EN中文

Journal of Resources and Ecology >

2024 , Vol. 15 >Issue 6: 1416 - 1432

DOI: https://doi.org/10.5814/j.issn.1674-764x.2024.06.003

Resource Utilization and Green Development

Two-sided and Heterogeneous Effects of Environmental Regulation on the Technological Innovation Efficiency of Iron and Steel Enterprises in China

  • GE Zehui , 1, 2 ,
  • SUN Xiaojie , 1, * ,
  • GUO Zhiyuan 1
Expand
  • 1. School of Management and Economics, University of Science and Technology Beijing, Beijing 100083, China
  • 2. The Institute of Low Carbon Operations Strategy for Beijing Enterprises, University of Science and Technology Beijing, Beijing 100083, China
* SUN Xiaojie, E-mail:

GE Zehui, E-mail:

Received date: 2023-09-01

  Accepted date: 2024-03-20

  Online published: 2024-12-09

Supported by

The National Natural Science Foundation of China(71871016)

The Fundamental Research Funds for the Central Universities(FRF-DF-20-68)

Fold

Abstract

In this study, the relationship between environmental regulations and technological innovation efficiency is empirically examined via panel data from 33 iron and steel enterprises (ISs) in China between 2015 and 2021. The results show that the average “innovation compensation effect” of environmental regulations on the technological innovation efficiency of ISs exceeds the average “compliance cost effect”, thus resulting in a clearly positive net effect. Both the two-sided effects and the net effects vary across different years, geographical regions, and types of property rights. As the quantile of technological innovation efficiency increases, the positive influence of environmental regulations tends to increase. Furthermore, the strengthening of financing constraints and firm competitiveness enhances the positive impact of environmental regulations on the technological innovation efficiency of ISs. Additionally, a double-threshold effect of environmental regulations on the technological innovation efficiency of ISs is revealed in this study. The realisation of the Porter hypothesis occurs when financing constraints and firm competitiveness fall within specific threshold intervals. This research not only deepens our understanding of the relationship between environmental regulations and the technological innovation efficiency of ISs but also provides valuable policy insights for optimising environmental regulations to facilitate targeted improvements in the level of technological innovation efficiency.

Key words： environmental regulation; technological innovation efficiency; two-sided effects; heterogeneous effects; threshold effects

Cite this article

GE Zehui , SUN Xiaojie , GUO Zhiyuan . Two-sided and Heterogeneous Effects of Environmental Regulation on the Technological Innovation Efficiency of Iron and Steel Enterprises in China[J]. Journal of Resources and Ecology, 2024 , 15(6) : 1416 -1432 . DOI: 10.5814/j.issn.1674-764x.2024.06.003

1 Introduction

Climate change and energy insecurity are pressing global environmental challenges. The production of the steel industry, which heavily relies on fossil fuels, is closely linked to both greenhouse gas emissions and potential energy insecurity (Wu et al., 2023). Notably, China faces significant pressure to reduce carbon dioxide emissions and address energy security concerns.
In recent years, China’s iron and steel industry has been developing rapidly. In 2020, China’s crude steel output exceeded one billion tons, ranking first in the world in terms of crude steel output. Based on data from the World Steel Association, 29 Chinese iron and steel enterprises (ISs) were ranked among the top 50 global ISs in terms of production in 2020. However, the long-term negative externalities of steel production cannot be ignored, as steel production has been argued to come at the expense of health and sustainable development potential (Mele and Magazzino, 2020). China is the world’s largest consumer of energy and its largest emitter of carbon dioxide. In 2020, China’s primary energy consumption accounted for 26.13% of total global energy consumption, and carbon dioxide emissions reached 9.9 billion t, accounting for 28.76% of the world’s total. China’s steel production relies heavily on coal-based fossil fuels, which are important factors leading to high carbon emissions and resulting in a series of environmental problems (e.g., Liu et al., 2016; Subramani et al., 2018). To address the energy-intensive and polluting nature of the iron and steel industry and improve its development model, governments at all levels in China have attempted to design and implement environmental regulations. Enhancing the technological innovation efficiency of ISs is a critical issue for deepening the implementation of innovation-driven development strategies, thus promoting the overall development of China and fulfilling the requirements necessary for building a modern socialist country. Additionally, the Chinese government is committed to achieving “carbon peaking” by 2030 and “carbon neutrality” by 2060. Governments in China look to encourage enterprises to increase their investment in technological research and to facilitate continuous improvements in enterprise innovation levels through environmental regulations (Liu et al., 2018). The technological innovation efficiency of ISs serves as a crucial indicator reflecting their level of technological innovation performance. Therefore, this study aims to explore the two-sided and heterogeneous effects of environmental regulations on the technological innovation efficiency of ISs.
There are generally two main views regarding the relationship between environmental regulations and the technological innovation efficiency of enterprises. The first view claims that the implementation of environmental regulations promotes an increase in the investment of heavily polluting enterprises in technological innovation and facilitates the performance of technological innovation activities, thereby improving the technological innovation efficiency of enterprises. This view aligns with the Porter hypothesis (Porter and Linde, 1995). The second view posits that under the urgent pressure of environmental regulations, the high cost and lengthy renovation process for polluting enterprises lead to a “crowding out” of research and development funds, thus resulting in a negative impact of environmental regulations on the technological innovation efficiency of enterprises (Ouyang et al., 2020). The studies that analyse the impact of environmental regulations employ a relatively limited scope, focusing solely on the “innovation compensation effect” or the “compliance cost effect”. However, the net effect of environmental regulations on the technological innovation efficiency of ISs arises from opposing two-sided effects that serve to counterbalance each other. The net effect of environmental regulations remains uncertain due to the presence of such two-sided effects (Wu et al., 2020). Therefore, it is necessary to analyse both the two-sided and net effects of environmental regulations through the effect decomposition method.
Furthermore, traditional regression models generally employ the mean regression method, where the estimated coefficient represents the mean effect. This approach does not enable observing the changing process of the impact of environmental regulations on the technological innovation efficiency of ISs. This research aims to explore the changing trend of the extent of the impact of environmental regulations as the quantiles of technological innovation efficiency increase. Financing constraints are critical factors that affect the technological innovation efficiency of enterprises (Fang et al., 2020). Furthermore, market competition can incentivise firms to invest in R&D to increase the efficiency of technological innovation (Aghion et al., 2005). Therefore, the moderating role of financing constraints and firm competitiveness merits further attention.
Moreover, few studies have examined the nonlinear relationship between environmental regulations and the technological innovation efficiency of ISs. Notably, our paper suggests the potential presence of threshold effects in this relationship. It is important to conduct research to uncover the true nature of the relationship (determining whether it exhibits distortion or discontinuity) between environmental regulations and the technological innovation efficiency of ISs, particularly once the thresholds of financing constraints and firm competitiveness are reached. This paper aims to empirically examine the two-sided and heterogeneous effects of environmental regulations on IS technological innovation efficiency. Our study offers a more nuanced perspective on the relationship between environmental regulations and the technological innovation efficiency of ISs.
This paper offers three main contributions. First, the two-sided and net effects of environmental regulations on the technological innovation efficiency of ISs are both measured in this study. Specifically, a two-tier stochastic frontier analysis (two-tier SFA) is conducted. The advantage of this model lies in its ability to assess not only the two-sided and net effects of environmental regulation but also the disparities in the effects across different years and regions. Second, the heterogeneous impacts of environmental regulations are explored under different quantiles of technological innovation efficiency in this paper. Specifically, the ISs exhibiting varying technological innovation efficiencies are classified into five groups. According to the estimation of the Bayesian quantile regression (BQR) model, there are differences in the effects of environmental regulations. The BQR results enrich the rules covering the variation in regression coefficients of the explanatory variables and capture the extent of the impact of environmental regulations on technological innovation efficiency under different quantiles. Third, from the perspective of the moderating effect of financing constraints and firm competitiveness, the nonlinear impact of environmental regulations on the technological innovation efficiency of ISs is analysed. The study employs the panel threshold regression (PTHR) model to examine these threshold effects.
The remainder of this paper is structured as follows. In Section 2, the related literature is reviewed, and the research hypotheses are developed. In Section 3, the materials and methods are presented. In Section 4, the results are presented, and discussions are offered. In Section 5, the conclusions and policy implications are summarised.

2 Literature review and hypotheses

Hardin (1968) claimed that the limited resources of the earth generally have the attributes of public goods because property rights cannot be defined or the cost of such definition is too high. This is an important source of the “tragedy of the commons” that results from the impact of limited resources on quality of life. With the rapid development of ISs, the output of products has increased, and the resulting negative externalities are difficult to eliminate automatically through market mechanisms. Environmental regulations have become the main, effective tool for solving negative externalities, as they internalise costs, improve the technological innovation efficiency of ISs, and increase social welfare. However, the effectiveness of environmental regulations in promoting improvements in technological innovation efficiency is not universally guaranteed.

