than that of the industry and research institute subsystems. Moreover, there are regional differences in IUR collaborative innovation, and many provinces are below the average value. China’s IUR collaborative innovation has plenty of room to improve.

Although earlier research has confirmed the impact of technological innovation on energy intensity (Chakraborty and Mazzanti, 2020; Hille and Lambernd, 2020), more information must be needed about how IUR collaborative innovation affects energy intensity. By constructing an econometric test model, we empirically examined the effect of IUR collaborative innovation on energy intensity. This finding confirms the critical role of IUR collaborative innovation in improving the regional technological innovation level, revealing that energy intensity can be reduced by increasing IUR collaborative innovation level. Several robustness tests are performed to test whether the estimates are robust. Our results enrich the existing explanation about the driving factors of energy intensity.

Moreover, we collect novel data on the turnover of 30 Chinese provincial leaders from 2010 to 2018 and analyze whether political turnover moderates the impact of IUR collaborative innovation on energy intensity. While previous studies pay more attention to how political turnover affects economic growth (Li and Zhou, 2005), corporate strategy (Choi et al., 2021), and innovation (Pertuze et al., 2019), there is little research on the impact of political turnover on energy intensity. This is the first time to analyze the role of political turnover in the effects of IUR collaborative innovation on energy intensity. This result supports the theory of promotion tournaments for Chinese officials and expands related research on the relationship between officials and energy.

Further, we analyze the regional heterogeneity of the dampening effect of IUR collaborative innovation on energy intensity. We find that IUR collaborative innovation only in the eastern region negatively affects energy intensity. This result may be related to the distribution of energy resources and economic growth level in China. The heterogeneity of moderating effects shows that political turnover significantly reinforces the restraining impact of IUR collaborative innovation on energy intensity in the central and western regions. This may be related to the eastern region’s complete market mechanism and property rights system. The heterogeneity analysis results help policymakers propose practical, targeted, and feasible solutions to reduce energy intensity.

Conclusions and implications

Due to the challenges of climate change and energy depletion, reducing energy intensity has become an interesting topic. The effect of technological innovation on energy intensity has attracted much attention, and many results have been achieved. Unlike the previous literature, this study analyzes the relationship between innovation and energy intensity based on IUR collaboration using 30 Chinese provincial data from 2010 to 2018. More importantly, we analyze the moderating role of political turnover and explore the boundary conditions of the effect of IUR collaborative innovation on energy intensity, which is not covered in previous studies. The instrumental variable method, variable substitution, grouped regression, and counterfactual simulation are used to test the result’s robustness.

The findings are as follows. First, the coefficient between IUR collaborative innovation and energy intensity is negatively significant, indicating that IUR collaborative innovation has an inhibitory effect on energy intensity. Second, political turnover exacerbates the negative impact of collaborative innovation on energy intensity. When political turnover occurs, the inhibitory effect of IUR collaborative innovation on energy intensity becomes stronger. Third, the regional heterogeneity analysis shows that there needs to be more evidence that the inhibitory effect of IUR collaborative innovation on energy intensity exists in the central and western regions. Moreover, the moderating role of political turnover is not significant in the eastern region due to the higher level of marketization in the eastern region.

Based on the results, some policy implications are put forward. First, much importance should be attached to IUR collaborative innovation, especially in the central and western regions. In China, the innovation base is the cooperation carrier to promote IUR collaborative innovation and transfer scientific achievements. Promoting the construction of the IUR collaborative innovation base lies in focusing on core technology areas, organizing teams to gather for research, and exploring sustainable collaboration mechanisms between industry, academia, and research. At the same time, based on regional innovation resource endowment, the government should build an innovation resource-sharing platform to promote the complementary advantages and efficient use of regional resources, emphasizing the flow of scientific and technological talents among provinces. Attach importance to improving energy efficiency by enhancing R&D capabilities and coordinating regional IUR collaborative innovation. Second, newly appointed officials, especially provincial secretaries, must consider the impact the new policies they actively pursue after assuming office may have on the IUR collaborative innovation behavior of the parties involved. Local governments should provide financial support and policy preferences to strengthen collaborative innovation platforms construction and inter-regional cooperation. Third, the evaluation system for officials should be improved.

The following are some limitations of our study. First, the data on political turnover can be further disaggregated by other variables, including promotion, leveling, and demotion, to explore the role of political turnover in the effect of IUR collaborative innovation on energy intensity under different conditions. Second, future research can focus on China’s municipal level to calculate the degree of synergy of IUR collaborative innovation of cities and further explore its impact on energy intensity.