Do corporate green investments improve environmental performance? Evidence from the perspective of efficiency

Yutao Chen; Jian Feng

doi:10.1080/21697213.2019.1625578

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

DeepDyve requires Javascript to function. Please enable Javascript on your browser to continue.

Do corporate green investments improve environmental performance? Evidence from the perspective of efficiency

Chen, Yutao; Feng, Jian 2019-01-02 00:00:00 CHINA JOURNAL OF ACCOUNTING STUDIES 2019, VOL. 7, NO. 1, 62–92 https://doi.org/10.1080/21697213.2019.1625578 ARTICLE Do corporate green investments improve environmental performance? Evidence from the perspective of eﬃciency Yutao Chen and Jian Feng School of Accounting, Southwestern University of Finance and Economics, Chengdu, China ABSTRACT KEYWORDS Corporate green investment By linking green investments to corporate environmental perfor- eﬃciency; green mance from the perspective of eﬃciency, this paper quantiﬁes and investments; environmental evaluates ﬁrm-level green investment eﬃciency, with SBM-DEA performance; local approach and hand-collected emission data of Chinese listed environmental enforcement; companies in polluting industries. We ﬁnd that corporate green SBM-DEA investment eﬃciency is overall low, primarily due to excessive green investments, suggesting that managers only invest exten- sively in environmental dimensions without considering the eﬃ- cient allocation and value-creating use of limited resources. Further, we conduct the Tobit regression and demonstrate that local environmental enforcement has an inverted U-shaped eﬀect on green investment eﬃciency of polluting ﬁrms. Notably, this eﬀect is only statistically signiﬁcant in Non-SOEs and small-scale enterprises. Moreover, we document a U-shaped relationship between local environmental enforcement and excessive green investments, and this relationship is also only pronounced in Non- SOEs and small-scale ﬁrms. Our ﬁndings indicate that local govern- ments should optimize corporate green investment eﬃciency through diﬀerentiated environmental regulation. 1. Introduction Enterprises are major consumers of resources and major producers of environmental problems. Due to the increasing concern of environmental issues, polluting ﬁrms are facing more and more stringent environmental regulation. The key to solving the contradiction between economic growth and environmental protection lies in incorpor- ating environmental factors into corporate investment decision-making process (Pearce, Markandya, & Barbier, 1989). Green investments, a particular type of corporate social responsibility (CSR) activity that involves the allocation of ﬁnancial and intangible resources of ﬁrms to transform environmental strategies and objectives into corporate actions and higher environmental performance (Ateş, Bloemhof, van Raaij, & Wynstra, 2012; Martin & Moser, 2016), play a vital role in achieving sustainable development and maximizing social value for enterprises. What green investments should essentially aim at is the reduction of environmental pollution. However, in reality, managerial incentives CONTACT Yutao Chen yutao.chen@hotmail.com School of Accounting, Southwestern University of Finance and Economics, 555 Liutai Avenue, Wenjiang District, Chengdu 611130, China Paper accepted by Kangtao Ye. © 2019 Accounting Society of China CHINA JOURNAL OF ACCOUNTING STUDIES 63 for green investments include regulatory preemption, green images creating and pro- duction cost savings (Maxwell & Decker, 2006), and managers usually utilize qualitative environmental disclosure as the channel to manage investors’ impressions with the purpose of mediating the eﬀect of poor environmental performance on corporate reputation (Cho, Guidry, Hageman, & Patten, 2012). Therefore, we call into question whether ﬁrm-level green investments eﬀectively improve corporate environmental performance. The resources of the enterprise are limited, and the green investments could not generate direct economic beneﬁts. To some extent, green investments transfer ﬁrm resources to other outside stakeholders, which leads to the tension between traditional performance objectives and societal value objectives. Under the stakeholder theory, ﬁrms need to strike a balance between shareholder interests and the interests of non- shareholder stakeholders (Carroll, 1991). On the one hand, shareholders are reluctant to bear the opportunity costs of green investments; on the other hand, the majority of stakeholders hope that polluting ﬁrms would invest a lot of money to reduce environ- mental pollution. Thus, it is crucial for polluting ﬁrms to optimize