2.1 Environmental regulations and technological innovation efficiency

The relationship between environmental regulations and the technological innovation efficiency of enterprises has long been a subject of controversy. There are two different theoretical perspectives regarding this relationship, namely, neoclassical theory, which is also known as the pollution haven theory, and the Porter theory. According to neoclassical theory, environmental regulations increase the production costs of pollution control, energy conservation, and emission reduction for enterprises; this, in turn, may limit investments in technological innovation, thus hindering the improvement of technological innovation efficiency. The pollution haven hypothesis posits that the impact of environmental regulations on the industrial sector increases the cost of production for enterprises (Palmer et al., 1995). In contrast, moderate environmental regulations can exert an innovation compensation effect, thus enabling enterprises to engage in innovation activities (Su and Xu, 2022) and ultimately enhancing the level of technological innovation efficiency.
According to most scholars, environmental regulations promote the improvement of technological innovation efficiency in enterprises. In cities powered by coal, the SO2 removal rate is used as a measure of governments’ environmental regulation intensity; this serves as a compliance cost for regulation implementation. To comply with environmental protection standards, enterprises incur costs related to governance, which may lead to trade-offs with the resources allocated to manpower and R&D; over time, this can lead to the Porter hypothesis. Under the pressure of early costs, coal enterprises are forced to carry out technological innovation (Qian et al., 2021). As studied by Cheng and Kong (2022), environmental regulations enhance green total factor productivity by narrowing the technological gap. This, in turn, leads to improved marginal profits for enterprises and fosters technological innovation efficiency.
However, the impact of environmental regulations on the technological innovation efficiency of ISs remains uncertain. On the one hand, when heavily polluting companies exhibit enthusiasm and a sense of responsibility for environmental governance, they pursue technological innovation (Blackman, 2010). Enterprises increase R&D investment and their number of patent applications, contributing to the improvement of the technological innovation efficiency of ISs. On the other hand, innovation activities possess inherent properties of high risk, high investment, long cycles, and high uncertainty (He and Tian, 2013). Stakeholders can overestimate the cost of financing and underestimate market risks due to information asymmetry (Jensen and Meckling, 1976). The participation of enterprises in innovation activities can lead to a decrease in marginal returns, which, in turn, leads to a decrease in the efficiency of technological innovation. Therefore, researchers should account for the two-sided effects of environmental regulations.
The impact of environmental regulations on the technological innovation efficiency of ISs varies across different regions and market conditions. Yin et al. (2015) found that environmental regulations in the eastern region are more stringent than those in the central and western regions. Some scholars find that reasonable environmental regulations produce “innovation compensation effects” and partially or completely offset their “compliance cost effects” (Ramanathan et al., 2018; Cai et al., 2020). Yuan and Zhang (2020) reported that improving the technological innovation efficiency of enterprises alleviates the inhibitory effect of environmental regulations on industrial development. Other studies have shown that, owing to the limitations of the intensity of environmental regulations, absorptive capacity, and economic development, the relationship between environmental regulations and the technological innovation efficiency of enterprises is complex (Shen et al., 2019). As the quantile of the technological innovation efficiency of ISs improves, the heterogeneous effects of environmental regulations merit further discussion.
The following hypotheses are proposed based on the above findings:
Hypothesis 1a: Environmental regulations have both two-sided and heterogeneous effects on the technological innovation efficiency of iron and steel enterprises.
Hypothesis 1b: The innovation compensation effect of environmental regulations is greater than the compliance cost effect of environmental regulations.

2.2 Moderating role of financing constraints and firm competitiveness

The traditional view holds that financing constraints exert a negative effect on the technological innovation efficiency of enterprises (Andersen, 2017; Zhou, 2017). This is due mainly to the long investment return cycle and high level of investment risk associated with enterprise technological innovation. Moreover, the information asymmetry between investors and enterprises makes it difficult for enterprises to secure long-term and stable financing support (Hottenrott and Peters, 2012). Zhang et al. (2015) found that traditional financial institutions are often reluctant to lend to high-risk and low-reward technological innovation projects. These factors contribute to a reduction in the level of R&D investment, which consequently impacts the technological innovation efficiency of enterprises. However, financing constraints may be conducive to enhancing technological innovation efficiency (Xie et al., 2013). Financing constraints compel firms to make decisions with respect to optimal innovation input and efficient resource allocation (Luo, 2011). This approach adequately fulfils the need for innovation funds and facilitates the transition of technological innovation in ISs from being factor-driven to being efficiency-driven. In addition, Lin and Ma (2022) argued that technological innovation behaviours lead to environmental governance externalities and knowledge spillovers. Financing constraints mitigate the innovation crowding out effect of environmental regulation through the provision of R&D subsidies and tax cuts. Some scholars have reported that financial subsidies and tax incentives promote improvements in the technological innovation efficiency of enterprises and alleviate the inhibitory effect of financing constraints (Shen et al., 2022). As the level of financing constraints intensifies, environmental regulations may prove more advantageous in enhancing the technological innovation efficiency of ISs.
Improving firm competitiveness is crucial for enhancing the technological innovation efficiency of enterprises. The enhancement of firm competitiveness is instrumental to enterprise differentiation in terms of both technological progress and risk tolerance (Anwar, 2018). Hermundsdottir and Aspelund (2021) argued that there is a positive correlation between corporate competitiveness and sustainable innovation. Furthermore, the synergistic combination of environmental regulations and firm competitiveness enable ISs to achieve greater “innovative compensation.” Several scholars have demonstrated that the joint influence of environmental regulations and enhanced firm competitiveness encourages enterprises to engage in technological innovation activities (Horbach et al., 2012). Moreover, knowledge sharing and organisational innovation are vital prerequisites for enhancing firm competitiveness (Arsawan et al., 2022). As firm competitiveness improves, the knowledge sharing and organisational innovation framework within enterprises is established, thereby reinforcing the positive impact of environmental regulations on the technological innovation efficiency of ISs.
Based on the above analysis, the following hypotheses are proposed:
Hypothesis 2: Financing constraints have a positive moderating effect on the relationship between environmental regulations and the technological innovation efficiency of iron and steel enterprises.
Hypothesis 3: Firm competitiveness has a positive moderating effect on the relationship between environmental regulations and the technological innovation efficiency of iron and steel enterprises.
Although extensive research has been conducted on the relationship between environmental regulations and the technological innovation efficiency of enterprises, several limitations remain. First, most researchers explore the relationship between environmental regulations and technological innovation efficiency by using panel fixed effects regression, the spatial Durbin model, or the mixed least squares method and fail to scientifically measure the two-sided effects of environmental regulations. Second, the methods used in the literature not only fail to address the heterogeneous impacts of environmental regulations on the technological innovation efficiency of ISs across different quantiles but also fail to examine whether the moderating role of financing constraints and firm competitiveness increases with improvements in technological innovation efficiency. Third, the literature has failed to account for the threshold effect. When financing constraints and firm competitiveness reach a specific threshold, a nonlinear impact of environmental regulations on the technological innovation efficiency of ISs may occur. Therefore, utilising a combination of two-tier SFA, BQR, and PTHR to examine the two-sided, heterogeneous, and nonlinear effects of environmental regulations on the technological innovation efficiency of ISs is recommended.
The mechanism by which environmental regulations effect the technological innovation efficiency of ISs is illustrated in Fig. 1.
Fig. 1 Effect mechanism assessed in this study

3 Data and methodology

3.1 Data description and analysis

3.1.1 Dependent variable

Since traditional data envelopment analysis (DEA) uses only the initial input and the final output of the innovation chain to measure efficiency, the intermediate output and reinput are not considered. To link the two phases of input‒output activities and reveal the “black box” in the calculation process, Kao and Hwang (2008) proposed a two-stage chained network DEA model. This model reflects the correlation between the two stages by setting the intermediate input‒output of the two stages to the same weight. Wang et al. (2016) considered total profits as the total output of two-stage technological innovation efficiency. To measure the technological innovation efficiency of ISs while considering the input lag, a relational network DEA model is constructed based on the value innovation chain theory proposed by Hansen and Birkinshaw (2007). The two-stage chained network process of the technological innovation of ISs is shown in Fig. 2 (the details of the method and efficiency values are given in Appendices B-C).
Fig. 2 Two-stage technological innovation efficiency (TIE) input-output evaluations
Owing to the time lag between the initial input and the final output of technological innovation activities in ISs, the initial input for technological innovation adopts data covering the period of 2015 to 2019, the intermediate output corresponds to the data from 2016 to 2020, and the commercialisation output (final output) is based on data covering the period of 2017 to 2021. The data are from the China Stock Market & Accounting Research (CSMAR) database, the Chinese Research Data Services (CNRDS) database, and the annual reports of enterprises.

3.1.2 Independent variable

Environmental regulation (ER) is the core independent variable used in this study. We refer to Wang et al. (2022) and Lu et al. (2021), who used investment in governance per unit of pollutant as a proxy variable for the intensity of environmental regulations. The formula is as follows:
$ E R I_{i t}=\frac{S E I_{i t}}{T E_{i t}}=\frac{E I_{i t} / \overline{E I}_{i t}}{\sum S E_{i j t}}$,
$ S E_{i j t}=\frac{E_{i j t}-\min E_{i j t}}{\max E_{i j t}-\min E_{i j t}}$,
where $ E R I_{i t}$ represents the intensity of environmental regulations; $ S E I_{i t}$ is the dimensionless value of pollutant emissions ① in year t of province i and is the standardised form of Eijt; TEit represents the total pollutants treated; EIit denotes the completed investment in industrial pollution control in year t of province i; and EIit denotes its mean value. Eijt denotes pollutant emissions; and SEijt is the standardized form of Eijt.

3.1.3 Moderator variables

The moderating variables are financing constraints and corporate competitiveness. Hadlock and Pierce (2010) suggested constructing the size-age (SA) index by using two exogenous variables: firm size and age. Firm competitiveness refers to the ability of an enterprise to consistently satisfy customer needs while generating profits. This capability is achieved by providing goods and services in the marketplace that deliver greater customer value than competitors do. We measure the market competitiveness of enterprises by assessing the firm’s market share (MS), which is determined by the proportion of the firm’s operating income to the total turnover of the IS industry.

3.1.4 Control variables

The control variables selected include the scale of R&D investment, assets and liabilities, technical market turnover, technology finance, and regional economic growth (Palmié et al., 2020; Huang et al., 2021; Liu and Liu, 2022). To mitigate sample dispersion and heteroscedasticity, a logarithmic transformation is applied to certain variables.
Using 2015 as the base period, the firm’s intangible assets are adjusted for inflation by using the industrial producer price index. For the macro data at the provincial level, the GDP deflator is applied to adjust the GDP and technology market turnover of provinces for inflation. The linear interpolation method is employed to fill in a small number of missing data. Then, 33 steel enterprises with relatively complete data covering the period from 2015 to 2021 are selected. These sample enterprises are listed on the Shanghai and Shenzhen stock exchanges in China. In this study, enterprise-level data taken from the CSMAR database, CNRDS database, China Environmental Statistical Yearbook, and firm annual reports are used. Appendix A provides a list of enterprises along with their stock codes and property rights classification (The Appendix content can be found on the journal website). Table 1 presents the definitions of the variables.
Table 1 Definitions of the main variables
Variable nature Variable name Variable definitions Provincial or enterprise level data Symbolic representation
Dependent variable Technological innovation efficiency of ISs The comprehensive efficiency which is obtained by multiplying the stage efficiencies is the proxy variable of the technical efficiency. Enterprise level TIE
Core explanatory variable Environmental regulation The investment in the treatment of pollutants per unit, i.e., the intensity of environmental regulation Provincial level ER
Moderator variable Financing constraints Financing Constrained SA Index Enterprise level SA
Firm competitiveness The proportion of the firm’s operating income to the total industry revenue, i.e., the market share Enterprise level MS
Control variable R&D investment scale R&D investment as a percentage of operating income for the year Enterprise level R&D
Assets and liabilities ratio Total corporate liabilities as a percentage of total corporate assets Enterprise level lnDAR
Technical market turnover Technology market turnover by province Provincial level lnTM
Financial technology level The comprehensive score of science and technology finance in each province as calculated via the entropy weight method② Provincial level TF
Regional economic growth level Regional GDP for the year Provincial level lnGDP

② The entropy weight method is used to determine the weights of the four indicators: urban science and technology employment, urban financial practitioners, local financial science and technology expenditure, and the balance of the RMB loans of financial institutions in each province.