the input-output eﬃciency of green investments. In this paper, by linking green investments to corporate environmental performance from the perspective of eﬃciency, we quantify and conduct an objective and comprehensive assessment of corporate green investment eﬃciency (GIE). Furthermore, there is a consensus among researchers that government is the major determinant of corporate environmental practices (Delmas & Toﬀel, 2004; Henriques & Sadorsky, 1996; Porter & van der Linde, 1995; Stoever & Weche, 2018; Wang, Wu, & Zhang, 2018a; Zhao, Zhao, Zeng, & Zhang, 2015). China has adopted a decentralization regime of environmental regulation. The central government main- tains the authority over the planning, designing, and formulation of environmental policies, while local governments are mainly responsible for the enforcement of speciﬁc environmental policies (Zheng, 2007). Such environmental decentralization gives local governments considerable discretion in implementing the same environmental regula- tions (Zhang, Chen, & Guo, 2018). Thus, we are also interested in the eﬀect of local environmental enforcement on ﬁrm-level green investment eﬃciency. Previous research has primarily concentrated on green investments rather than green investment eﬃciency. A growing body of literature has investigated the determinants of green investment behaviors and strategies (Ateş et al., 2012; Bahn, Chesney, & Gheyssens, 2012; Costa-Campi, García-Quevedo, & Martínez-Ros, 2017;Eyraud,Clements,&Wane, 2013; Kim, 2013; Maggioni & Santangelo, 2017;Schaltenbrand,Foerstl,Azadegan,&Lindeman, 2016; Song, Yao, Yu, & Shen, 2017). Several studies reveal the trends and schemes of green investments (Eyraud et al., 2013; Karásek & Pavlica, 2016), and examine investors’ response to green investments (Martin & Moser, 2016). Another scream of literature explores the relationship between green investments and ﬁrm performance, including environmental performance, productivity, and export performance (Antonietti & Marzucchi, 2014;Bostian, Färe, Grosskopf, & Lundgren, 2016;Lundgren & Zhou, 2017). There is a paucity of empirical research to directly combine corporate green investments with environmental performance from the point of eﬃciency. Most researchers mainly focus on environmental aspects and assess eco-eﬃciency (Burnett & Hansen, 2008; Figge & Hahn, 2013;Hua,Bian,&Liang, 2007; Yu,Huang,&Luo, 2018)orenvironmental eﬃciency (Chang, Zhang, Danao, & Zhang, 2013; Jiang, Folmer, & Bu, 2016; Li, Fang, Yang, Wang, & Hong, 2013; Song & Zhou, 2016). As for 64 Y. CHEN AND J. FENG environmental regulation, recent studies have examined the association of environmental regulation with productivity (Wang et al., 2018a) or corporate environmental actions (Wang, Wijen, & Heugens, 2018b). It is not yet known whether the enforcement of local environ- mental regulation can aﬀect green investment eﬃciency of polluting ﬁrms. In this study, we call researchers’ attention to the green investment eﬃciency by (1) quantifying ﬁrm-level green investment eﬃciency with nonparametric eﬃciency evalua- tion and (2) exploring the association between local environmental enforcement and corporate green investment eﬃciency. We choose Chinese listed companies in polluting industries as our sample for two reasons. First, it is more direct and accurate to use pollutant emissions as proxies for environmental performance, and the availability of emission data is made possible by the pollutant disclosure regime switch in China. In December 2016, China Securities Regulatory Commission (CSRC) revised the information disclosure guidelines for publicly issuing securities, which requires listed companies that belong to National Specially Monitored Firms (NSMF) program to mandatorily disclose detailed pollutants information in their annual reports. Second, to the best of our knowledge, there is seldom empirical evidence for ﬁrm-level GIE of Chinese listed companies that are crucial components of emerging capital markets. Developing coun- tries are facing severe environmental challenges, however, prior research has only to a limited extent explored how governments inﬂuence corporate environmental practices in emerging economies. In the ﬁrst stage, in order to evaluate green investment eﬃciency of polluting ﬁrms, we hand-collect ﬁrm-level emission data and apply Data Envelopment Analysis (DEA) based on slack-based measure (SBM) approach, with greeninvestmentsasinputsand pollutant emis- sions as undesirable outputs, to examine GIE of Chinese listed polluting ﬁrms in 2016 and 2017. DEA is a nonparametric method for measuring the eﬃciency of peer decision-making units (DMUs) with multiple inputs and