3.1.5 Data description

Table 2 presents the descriptive statistics of all the variables used in the empirical model. The standard deviation (SD) and the maximum value indicate significant variations in the technological innovation efficiency of ISs, the environmental regulations faced by enterprises, and other factors. The median and tertile of TIE show that most ISs are at a low level of technological innovation efficiency. ISs have not reached the state of effective input-output (a TIE of less than 1), and most enterprises are faced with low environmental regulations. Furthermore, the table reveals other findings. The mean and median values (prelogarithm data) of the debt ratios fall within the optima l range (40%-60%). There is a large gap in the technology market turnover between provinces; this indicates that the development of technology finance is still in its early stages and highlights the issues of imbalance and inadequacy in technology finance. In recent years, the regional economic disparity among provinces has not diminished, and economic growth has been uneven.
Table 2 Descriptive statistics of the main variables
Variables Observations Mean SD Min Q(0.25) Median Q(0.75) Max P value of the K-S normality test P value of the S-W normality test
TIE 165 0.1533 0.2017 0.0055 0.0279 0.0549 0.2108 0.9869 <0.001 <0.001
ER 165 0.8269 1.0592 0.1152 0.2751 0.4984 1.0836 8.1899 <0.001 <0.001
SA 165 6.8597 1.6527 3.4435 5.5747 6.9203 8.0490 10.6428 0.200 0.022
MS 165 0.5601 0.7145 0.0040 0.1168 0.3366 0.7494 4.5317 <0.001 <0.001
R&D 165 2.2858 1.4414 0.0400 1.1300 2.3800 3.4300 5.7600 0.001 <0.001
lnDAR 165 3.8932 0.4707 2.4060 3.5482 4.0445 4.2650 4.7165 <0.001 <0.001
lnTM 165 5.2384 1.4396 1.1086 4.3580 5.5309 6.3507 8.3933 0.001 0.001
TF 165 0.2286 0.1373 0.0569 0.1222 0.1831 0.3373 0.5950 <0.001 <0.001
lnGDP 165 10.3377 0.6176 8.8233 9.9924 10.3339 10.7391 11.4261 <0.001 <0.001

Note: The prefix “ln” before the explanatory variables denotes that it takes the logarithmic form. Both the Kolmogorov-Smirnov (K-S) test and Shapiro-Wilk (S-W) test are normality tests used to test whether each main variable has a normal distribution.

3.2 Methodology and model specification

3.2.1 Benchmark model

The purpose of this study is to analyse the impact of environmental regulations on the level of technological innovation efficiency. The basic model is as follows:
$\begin{aligned} T I E_{i t}= & \beta_{0}+\beta_{1} E R_{i t-2}+\beta_{2} R \& D_{i t-2}+ \\ & \beta_{3} \ln D A R_{i t-2}+\beta_{4} \ln T M_{i t-2}+ \\ & \beta_{5} T F_{i t-2}+\beta_{6} \ln G D P_{i t-2}+\alpha_{i}+\delta_{t-2}+\varepsilon_{i t-2} \end{aligned}$
where i and t represent province and time, respectively ($i=1,2,3, \cdots, 20 ; t=2017,2018,2019,2020,2021$);$\alpha_{i}$and $\delta_{t-2}$ indicate the enterprise effects and the time effects, respectively; $\varepsilon_{i t-2}$ is the idiosyncratic error term, and we assume that the errors ($\varepsilon_{i t-2}$) are independently and identically distributed; $\beta_{0}$ is a constant; and $\beta_{1}, \beta_{2}, \cdots, \beta_{6}$ are the coefficients of the variables on the right-hand side of the equation.
The efficiency of the technological innovation of ISs may be affected by the previous period. Therefore, based on Eq. (3), the lag phase of the technological innovation efficiency of ISs ($T_{I E}-1$) is introduced as an explanatory variable. The following equation can be obtained:
$\begin{aligned} T I E_{i t}= & \beta_{0}+\beta_{1} T I E_{i t-1}+\beta_{2} E R_{i t-2}+\beta_{3} R \& D_{i t-2}+ \\ & \beta 4 \ln D A R_{i t-2}+\beta_{5} \ln T M_{i t-2}+\beta_{6} T F_{i t-2}+ \\ & \beta_{7} \ln G D P_{i t-2}+\alpha_{i}+\delta_{t-2}+\varepsilon_{i t-2} \end{aligned}$
The moderation effect model is constructed on the basis of Baron and Kenny (1986). There are moderation effects of financing constraints and firm competitiveness on the association between environmental regulation and the technological innovation efficiency of ISs. That is,
$\begin{aligned} T I E_{i t}= & \beta 0+\beta_{1} E R_{i t-2}+\beta_{2} S A_{i t-2}+\beta_{3} R \& D_{i t-2}+ \\ & \beta 4 \ln D A R_{i t-2}+\beta_{5} \ln T M_{i t-2}+\beta 6 T F_{i t-2}+ \\ & \beta 7 \ln G D P_{i t-2}+\alpha_{i}+\delta_{t-2}+\varepsilon_{i t-2} \end{aligned}$
$\begin{aligned} T_{T E}^{i t}= & \beta_{0}+\beta_{1} E R_{i t-2}+\beta_{2 S} S A_{i t-2}+\beta_{3} R \& D_{i t-2}+ \\ & \beta 4 \ln D A R_{i t-2}+\beta_{5} \ln T M_{i t-2}+\beta 6 T F_{i t-2}+ \\ & \beta 7 \ln G D P_{i t-2}+\beta 8 E R_{i t-2} \times S A_{i t-2}+\alpha_{i}+ \\ & \delta_{t-2}+\varepsilon_{i t-2} \end{aligned}$
$\begin{aligned} T I E_{i t}= & \beta_{0}+\beta_{1} E R_{i t-2}+\beta_{2} M S_{i t-2}+\beta_{3} R \& D_{i t-2}+ \\ & \beta_{4} \ln D A R_{i t-2}+\beta_{5} \ln T M_{i t-2}+\beta_{6} T F_{i t-2}+ \\ & \beta_{7} \ln G D P_{i t-2}+\alpha_{i}+\delta_{t-2}+\varepsilon_{i t-2} \end{aligned}$
$\begin{aligned} T I E_{i t}= & \beta_{0}+\beta_{1} E R_{i t-2}+\beta_{2} M S_{i t-2}+\beta_{3} R \& D_{i t-2}+ \\ & \beta_{4} \ln D A R_{i t-2}+\beta 5 \ln T M_{i t-2}+\beta 6 T F_{i t-2}+ \\ & \beta_{7} \ln G D P_{i t-2}+\beta_{8} E R_{i t-2} \times M S_{i t-2}+ \\ & \alpha_{i}+\delta_{t-2}+\varepsilon_{i t-2} \end{aligned}$
where $E R_{i t-2} \times S A_{i t-2}$ and $E R_{i t-2} \times M S_{i t-2}$ represent the interaction effects of ER with SA and ER with MS, respectively.

3.2.2 The two-tier stochastic frontier model

On the basis of Eq. (3), we obtain $\varepsilon_{i t-2}$, and then, the two-sided effect of environmental regulations on the technological innovation efficiency of ISs can be decomposed via two-tier SFA. The two-tier stochastic frontier model has recently attracted increased interest in applied work (Parmeter, 2018). In this paper, the work of Kumbhakar and Parmeter (2009) and Song and Han (2022) is referenced to set up a two-tier SFA for the two-sided effect of environmental regulations on the technological innovation efficiency of ISs:
$\begin{aligned} T I E_{i t} & =i\left(x_{i t-2}\right)+E R_{i t-2}+\omega_{i t-2}+\mu_{i t-2}+\varepsilon_{i t-2} \\ & =i\left(x_{i t}\right)+\xi_{1 i t}=x_{i t-2}{ }^{\prime} \beta+E R_{i t-2}+\xi_{i t-2} \end{aligned}$
where i ( x i t − 2 ) represents the frontier level of the technological innovation efficiency of ISs and i ( x i t − 2 ) = x i t − 2 ' β in a fully competitive market. i ( x i t − 2 ) reflects the characteristics of interprovincial and interenterprise conditions ( x i t − 2 contains all control variables, i.e., the set of variables R&D, lnDAR, lnTM, TF, and lnGDP). β is the parameter vector that reflects the characteristics of interprovincial and interenterprise conditions. ξ i t − 2 is the compound residual error, i.e., ξ i t − 2 = ω i t − 2 − μ i t − 2 + ε i t − 2. ω i t − 2 is the upwards bias, and μ i t − 2 is the downwards bias, which reflects the positive or negative effects of environmental regulations on the optimal level of technological innovation efficiency. When ω i t − 2 ≥ 0, μ i t − 2 ≥ 0 is met. When ω i t − 2 = 0, environmental regulations only exert a negative effect on the technological innovation efficiency of ISs and this effect becomes a unilateral characteristic. Conversely, if μ i t − 2 = 0, environmental regulations only exerts a positive effect on technological innovation efficiency.
Because the OLS may have residual bias if it does not meet the assumption of an expected residual value of zero, the maximum likelihood estimation (MLE) method can be used to obtain a satisfactory result. Assume that ε i t − 2 is subject to a normal distribution with a mean value of zero and a variance of σ ε 2, i.e. ε i t − 2 ∼ i i d N ( 0 , σ ε 2 ), iid represents the independently and identically distributed. ω i t − 2 and μ i t − 2 are subject to an exponential distribution, i.e., ω i t − 2 ∼ i i d EXP ( σ ω , σ ω 2 ), u i t − 2 ∼ i i d EXP ( σ μ , σ μ 2 ), and the error terms are independent of each other and have no correlation with the other variables. Next, the probability density function for calculating the compound error term can be obtained as follows:
f ( ξ i t ) = ( σ ω + σ μ ) − 1 exp ( a i t ) Φ ( c i t ) + exp ( b i t ) ∫ − h ∞ ϕ ( z ) d ( z ) = ( σ ω + σ μ ) − 1 exp ( a i t ) Φ ( c i t ) + exp ( b i t ) ϕ ( h i t )
where Φ ( c i t ) and ϕ ( h i t ) are the probability density function and cumulative distribution function of the standard normal distribution, respectively. The partial parameters can be obtained as follows:
a i t = σ μ − 1 ξ i t + σ ε 2 ( 2 σ μ ) − 1 ; b i t = σ ω − 1 − ξ i t + σ ε 2 ( 2 σ ω ) − 1 ; h i t = ξ i t σ ε − 1 − σ ε σ ω − 1 ; c i t = − ξ i t σ ε − 1 − σ ε σ μ − 1
The log-likelihood function can be obtained when the sample contains n observations:
ln L ( X ; θ ) = − n ln ( σ ω + σ μ ) + ∑ i = 1 n ln e a i t Φ ( c i t ) + e b i t Φ ( h i t )
where θ = β , σ ε , σ μ , σ ω '. The conditional expectations of μ i t and ω i t are as follows:
f ( μ i t ξ i t ) = λ exp ( − λ μ i t ) Φ μ i t / σ ε + h i t ×   Φ ( h i t ) + exp ( a i t − b i t ) Φ ( c i t ) − 1 ; f ( ω i t ξ i t ) = λ exp ( − λ ω i t ) Φ ω i t / σ ε + c i t exp b i t − a i t − 1 ×   Φ ( h i t ) + exp ( a i t − b i t ) Φ ( c i t ) − 1
where λ = 1 / σ μ + 1 / σ ω.
The comprehensive (net) effect of environmental regulations on the technological innovation efficiency of ISs can be obtained by comparing the promotion and inhibition effects via Eq. (14):
N S = E 1 − e − w i t ξ i t − E 1 − e − μ i t ξ i t = E ( e − w i t − e − μ i t ξ i t )
Given that the parameter σ μ appears in both a i t and c i t, while σ ω appears in b i t and h i t, these parameters are identifiable. Through decomposition, the two-sided effect of environmental regulations on the technological innovation efficiency of ISs can be derived, which enables the calculation of net effects through the comparison of bilateral effects. The magnitude of bilateral effects is not predetermined by subjective assumptions but rather is entirely determined by the objective decomposition results of the model.

3.2.3 Bayesian quantile regression

As observed in Section 3.1.5, the data for the main variables do not follow a normal distribution. In such cases, the coefficient estimator of ordinary least squares (OLS) techniques becomes biased. Moreover, the OLS estimation only obtains the mean effect of environmental regulations on the technological innovation efficiency of ISs but cannot determine the impact of environmental regulations on technological innovation efficiency in different quantiles. Koenker and Bassett (1978) introduce the quantile regression (QR) model, which models quantiles as a function of predictors.
Although the large sample theory of QR has been extensively studied, there are computational deficiencies in its parameter estimation, and the model cannot accurately predict the posterior distributions of regression coefficients. Furthermore, QR does not possess parametric likelihoods. The BQR addresses the shortcomings of traditional quantile regression and uses a likelihood function based on an asymmetric Laplace distribution (Yu and Moyeed, 2001). QR vector coefficients exhibit a conditional conjugate property, which enables the construction of an efficient MCMC algorithm to fit the QR model. As a result, the conditional posterior value of the variable regression coefficient is updated, and the Gibbs sampling algorithm of the BQR (Kozumi and Kobayashi, 2011) is employed to analyse the heterogeneous effects of environmental regulations on the technological innovation efficiency of ISs. From this perspective, the BQR is more effective than OLS and QR are for the purpose of this study. The BQR not only provides the posterior distribution of explanatory variables during the Gibbs sampling fitting process but also enhances the reliability of the regression results. Furthermore, the standard deviation of the predicted value of the regression coefficient is significantly smaller than that of QR.
The ISs exhibiting different quantiles of technological innovation efficiency are divided into 5 groups, which are equivalent to the quantile τ set at the 5th, 25th, 50th, 75th, and 95th percentiles. The quantile function is constructed as follows:
Q T I E i t ( τ β ) = β 0 τ + β 1 τ E R i t − 2 + β 2 τ S A i t − 2 + β 3 τ R & D i t − 2 +   β 4 τ ln D A R i t − 2 + β 5 τ ln T M i t − 2 + β 6 τ T F i t − 2 +   β 7 τ ln G D P i t − 2 + ε i t − 2
Q T I E i t ( τ β ) = β 0 τ + β 1 τ E R i t − 2 + β 2 τ S A i t − 2 + β 3 τ R & D i t − 2 +   β 4 τ ln D A R i t − 2 + β 5 τ ln T M i t − 2 + β 6 τ T F i t − 2 +   β 7 τ ln G D P i t − 2 + β 8 τ E R i t − 2 × S A i t − 2 + ε i t − 2
Q T I E i t ( τ β ) = β 0 τ + β 1 τ E R i t − 2 + β 2 τ M S i t − 2 + β 3 τ R & D i t − 2 +   β 4 τ ln D A R i t − 2 + β 5 τ ln T M i t − 2 + β 6 τ T F i t − 2 +   β 7 τ ln G D P i t − 2 + ε i t − 2
Q T I E i t ( τ β ) = β 0 τ + β 1 τ E R i t − 2 + β 2 τ M S i t − 2 + β 3 τ R & D i t − 2 +   β 4 τ ln D A R i t − 2 + β 5 τ ln T M i t − 2 + β 6 τ T F i t − 2 +   β 7 τ ln G D P i t − 2 + β 8 τ E R i t − 2 × M S i t − 2 + ε i t − 2

3.2.4 Panel threshold regression

The traditional linear regression model can be used to examine only the stable relationships among the research variables and does not consider the nonlinear relationships among variables. Building upon the linear model, Hansen (1999) proposed a panel threshold regression (PTHR) model to construct a piecewise function of the explanatory variables under investigation. The PTHR can be used not only to estimate the threshold value but also to assess its validity and significance. Considering the potential threshold effect of financing constraints and firm competitiveness, environmental regulations may have a nonlinear effect on the technological innovation efficiency of ISs.
In this section, we utilise a PTHR to analyse the nonlinear relationship between environmental regulations and the technological innovation efficiency of ISs. The threshold model is presented as follows:
T I E i t = ϕ 0 + ϕ 1 E R i t − 2 × I ( S A i t − 2 ≤ θ 1 ) + ϕ 2 E R i t − 2 ×                   I ( θ 1 < S A i t − 2 < θ 2 ) + ϕ 3 E R i t − 2 ×                   I ( S A i t − 2 ≥ θ 2 ) + β ′ x i t − 2 + ε i t − 2
T I E i t = ϕ 0 + ϕ 1 E R i t − 2 × I ( M S i t − 2 ≤ θ 1 ) + ϕ 2 E R i t − 2 ×                   I ( θ 1 < M S i t − 2 < θ 2 ) + ϕ 3 E R i t − 2 ×                   I ( M S i t − 2 ≥ θ 2 ) + β ′ x i t − 2 + ε i t − 2
where I represents the indicator function, which takes a value of 1 if the corresponding condition is satisfied and 0 otherwise; θ 1 and θ 2 are the threshold values; and the relationship between θ 1 and θ 2 satisfies θ 1 < θ 2. The empirical framework of the study is shown in Fig. 3.
Fig. 3 Empirical framework of this study

4 Empirical results and discussion

4.1 Benchmark model and moderator selection results

The results of Eqs. (15)-(18) are shown in Table 3. First, the Hausman test is conducted, which reveals that using fixed-effects regression is more appropriate than using random-effects regression. Columns (1)-(2) present the estimated results of the fixed and random effects regressions, respectively. The results indicate a significant positive impact of environmental regulations. To address the issue of endogeneity in the model, a two-step SYS-GMM estimation of Eq. (4) is employed in this study. The results of the Sargan test support the null hypothesis that the instrumental variable is exogenous and confirm the reasonable selection of the instrumental variable. Hence, the estimation results from the two-step SYS-GMM are considered reliable. The estimate in Column (3) reveals that for every 1% increase in environmental regulations, the technological innovation efficiency of ISs is expected to increase by 0.0199%. This finding highlights the significant contribution of environmental regulations to enhancing the technological innovation efficiency of ISs. Column (4) presents the estimated result via pooled OLS. Although the coefficient of environmental regulations is small and nonsignificant, it merits consideration. Columns (5)-(6) present the estimation of the main effect of environmental regulations and the estimation of the moderating effect after incorporating the interaction term between environmental regulations and financing constraints. The coefficient of the interaction term between environmental regulation and financing constraints is positive. Financing constraints strengthen the positive impact of environmental regulation on the technological innovation efficiency of ISs. Columns (7)-(8) present the estimates of the moderating effect of firm competitiveness. The results show that the interaction between environmental regulation and firm competitiveness is positive. Similarly, firm competitiveness is believed to strengthen the positive effect of environmental regulations on the technological innovation efficiency of ISs.
Table 3 Results of the benchmark model and moderator selection
Model
variables		 FE RE Two-step SYS-GMM OLS OLS OLS OLS OLS VIF
TIE (1) TIE (2) TIE (3) TIE (4) TIE (5) TIE (6) TIE (7) TIE (8) -
L.TIE - - 0.0054 - - - - - -
- - (0.13) - - - - - -
ER 0.0196** 0.0182* 0.0199*** -0.0094 0.0293** -0.1963** 0.0358** -0.0435 2.36
(2.25) (1.91) (3.21) (-0.68) (2.12) (-2.39) (2.10) (-1.40) -
SA - - - - -0.0407*** -0.0588*** - - 3.93
- - - - (-4.29) (-5.18) - - -
MS - - - - - - -0.0588** -0.1077*** 4.02
- - - - - - (-2.55) (-3.89) -
R&D -0.0427** -0.0708*** -0.0237* -0.0975*** -0.1034*** -0.1020*** -0.0987*** -0.1034*** 1.29
(-2.07) (-4.12) (-1.69) (-10.23) (-11.30) (-11.37) (-10.52) (-11.14) -
lnDAR 0.0685 0.0014 0.1122*** -0.0248 0.0558* 0.0379 0.0003 0.0044 1.83
(1.36) (0.05) (3.55) (-0.87) (1.70) (1.15) (0.01) (0.15) -
lnTM 0.0584* 0.0468** -0.0304 0.0389*** 0.0520*** 0.0364*** 0.0487*** 0.0433*** 2.61
(1.88) (2.45) (-1.61) (2.94) (4.03) (2.63) (3.59) (3.25) -
TF -1.9504** -0.5489*** -0.3151 -0.2256 -0.3286* -0.0740 -0.3480* -0.1093 5.35
(-2.37) (-2.60) (-1.38) (-1.13) (-1.71) (-0.35) (-1.72) (-0.51) -
lnGDP 0.3911 0.0575 0.5158*** -0.0359 0.0268 0.0317 0.0358** 0.0327 2.86
(0.82) (1.30) (3.49) (-1.07) (0.84) (1.01) (1.09) (1.02) -
ER×SA - - - - - 0.0225*** - - -
- - - - - (2.79) - - -
ER×MS - - - - - - - 0.0266*** -
- - - - - - - (3.01) -
Constant -3.9394 -0.4100 -5.3571*** -0.0583 -0.0467 0.1685 -0.1643 -0.0945 -
(-0.80) (-0.99) (-3.49) (-0.17) (-0.14) (0.50) (-0.47) (-0.28) -
R2 0.2816 0.2274 - 0.4133 0.4747 0.4996 0.4367 0.4677 -
Hausman- χ2 test chi2(6) = 17.62*** - - - - - - -
[P = 0.0073] - - - - - - -
Sargan test - - chi2(9) = 13.1935 - - - - - -
- - [P = 0.1540] - - - - - -
N 165 165 132 165 165 165 165 165 165