outputs (Emrouznejad & Yang, 2017). This method is appro- priate to associate diﬀerent types of green investments with varieties of pollutant emissions, and is able to overcome the interference of the subjective factors that traditional performance evaluation methods suﬀer when setting weights and avoid possible bias or errors in the model speciﬁcation. The SBM approach in DEA is ﬁrst proposed by Tone (1997) and further improved by Tone (2001) to deal with undesirable factors as outputs, which allows us to use pollutant emissions as undesirable outputs in SBM-DEA tests. By in-depth analyses of diﬀerent models, weﬁnd that the corporate green investment eﬃciency is overall low, primarily due to excessive green investments. The results indicate that managers only invest extensively in environmen- tal dimensions for compliance reasons, neglecting the eﬃcient allocation and value-creating use of limited resources. In the second stage analysis, we conduct the Tobit regression to examine the relationship between local environmental enforcement and green investment eﬃciency of polluting ﬁrms. We ﬁrst ﬁnd that local environmental enforcement has an inverted U-shaped eﬀect on green investment eﬃciency. Notably, this eﬀect is only statistically signiﬁcant in non-state- owned enterprises and small-scale enterprises. Further, we document that there exists a U-shaped relationship between local environmental enforcement and excessive green investments, and this relationship is only pronounced in Non-SOEs and small-scale ﬁrms as well. Our research provides several contributions. First, our study extends green ﬁnance and environmental accounting literature by demonstrating the importance of green CHINA JOURNAL OF ACCOUNTING STUDIES 65 investment eﬃciency. Previous literature focuses almost exclusively on green investments (Antonietti & Marzucchi, 2014;Ateş et al., 2012; Bahn et al., 2012;Bostian et al., 2016;Doval & Negulescu, 2014; Eyraud et al., 2013; Inderst, Kaminker, & Stewart, 2012; Karásek & Pavlica, 2016;Kim, 2013; Maggioni & Santangelo, 2017; Schaltenbrand et al., 2016; Song et al., 2017; Voica, Panait, & Radulescu, 2015). Nevertheless, we quantify ﬁrm-level green investment eﬃciency and provide the initial evidence, which suggests the main problem of low- eﬃciency ﬁrms is that managers ignore the eﬃcient allocation and value-creating use of resources in terms of reducing pollutant emissions. Second, our study contributes to CSR literature by dealing with the integration of environmental performance into corporate green investment decisions from the perspective of eﬃciency. The environment is asigniﬁcant element of CSR (Huang & Watson, 2015), which means our research should be based on stakeholder theory. Our study adds value to this strand of literature in that we show an eﬀective way to help resolve the conﬂict of interests between shareholders and non-shareholder stakeholders, which is optimizing corporate green investment eﬃciency. Third, our study is related to research that examines the eﬀect of environmental regulation on corporate behaviors and environmental practices. We shed new light on how the local environmental enforcement aﬀects green investment eﬃciency of polluting ﬁrms in emer- ging economies, which helps to pursue the path of environmental regulatory reform to a greater depth and width and furthers our understanding of how to confront environ- mental challenges more eﬀectively. We believe our ﬁndings have implications for managers and policymakers. Our results suggest that managers ought to recognize the importance of identifying and reducing the low-value allocation of limited resource, and indicate that local governments should optimize corporate green investment eﬃciency through diﬀerentiated environmental regulation. Regarding Non-SOEs and small-scale ﬁrms, local governments could adopt prudent environmental regulation to improve their green investment eﬃciency by restraining excessive green investments. As for SOEs and large-scale ﬁrms, local govern- ments should make more use of market-based means, such as environmental taxes and tradable emission permit, encourage managers to actively take corporate social respon- sibilities instead of passively catering to government regulation, and guide the majority of stakeholders to play a supervisory role. The rest of this paper proceeds as follows. Section 2 provides background information about our setting and discusses related research. Section 3 presents the research design. Section 4 describes the sample and provides descriptive statistics. Section 5 provides the SBM-DEA tests and the Tobit regression results. Section 6 concludes. 