Note: The prefix “ln” before the explanatory variables denotes that it takes the logarithmic form. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively. The figures displayed in () are the posterior standard errors of the coefficients, and those displayed in [] are the P values of the corresponding test statistics. L.TIE indicates the one-period lag of the dependent variable. Column (4) presents the estimated result using pooled OLS. Columns (5)-(6) represent the estimation of the main effect of environmental regulation and the estimation of the moderating effect after incorporating the interaction term.Columns (7)-(8) present the estimates of the moderating effect of firm competitiveness after incorporating the interaction term.

4.2 Two-sided effects of ER

To quantify the specific positive and negative effects of environmental regulations on the technological innovation efficiency of ISs, decomposing the estimated results and calculating the net effect are necessary. The results of the variance decomposition are presented in Table 4. The variance ratio of the promotion effect is 94.31%, whereas the variance ratio of the inhibition effect is only 5.69%. The average promotion effect of environmental regulations on the improvement of the technological innovation efficiency of ISs is 0.1401, which significantly surpasses the average inhibition effect of 0.0344. The net effect of environmental regulation son the technological innovation efficiency of ISs is E( ω – μ )= σ ω – σ μ =0.1057, which emphasises the positive impact of environmental regulations. Thus, the impact of environmental regulations on the technological innovation efficiency of ISs is two-fold and exerts a positive net effect, confirming Hypothesis 1b.
Table 4 Variance decomposition of the two-sided effects
Variance decomposition Definition Mathematical name ER decomposition results
The two-sided effects Promotion effect σ ω 0.1401
Inhibition effect σ μ 0.0344
Random error σ ν 0.0100
Variance decomposition Total variance of random error σ ν 2 + σ μ 2 + σ ω 2 0.0209
The proportion of the two-sided effects σ μ 2 + σ ω 2 / σ ν 2 + σ μ 2 + σ ω 2 0.9952
The proportion of the inhibition effect σ μ 2 / σ μ 2 + σ ω 2 0.0569
The proportion of the promotion effect σ ν 2 / σ μ 2 + σ ω 2 0.9431
The distribution of quantiles for the two-sided effects shown in Table 5 can reveal variations in their degrees of impact. The average inhibition effect of environmental regulation shows little variation between Q1, Q2, and Q3. Despite the increasing stringency of environmental regulations, its negative impact on technological innovation efficiency remains relatively low. This may be related to the size of the sample companies selected in this study. The selected ISs have high market value and large market share. One potential explanation is that small enterprises, owing to their cost and market power disadvantage, are significantly and negatively affected by excessively stringent environmental regulations, which can sometimes lead to their exit from the market. Large enterprises have relatively more resources, which makes it easier for them to share costs to reduce the crowding out of funds in innovation activities (Li et al., 2022). Owing to economies of scale, as firms grow in size, the cost allocation within firms follows a downwards-sloping long-run average cost curve (Bernard et al., 2010).
Table 5 Proportion and quantile distributions of the effects
Core explanatory variable Effects Number Mean SD Min Q(0.25) Median Q(0.75) Max
ER Promotion effect 165 14.01 15.28 2.76 3.82 8.51 17.98 81.88
Inhibition effect 165 3.44 1.89 2.76 2.76 2.76 2.83 14.08
Net effect 165 10.57 15.89 -11.31 0.99 5.75 15.22 79.12

Note: % refers to taking the percentiles of mean, SD, and Q. The same below.

In other words, large companies can mitigate the adverse effects of environmental regulation to some extent. In contrast, the promotion effects of Q(0.25), median, and Q(0.75) exhibit notable disparities, with the magnitude of the effect sequentially increasing. This suggests that the uncertainty in the net effect of environmental regulations stems primarily from the innovation compensation effect.
Table 6 presents the two-sided and net effects of environmental regulations on the technological innovation efficiency of ISs for the period of 2017-2021. With respect to time-varying characteristics, the net effects of environmental regulations on the technological innovation efficiency of industrial enterprises can be categorised into three phases: 2017-2018, 2019, and 2020-2021. The first phase has a positive net effect, the second phase has a negative net effect, and the third phase has a positive net effect. Overall, these changes are closely tied to the macroeconomic environment.
Table 6 Annual variation characteristics of the two-sided effects of ER on TIE
Year Effects Number Mean SD
2017 Promotion effect 33 13.17 14.37
Inhibition effect 33 5.73 3.66
Net effect 33 7.44 15.90
2018 Promotion effect 33 10.60 11.50
Inhibition effect 33 5.99 4.07
Net effect 33 4.61 13.37
2019 Promotion effect 33 7.04 6.11
Inhibition effect 33 8.77 7.60
Net effect 33 -1.73 11.24
2020 Promotion effect 33 13.06 12.33
Inhibition effect 33 1.00 0.19
Net effect 33 12.06 12.39
2021 Promotion effect 33 14.48 18.38
Inhibition effect 33 4.99 3.16
Net effect 33 9.49 19.40
The transition from a positive to a negative net effect that occurred from 2017 to 2019 can be attributed to production stoppages and capacity restrictions in the iron and steel industry. The positive net effect observed from 2020-2021 is closely related to China’s implementation of independent innovation and new infrastructure measures during the COVID-19 period. The Chinese iron and steel industry had strengthened its level of innovation, resulting in the emergence of new steel varieties or new technical equipment from steel enterprises, such as Baoshan Iron & Steel Co. Ltd., Hbis Company Limited, Pangang Group Vanadium & Titanium Resources Co. Ltd., and Shanxi Taigang Stainless Steel Co. Ltd.
Table 7 presents the two-sided and net effects of environmental regulations across different geographic regions. Clearly, there are variations in the inhibition and promotion effects of ISs across geographic regions, but the net effects are consistently positive. More specifically, the positive and negative effects of environmental regulations on the technological innovation efficiency of ISs in coastal provinces are smaller than those in inland provinces, whereas the net effect in coastal provinces is greater than that in inland provinces. The net effect varies from largest to smallest in the following order: central enterprises, western enterprises, and eastern enterprises. This suggests that the net effect of environmental regulation on the technological innovation efficiency of ISs is not solely determined by the level of regional economic growth. Rather, it is also influenced by industrial transfer policies and regional resource endowments; this underscores the importance of implementing environmental regulations while considering the variations in regional capabilities.
Table 7 Variation characteristics of the two-sided effects of ER on TIE by different regions
Different regions Effects Number Mean SD
Enterprises in coastal provinces Promotion effect 95 14.42 14.04
Inhibition effect 95 2.65 1.42
Net effect 95 11.77 14.46
Enterprises in inland provinces Promotion effect 70 14.67 16.85
Inhibition effect 70 3.89 2.16
Net effect 70 10.78 17.54
Enterprises in eastern provinces Promotion effect 95 12.89 14.03
Inhibition effect 95 3.50 2.04
Net effect 95 9.39 14.71
Enterprises in central provinces Promotion effect 30 15.71 16.27
Inhibition effect 30 4.67 0.12
Net effect 30 11.04 16.32
Enterprises in western provinces Promotion effect 40 12.44 12.24
Inhibition effect 40 1.80 0.20
Net effect 40 10.64 12.31
Table 8 presents the effect disparity between state-owned enterprises and nonstate-owned enterprises. The net effect is slightly lower for state-owned enterprises than for nonstate-owned enterprises. Thus, the positive impact of environmental regulation on enhancing the technological innovation efficiency of ISs is suggested to be somewhat more evident in nonstate-owned enterprises than in state-owned enterprises.
Table 8 Variation characteristics of the two-sided effects of ER on TIE based on property rights
Property right nature Effects Number Mean SD
State-owned enterprises Promotion effect 115 13.10 14.68
Inhibition effect 115 1.49 0.37
Net effect 115 11.61 14.79
Nonstate-owned enterprises Promotion effect 50 12.74 12.74
Inhibition effect 50 1.00 0.11
Net effect 50 11.74 12.79

4.3 The heterogeneous effects of ER and the moderating effects of SA and MS

To estimate the parameters presented in Eq. (10)-(13) via BQR, 11000 Gibbs iterations are conducted for this study, and the initial 1000 iterations are discarded. During the Gibbs sampling process, the sample regression coefficients can be derived from the posterior distribution of each parameter. Given that the results of the final 10000 iterations are stable and converge towards the ergodic mean of the regression coefficients, the mean value of the samples can be employed as estimates for the model parameters. Compared with the QR, the BQR results in a significantly lower posterior standard deviation for the coefficient estimates, indicating that the estimated results of the BQR are more reliable (refer to Appendix D for the estimation results of the traditional QR).
Table 9 lists (9)-(13) the main effects of environmental regulations on the technological innovation efficiency of ISs, whereas (14)-(18) indicate whether the impact of environmental regulations on the technological innovation efficiency of ISs is influenced by financing constraints. The coefficients of environmental regulations are positively and significantly correlated in all observed main-effect quantile regressions. That is, environmental regulations have a positive effect on technological innovation efficiency, which is consistent with the estimation results of Zhao et al. (2015) and Chakraborty and Chatterjee (2017). Under different quantiles of the technological innovation efficiency of ISs, there are significant differences in the coefficients. Thus, environmental regulations have significant heterogeneous effects on the technological innovation efficiency of ISs. Specifically, the largest coefficient is Q(0.95)=0.0504; this implies that a one-unit increase in ER would result in a 5.04% increase in per unit TIE at the 95th quantile. The coefficient estimates for the other quartiles are Q(0.05)= 0.0156, Q(0.25)=0.0144, Q(0.50)=0.0228, and Q(0.75)= 0.0287. These results indicate that environmental regulations have more pronounced effects on the technological innovation efficiency of ISs that already exhibit higher levels of technological innovation efficiency. With increasing quantiles, the value of the coefficient of financing constraints gradually decreases, but the significance tends to increase. This finding indicates that the negative impact of financing constraints is greater in ISs with higher technological innovation efficiency. The observed quantiles demonstrate a positive and statistically significant impact of the interaction term between environmental regulations and financing constraints on the technological innovation efficiency of ISs. In conjunction with the main effects results, financing constraints can be seen to exert a positive moderating influence. In addition, the coefficient of the interaction term increases with increasing quantiles; this shows that the moderating role of financing constraints is more prominent in ISs with higher technological innovation efficiency. One possible explanation for this is that certain firms face limitations in their R&D investments due to short-term financial constraints. They are “locked at the low end” of the market due to a lack of core technological competitiveness, and they subsequently only pursue increments in production (Fan et al., 2015). While innovation investment can directly intensify the financing constraints of enterprises, a strong innovation compensation effect as a result of technological innovation can alleviate these constraints and enable enterprises to obtain additional capital and market advantages (He et al., 2021), thus highlighting the synergistic effect of financing constraints and environmental regulations.
Table 9 Bayesian quantile estimates of ER and the moderating effects of SA
Model
variables		 (9) (10) (11) (12) (13) (14) (15) (16) (17) (18)
TIE-Q(0.05) TIE-Q(0.25) TIE-Q(0.50) TIE-Q(0.75) TIE-Q(0.95) TIE-Q(0.05) TIE-Q(0.25) TIE-Q(0.50) TIE-Q(0.75) TIE-Q(0.95)
ER 0.0156** 0.0144* 0.0228* 0.0287* 0.0504*** -0.0433 -0.0910** -0.1637*** -0.2365*** -0.2620***
(0.0001) (0.0001) (0.0001) (0.0002) (0.0002) (0.0003) (0.0004) (0.0006) (0.0007) (0.0009)
SA 0.0008 -0.0055 -0.0277*** -0.0475*** -0.0779*** -0.0062 -0.0156*** -0.0421*** -0.0704*** -0.1060***
(0.0000) (0.0000) (0.0001) (0.0001) (0.0002) (0.0000) (0.0001) (0.0001) (0.0001) (0.0002)
R&D -0.0136*** -0.0377*** -0.0844*** -0.1090*** -0.1355*** -0.0121*** -0.0363*** -0.0772*** -0.1034*** -0.1396***
(0.0000) (0.0001) (0.0001) (0.0001) (0.0001) (0.0000) (0.0001) (0.0001) (0.0001) (0.0001)
lnDAR 0.0005 0.0104 0.0393* 0.0891*** 0.1712*** -0.0024 -0.0026 0.0187 0.0686** 0.1561***
(0.0001) (0.0001) (0.0002) (0.0003) (0.0006) (0.0001) (0.0001) (0.0002) (0.0003) (0.0005)
lnTM 0.0047 0.0098* 0.0433*** 0.0685*** 0.0989*** -0.0026 0.0015 0.0212 0.0434*** 0.0800***
(0.0000) (0.0001) (0.0001) (0.0001) (0.0002) (0.0000) (0.0001) (0.0001) (0.0001) (0.0002)
TF 0.0125 0.0289 -0.3849** -0.3373* -0.0318 0.0962 0.1393 -0.0954 -0.1132 0.4022
(0.0005) (0.0010) (0.0019) (0.0020) (0.0028) (0.0006) (0.0010) (0.0021) (0.0021) (0.0029)
lnGDP -0.0005 0.0024 0.0462 0.0011 -0.1656*** -0.0053 -0.0006 0.0384 0.0253 -0.2075***
(0.0001) (0.0002) (0.0003) (0.0003) (0.0006) (0.0001) (0.0002) (0.0003) (0.0003) (0.0005)
ER*SA - - - - - 0.0060** 0.0111*** 0.0193*** 0.0286*** 0.0305***
- - - - - (0.0000) (0.0000) (0.0001) (0.0001) (0.0001)
Constant 0.0067 0.0394 -0.2817 0.1293 1.7505*** 0.1457 0.2299 0.0483 0.2314 2.5217***
(0.0009) (0.0017) (0.0030) (0.0031) (0.0068) (0.0012) (0.0018) (0.0031) (0.0028) (0.0058)

Note: The prefix “ln” before the explanatory variables denotes that it takes the logarithmic form. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively. The figures displayed in () are the posterior standard errors of the coefficients. (9)-(13) present the impact of environmental regulation on technical efficiency under financing constraints at all the observed quantiles. (14)-(18) add the interaction effects between environmental regulations and financing constraints.

With respect to the main effect (Table 10), firm competitiveness has a significantly negative effect on ISs with high technological innovation efficiency. The interaction coefficient between environmental regulations and firm competitiveness is not statistically significant at the 5th quantile, but it becomes positive and statistically significant at the 25th quantile and above. Specifically, the largest coefficient is Q(0.95)= 0.0420, which indicates that a one-unit increase in the interaction between environmental regulations and firm competitiveness leads to a 4.20% increase in the per unit technological innovation efficiency of ISs at the 95th quantile. The estimated coefficients at the other quantiles are as follows: Q(0.05)=0.0042, Q(0.25)=0.0095, Q(0.50)=0.0211, and Q(0.75)=0.0392. These results indicate that the interaction between environmental regulations and firm competitiveness is stronger in ISs that already exhibit higher levels of technological innovation efficiency. Aghion et al. (2005) explain the positive impact of firm competitiveness as arising from the “escape competition effect”. Furthermore, the influence of firm competitiveness is associated with firm size, with a more pronounced effect on larger firms (Askenazy et al., 2013). In summary, the positive impact of environmental regulations on the technological innovation efficiency of ISs is strengthened with the improvement of firm competitiveness. Consequently, H2 and H3 are supported.
Table 10 Bayesian quantile estimates of ER and the moderating effects of MS
Model
variables		 (19) (20) (21) (22) (23) (24) (25) (26) (27) (28)
TIE-Q(0.05) TIE-Q(0.25) TIE-Q(0.50) TIE-Q(0.75) TIE-Q(0.95) TIE-Q(0.05) TIE-Q(0.25) TIE-Q(0.50) TIE-Q(0.75) TIE-Q(0.95)
ER 0.0153** 0.0142 0.0230 0.0364** 0.0604*** 0.0022 -0.0099 -0.0287 -0.0713*** -0.0632**
(0.0001) (0.0001) (0.0001) (0.0002) (0.0002) (0.0001) (0.0002) (0.0002) (0.0003) (0.0003)
MS 0.0003 -0.0047 -0.0272 -0.0708*** -0.0913*** -0.0085 -0.0258 -0.0577** -0.1173*** -0.1624***
(0.0001) (0.0001) (0.0002) (0.0002) (0.0001) (0.0001) (0.0002) (0.0003) (0.0002) (0.0003)
R&D -0.0136*** -0.0346*** -0.0716*** -0.1041*** -0.1423*** -0.0131*** -0.0373*** -0.0725*** -0.1086*** -0.1410
(0.0000) (0.0001) (0.0001) (0.0001) (0.0004) (0.0000) (0.0001) (0.0001) (0.0001) (0.0001)
lnDAR 0.0023 0.0022 0.0036 0.0128 0.0443 0.0006 -0.0015 0.0001 0.0037 0.0477
(0.0001) (0.0001) (0.0002) (0.0003) (0.0002) (0.0001) (0.0001) (0.0002) (0.0003) (0.0003)
lnTM 0.0043 0.0092 0.0337 0.0640*** 0.0966*** 0.0012 0.0068 0.0285** 0.0509*** 0.0910***
(0.0000) (0.0001) (0.0001) (0.0001) (0.0002) (0.0000) (0.0001) (0.0001) (0.0001) (0.0002)
TF 0.0173 0.0453 -0.2536 -0.3450* -0.3626 0.0607 0.1108 -0.1125 -0.1133 0.1075
(0.0005) (0.0010) (0.0017 (0.0020) (0.0032) (0.0006) (0.0011) (0.0019) (0.0019) (0.0026)
lnGDP -0.0013 0.0015 0.0367 0.0041 -0.0992 -0.0044 -0.0022 0.0341 0.0271 -0.1581***
(0.0001) (0.0002) (0.0003) (0.0003) (0.0007) (0.0001) (0.0002) (0.0003) (0.0003) (0.0005)
ER*MS - - - - - 0.0042 0.0095* 0.0211** 0.0392*** 0.0420***
- - - - - (0.0000) (0.0001) (0.0001) (0.0001) (0.0001)
Constant 0.0147 0.0369 -0.2399 0.1167 0.0604 0.0706 0.1195 -0.1561 0.0251 1.7959***
(0.0009) (0.0017) (0.0029) (0.0035) (0.0078) (0.0010) (0.0018) (0.0030) (0.0031) (0.0059)

Note: The prefix “ln” before the explanatory variables denotes that it takes the logarithmic form. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively. The figures displayed in () are the posterior standard errors of the coefficients. (19)-(23) present the impact of environmental regulation on technical efficiency under firm competitiveness at all the observed quantiles. (24)-(28) add the interaction effects between environmental regulations and firm competitiveness.

4.4 Threshold effects between ER and TIE

The results of the threshold effect test are shown in Table 11. A dual-threshold effect of financing constraints is observed, with threshold values of 7.8473 and 8.4537. Furthermore, a dual-threshold effect of firm competitiveness is observed, with thresholds of 0.2595 and 0.3336. These thresholds successfully pass the F statistic test.
Table 11 Threshold effects test
Threshold variable Regime
variable		 Threshold type BS F statistics P value The existence of the threshold Threshold value 95% threshold confidence interval
Lower Higher
SA Single 300 14.40 0.0933 Yes 8.4537 8.4231 8.6278
ER Double 300 25.17 0.0133 Yes 7.8473 7.6620 7.9461
Triple 300 8.61 0.8367 No 5.7277 5.6341 5.8174
MS Single 300 25.55 0.0067 Yes 0.3336 0.3172 0.3478
ER Double 300 11.96 0.0800 Yes 0.2595 0.2333 0.2832
Triple 300 9.44 0.1333 No 0.1869 0.1732 0.1923

Note: The regime variable is the core explanatory variable that is affected by the threshold variable. BS represents the number of times to bootstrap.

The regression coefficients for the variables are presented in Table 12. Our findings reveal a significant positive impact of environmental regulations on the technological in-novation efficiency of ISs within the range of 7.8473<SA< 8.4537. However, when SA≤7.8473 and SA≥8.4537, the coefficient of the effects of environmental regulations on ISs’ technological innovation efficiency approaches zero and fails to achieve statistical significance. This suggests that maintaining financing constraints within a specific range can effectively guide the promotion of environmental regulations to enhance ISs’ technological innovation efficiency. This result aligns with the research of Lin et al. (2022), which validated the role of financing constraints in augmenting enterprise innovation efficiency.
Table 12 Results of the double threshold regression
Variables (29) Variables (30)
ER (SA≤7.8473) 0.0151 ER (MS≤0.2595) 0.0829**
(0.33) (1.97)
ER (7.8473<SA<8.4537) 0.3275*** ER (0.2595<MS<0.3336) 0.2567***
(5.83) (5.52)
ER (SA≥8.4537) 0.0159 ER (MS≥0.3336) 0.0121
(1.10) (0.85)
R&D -0.0336*** R&D -0.0382***
(-2.78) (-3.14)
lnDAR 0.0360 lnDAR 0.0423
(0.78) (0.91)
lnTM 0.0397 lnTM 0.0543*
(1.37) (1.88)
TF -1.0471** TF -0.9427*
(-2.12) (-1.87)
lnGDP 0.4311** lnGDP 0.4104*
(1.96) (1.83)
Constant -4.3792** Constant -4.2785*
(-1.96) (-1.88)
R2 0.3568 R2 0.3490

Note: The prefix “ln” before the explanatory variables denotes that it takes the logarithmic form. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively. The figures displayed in () are the t values of the coefficients.

Furthermore, when MS≤0.2595 and 0.2595<MS< 0.3336, environmental regulations have significant positive effects on the technological innovation efficiency of ISs, with more pronounced marginal effects observed in the latter interval. However, when MS≥0.3336, the coefficient of environmental regulations diminishes and fails to achieve statistical significance. This can be attributed to fierce competition, which can discourage collaboration and knowledge sharing among firms, thus impeding the realisation of the Porter hypothesis (Porter and Linde, 1995; de Bettignies et al., 2023).
The threshold effects exhibit both temporal and spatial heterogeneity, thus leading to variations in the distribution of enterprises across different years. Table 13 provides an overview of the stock codes and corresponding durations for the different threshold intervals related to financing constraints and firm competitiveness. The majority of ISs encounter financing constraints within the lower range, accounting for 69.09% of the total observations. Conversely, a small proportion of enterprises, representing 14.55% of the total observations, experience strengthened effects attributable to financing constraints. This suggests that the potential of financing constraints to augment the impact of environmental regulation on the technological innovation efficiency of ISs is limited. Notably, close to half of the enterprises, comprising 49.00% of the total observations, present a market share below the threshold of 0.3336. This finding underscores the substantial enhancement in the impact of environmental regulations on technological innovation efficiency through the advancement of firm competitiveness in these specific instances.
Table 13 The specific number of enterprises and years in different threshold intervals
Threshold interval Enterprises stock code Sum
SA≤7.8473 002756(5), 002478(5), 002443(5), 002318(5), 002110(5), 002075(5), 000923(5), 000778(5), 603878(5), 601005(5), 601003(5), 600581(5), 600569(5), 600516(5), 600507(5), 600399(5), 600307(5), 600295(5), 600282(5), 600231(5), 600126(5), 000708(4), 000629(4), 000761(1) 114
7.8473<SA<8.4537 600808(5), 000932(5), 000825(5), 600022(5), 000761(4), 000959(1), 000708(1), 000629(1) 27
SA≥8.4537 600019(5), 000709(5), 600010(5), 000898(5), 000959(4) 24
MS≤0.2595 002756(5), 002478(5), 002443(5), 002318(5), 000923(5), 000629(5), 603978(5), 600516(5), 600507(5), 600399(5), 002075(4), 000708(4), 601005(3), 600581(3), 002110(2), 600231(2), 600295(1), 600126(1) 71
0.2595<MS<0.3336 600231(3), 600581(2), 600295(2), 000959(1), 600569(1), 600126(1) 10
MS≥0.3336 000932(5), 000898(5), 000825(5), 000778(5), 000761(5), 000709(5), 601003(5), 000761(5), 000709(5), 601003(5), 600808(5), 600307(5), 600282(5), 600022(5), 600019(5), 600010(5), 600569(4), 002110(3), 000959(3), 600126(3), 601005(2), 600295(2), 000708(1) 84

Note: The enterprise name is replaced by a stock code. The figures displayed in () are the number of years for the enterprise in each threshold interval.

5 Conclusions and policy implications

The net effect of environmental regulations is obtained by comparing the two-sided effects of environmental regulations on the technological innovation efficiency of iron and steel enterprises. With the increase in the quantiles of the technological innovation efficiency of iron and steel enterprises, heterogeneous effects of environmental regulations on technological innovation efficiency can be observed. The impact of environmental regulations on the technological innovation efficiency of iron and steel enterprises presents nonlinear characteristics. Therefore, by using two-tier stochastic frontier analysis and the Bayesian quantile regression model, both the two-sided and the heterogeneous effects of environmental regulations on the technological innovation efficiency of iron and steel enterprises are examined. The panel threshold regression model is further applied to explore whether the moderating mechanism of financing constraints and firm competitiveness can trigger the nonlinear effects of environmental regulations on technological innovation efficiency.
This research provides valuable results. First, although environmental regulations incentivise the improvement of the technological innovation efficiency of iron and steel enterprises overall, there are many different incentive effects in various years, regions, and types of property rights. In terms of the average effect, the promotion effect of environmental regulation on the technological innovation efficiency of iron and steel enterprises is 14.01%, and the inhibition effect is 3.44%, resulting in a net effect of 10.57%. Annually, during the 2017-2018 period, environmental regulations played a role in promoting the improvement of the technological innovation efficiency of iron and steel enterprises; it was inhibited in 2019, but it played a promoting role from 2020-2021. From a regional perspective, the incentive effect of environmental regulations in coastal areas is greater than that in inland areas; the incentive effect in the central area is the largest, and the incentive effect in the western area is the smallest. In terms of the nature of property rights, the incentive effect of environmental regulations on state-owned enterprises is slightly lower than that on nonstate-owned enterprises. Second, with the increase in the technological innovation efficiency of iron and steel enterprises, the incentive effect of environmental regulations on the improvement of technological innovation efficiency is characterised by an increasing marginal effect. Moreover, the enhancement of financing constraints and firm competitiveness can strengthen the effect of environmental regulations. Third, the incentive effect of environmental regulations on the improvement of the technological innovation efficiency of iron and steel enterprises has a certain inherent premise, and its specific effect is moderated by financing constraints and firm competitiveness. More precisely, the incentive effect of environmental regulations on the enhancement of technological innovation efficiency is manifested within a specific range of financing constraints, which specifically range from 7.8473 to 8.4537. Likewise, a significant incentive effect of environmental regulation on the enhancement of technological innovation efficiency can be observed when the market share of iron and steel enterprises falls below 0.3336.
The policy implications are clear and straightforward. Provinces should tailor energy-saving and emission reduction measures, set environmental investment standards and compensate for incurred environmental costs. To counterbalance misallocated R&D investments, the central government could establish funds for pollution control and SO2 emission reduction. Furthermore, linking corporate environmental reputation to creditworthiness and applying R&D achievements is crucial. Strengthening supervision will promote reputational competition and survival of the fittest, emphasising optimal R&D resource allocation. Finally, optimising financing conditions and fostering competitive environments for these enterprises, possibly through digitalisation and lessening institutional constraints on technology, can exploit environmental regulations as a catalyst for innovation, thus increasing technological innovation efficiency.

References

Publishing order | Descend order by publishing year | Descend order by cited within
[1]
Aghion P, Bloom N, Blundell R, et al. 2005. Competition and innovation: An inverted-U relationship. The Quarterly Journal of Economics, 120(2): 701-728.

[2]
Andersen D C. 2017. Do credit constraints favor dirty production? Theory and plant-level evidence. Journal of Environmental Economics and Management, 84: 189-208.

[3]
Anwar M. 2018. Business model innovation and SMEs performance— Does competitive advantage mediate? International Journal of Innovation Management, 22(7): 1850057. DOI: 10.1142/s1363919618500573.

[4]
Arsawan I W E, Koval V, Rajiani I, et al. 2022. Leveraging knowledge sharing and innovation culture into SMEs sustainable competitive advantage. International Journal of Productivity and Performance Management, 71(2): 405-428.

[5]
Askenazy P, Cahn C, Irac D. 2013. Competition, R&D, and the cost of innovation: Evidence for France. Oxford Economic Papers, 65(2): 293-311.

[6]
Baron R M, Kenny D A. 1986. The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51(6): 1173-1182.

DOI PMID

[7]
Bernard A B, Redding S J, Schott P K. 2010. Multiple-product firms and product switching. American Economic Review, 100(1): 70-97.

[8]
Blackman A. 2010. Alternative pollution control policies in developing countries. Review of Environmental Economics and Policy, 4(2): 234-253.

[9]
Cai X, Zhu B Z, Zhang H J, et al. 2020. Can direct environmental regulation promote green technology innovation in heavily polluting industries? Evidence from Chinese listed companies. Science of the Total Environment, 746: 140810. DOI: 10.5194/esd-2020-67-rc1.

[10]
Chakraborty P, Chatterjee C. 2017. Does environmental regulation indirectly induce upstream innovation? New evidence from India. Research Policy, 46(5): 939-955.

[11]
Cheng Z H, Kong S Y. 2022. The effect of environmental regulation on green total-factor productivity in China’s industry. Environmental Impact Assessment Review, 94: 106757. DOI: 10.1016/j.eiar.2022.106757.

[12]
de Bettignies J E, Liu H F, Robinson D T, et al. 2023. Competition and innovation in markets for technology. Management Science, 69(8): 4753-4773.

[13]
Fan H C, Lai E L C, Li Y A. 2015. Credit constraints, quality, and export prices: Theory and evidence from China. Journal of Comparative Economics, 43(2): 390-416.

[14]
Fang J Y, Gao C, Lai M Y. 2020. Environmental regulation and firm innovation: Evidence from National Specially Monitored Firms program in China. Journal of Cleaner Production, 271: 122599. DOI: 10.1016/j.jclepro.2020.122599.

[15]
Hadlock C J, Pierce J R. 2010. New evidence on measuring financial constraints: Moving beyond the KZ index. The Review of Financial Studies. 23( 5): 1909-1940.

[16]
Hansen B E. 1999. Threshold effects in non-dynamic panels: Estimation, testing, and inference. Journal of Econometrics, 93(2): 345-368.

[17]
Hansen M T, Birkinshaw J. 2007. The innovation value chain. Harvard Business Review, 85(6): 121-30, 142.

PMID

[18]
Hardin G. 1968. The tragedy of the commons: The population problem has no technical solution; it requires a fundamental extension in morality. Science, 162(3859): 1243-1248.

PMID

[19]
He J J, Tian X. 2013. The dark side of analyst coverage: The case of innovation. Journal of Financial Economics, 109(3): 856-878.

[20]
He Y Q, Ding X, Yang C C. 2021. Do environmental regulations and financial constraints stimulate corporate technological innovation? Evidence from China. Journal of Asian Economics, 72: 101265. DOI: 10.1016/j.asieco.2020.101265.

[21]
Hermundsdottir F, Aspelund A. 2021. Sustainability innovations and firm competitiveness: A review. Journal of Cleaner Production, 280: 124715. DOI: 10.1016/j.jclepro.2020.124715.

[22]
Horbach J, Rammer C, Rennings K. 2012. Determinants of eco-innovations by type of environmental impact—The role of regulatory push/pull, technology push and market pull. Ecological Economics, 78: 112-122.

[23]
Hottenrott H, Peters B. 2012. Innovative capability and financing constraints for innovation: More money, more innovation? Review of Economics and Statistics, 94(4): 1126-1142.

[24]
Huang J C, Zhao J, Cao J E. 2021. Environmental regulation and corporate R&D investment—Evidence from a quasi-natural experiment. International Review of Economics & Finance, 72: 154-174.

[25]
Jensen M C, Meckling W H. 1976. Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3(4): 305-360.

[26]
Kao C A, Hwang S N. 2008. Efficiency decomposition in two-stage data envelopment analysis: An application to non-life insurance companies in Taiwan. European Journal of Operational Research, 185(1): 418-429.

[27]
Koenker R, Bassett Jr G. 1978. Regression quantiles. Econometrica, 46(1): 33-50.

[28]
Kozumi H, Kobayashi G. 2011. Gibbs sampling methods for Bayesian quantile regression. Journal of Statistical Computation and Simulation, 81(11): 1565-1578.

[29]
Kumbhakar S C, Parmeter C F. 2009. The effects of match uncertainty and bargaining on labor market outcomes: Evidence from firm and worker specific estimates. Journal of Productivity Analysis, 31(1): 1-14.

[30]
Li X Z, Du K R, Ouyang X L, et al. 2022. Does more stringent environmental regulation induce firms’ innovation? Evidence from the 11th Five-year Plan in China. Energy Economics, 112: 106110. DOI: 10.1016/s0195-6663(22)00201-x.

[31]
Lin B Q, Ma R Y. 2022. Green technology innovations, urban innovation environment and CO2 emission reduction in China: Fresh evidence from a partially linear functional-coefficient panel model. Technological Forecasting and Social Change, 176: 121434. DOI: 10.1016/j.techfore.2021.121434.

[32]
Lin X W, Zhang Q H, Chen A H, et al. 2022. The bright side of financial constraint on corporate innovation: Evidence from the Chinese bond market. Finance Research Letters, 49: 103098. DOI: 10.1016/j.frl.2022.103098.

[33]
Liu X, Liu F Z. 2022. Environmental regulation and corporate financial asset allocation: A natural experiment from the new environmental protection law in China. Finance Research Letters, 47: 102974. DOI: 10.1016/j.frl.2022.102974.

[34]
Liu Y L, Li Z H, Yin X M. 2018. Environmental regulation, technological innovation and energy consumption—A cross-region analysis in China. Journal of Cleaner Production, 203: 885-897.

[35]
Liu Z, Davis S J, Feng K, et al. 2016. Targeted opportunities to address the climate-trade dilemma in China. Nature Climate Change, 6(2): 201-206.

DOI

[36]
Lu W X, Wu H C, Geng S S. 2021. Heterogeneity and threshold effects of environmental regulation on health expenditure: Considering the mediating role of environmental pollution. Journal of Environmental Management, 297: 113276. DOI: 10.1016/j.jenvman.2021.113276.

[37]
Luo M. 2011. A bright side of financial constraints in cash management. Journal of Corporate Finance, 17(5): 1430-1444.

[38]
Mele M, Magazzino C. 2020. A machine learning analysis of the relationship among iron and steel industries, air pollution, and economic growth in China. Journal of Cleaner Production, 277: 123293. DOI: 10.1016/j.jclepro.2020.123293.

[39]
Ouyang X L, Li Q, Du K R. 2020. How does environmental regulation promote technological innovations in the industrial sector? Evidence from Chinese provincial panel data. Energy Policy, 139: 111310. DOI: 10.1016/j.mad.2020.111310.

[40]
Palmer K, Oates W E, Portney P R. 1995. Tightening environmental standards: The benefit-cost or the no-cost paradigm? Journal of Economic Perspectives, 9(4): 119-132.

[41]
Palmié M, Wincent J, Parida V, et al. 2020. The evolution of the financial technology ecosystem: An introduction and agenda for future research on disruptive innovations in ecosystems. Technological Forecasting and Social Change, 151: 119779. DOI: 10.1016/j.techfore.2019.119779.

[42]
Parmeter C F. 2018. Estimation of the two-tiered stochastic frontier model with the scaling property. Journal of Productivity Analysis, 49(1): 37-47.

[43]
Porter M E, Linde C. 1995. Toward a new conception of the environment-competitiveness relationship. Journal of Economic Perspectives, 9(4): 97-118.

[44]
Qian X Y, Wang D, Wang J, et al. 2021. Resource curse, environmental regulation and transformation of coal-mining cities in China. Resources Policy, 74: 101447. DOI: 10.1016/j.resourpol.2019.101447.

[45]
Ramanathan R, Ramanathan U, Bentley Y. 2018. The debate on flexibility of environmental regulations, innovation capabilities and financial performance—A novel use of DEA. Omega, 75: 131-138.

[46]
Shen N, Liao H L, Deng R M, et al. 2019. Different types of environmental regulations and the heterogeneous influence on the environmental total factor productivity: Empirical analysis of China’s industry. Journal of Cleaner Production, 211: 171-184.

[47]
Shen T, Chen H H, Zhao D H, et al. 2022. Examining the impact of environment regulatory and resource endowment on technology innovation efficiency: From the microdata of Chinese renewable energy enterprises. Energy Reports, 8: 3919-3929.

[48]
Song W F, Han X F. 2022. Heterogeneous two-sided effects of different types of environmental regulations on carbon productivity in China. Science of the Total Environment, 841: 156769. DOI: 10.1016/j.scitotenv.2022.156769.

[49]
Su Y, Xu G. 2022. Low-carbon transformation of natural resource industry in China: Determinants and policy implications to achieve COP26 targets. Resources Policy, 79: 103082. DOI: 10.1016/j.resourpol.2022.103082.

[50]
Subramani L, Parthasarathy M, Balasubramanian D, et al. 2018. Novel Garcinia gummi-gutta methyl ester (GGME) as a potential alternative feedstock for existing unmodified DI diesel engine. Renewable Energy, 125: 568-577.

[51]
Wang Q W, Hang Y, Sun L C, et al. 2016. Two-stage innovation efficiency of new energy enterprises in China: A non-radial DEA approach. Technological Forecasting and Social Change, 112: 254-261.

[52]
Wang Y, Deng X Z, Zhang H W, et al. 2022. Energy endowment, environmental regulation, and energy efficiency: Evidence from China. Technological Forecasting and Social Change, 177: 121528. DOI: 10.1016/j.techfore.2022.121528.

[53]
Wu H T, Yu H, Ren S Y. 2020. How do environmental regulation and environmental decentralization affect green total factor energy efficiency: Evidence from China. Energy Economics, 91: 104880. DOI: 10.1016/j.eneco.2020.104880.

[54]
Wu R X, Tan Z Z, Lin B Q. 2023. Does carbon emission trading scheme really improve the CO2 emission efficiency? Evidence from China’s iron and steel industry. Energy, 277: 127743. DOI: 10.1016/j.energy.2023.127743.

[55]
Xie X M, Zeng S X, Peng Y F, et al. 2013. What affects the innovation performance of small and medium-sized enterprises in China? Innovation, 15(3): 271-286.

[56]
Yin J H, Zheng M Z, Chen J. 2015. The effects of environmental regulation and technical progress on CO2 Kuznets curve: An evidence from China. Energy Policy, 77: 97-108.

[57]
Yu K M, Moyeed R A. 2001. Bayesian quantile regression. Statistics & Probability Letters, 54(4): 437-447.

[58]
Yuan B L, Zhang Y. 2020. Flexible environmental policy, technological innovation and sustainable development of China’s industry: The moderating effect of environment regulatory enforcement. Journal of Cleaner Production, 243: 118543. DOI: 10.1016/j.jclepro.2019.118543.

[59]
Zhang W, Jin Y G, Wang J P. 2015. Greenization of venture capital and green innovation of Chinese entity industry. Ecological Indicators, 51: 31-41.

[60]
Zhao X L, Zhao Y, Zeng S, et al. 2015. Corporate behavior and competitiveness: Impact of environmental regulation on Chinese firms. Journal of Cleaner Production, 86: 311-322.

[61]
Zhou Y. 2017. Research on the performance, causes and governance of structural features of Chinese enterprises’ innovative financing constraints. Management World, (4): 184-185. (in Chinese)

Options
Outlines

/

〈 〉