2. Background and related research 2.1. Background As China’s economic development has entered a new normal situation, the tolerance of the ecological environment has reached or close to the upper limit. The traditional way of economic growth through excessively consuming resources has been unable to continue. China urgently needs to seek a new path of green economy transformation to solve the dilemma between environmental protection and economic growth. In recent years, China has tremendously invested in environmental governance. The total 66 Y. CHEN AND J. FENG amounts of environmental investments increased by eight times from 2000 to 2017 (Figure B1, Appendix B), however, the average proportion of annual environmental investments in GDP is about 1.36% (Figure B2, Appendix B), which is relatively low as for improving environmental conditions. Meanwhile, China has continuously strength- ened environmental regulation in recent years. The central government has conducted environmental protection supervision since January 2016 and has extended to 31 provincial-level administrative zones by the end of 2017. The supervision teams talked with tens of thousands of local politicians and inspected quite a lot of ﬁrms. As a result, these inspections held local governments and oﬃcials accountable for misconduct and ineﬃciency, and tremendous polluting ﬁrms were rectiﬁed and ﬁned. Regarding listed companies, China Securities Regulatory Commission (CSRC) revised the information disclosure guidelines for publicly issuing securities in December 2016, which requires listed companies in polluting industries to mandatorily disclose detailed pollutants information, such as the name of major pollutants, emission concentration and total amounts, excessive discharge, etc., in their annual reports. This regime switch setting of pollutant disclosure provides available emission data that we need as proxies for environmental performance and further quantify ﬁrm-level green investment eﬃciency. 2.2. Related research Prior literature focuses almost exclusively on green investments rather than green investment eﬃciency, including the determinants of green investment behaviors and strategies (Ateş et al., 2012; Bahn et al., 2012; Costa-Campi et al., 2017; Eyraud et al., 2013; Kim, 2013; Maggioni & Santangelo, 2017; Schaltenbrand et al., 2016; Song et al., 2017), the trends and schemes of green investments (Eyraud et al., 2013; Karásek & Pavlica, 2016), and investors’ reaction to green investments(Martin & Moser, 2016). Although there is research evaluating the green investment eﬃciency of the govern- ment (Kim, Lee, Park, Zhang, & Sultanov, 2015), researchers seldom examine ﬁrm-level green investment eﬃciency. Many researchers have primarily concentrated on environ- mental aspects and assess eco-eﬃciency (Burnett & Hansen, 2008; Figge & Hahn, 2013; Hua et al., 2007; Yu et al., 2018) or environmental eﬃciency (Chang et al., 2013; Jiang et al., 2016; Li et al., 2013; Song & Zhou, 2016). Another scream of literature explores the interactions among environmental investments, productivity, energy eﬃciency, and environmental performance (Bostian et al., 2016; Lundgren & Zhou, 2017). There is a paucity of empirical research to directly combine corporate green investments with environmental performance from the perspective of eﬃciency. As for environmental regulation, recent studies have examined the association of environmental regulation with productivity (Wang et al., 2018a) or corporate environmental actions (Wang et al., 2018b). However, there has been little discussion about the correlation between the enforcement of environmental regulation and corporate green investment eﬃciency. We highlight a few related papers and discuss how our study advances the literature. Kim et al. (2015) analyze the eﬃciency of the government’s green investments in three major new and renewable energy (NRE) sources in Korea by applying DEA, and they argue that strategic selection and focused investment help accomplish the policy objectives with fewer resources and budget. Lundgren and Zhou (2017) use DEA to calculate the Malmquist ﬁrm performance indexes that include productivity, energy CHINA JOURNAL OF ACCOUNTING STUDIES 67 eﬃciency, and environmental performance, and then apply a panel vector auto- regression (VAR) method to explore the dynamic and causal relationship between the three dimensions of ﬁrm performance and green investments, ﬁnding that improving energy eﬃciency is able to obtain various advantages. Some research applies DEA to test environmental eﬃciency. Li et al. (2013) use the Super-SBM model with undesirable outputs to calculate regional environmental eﬃ- ciency in China from 1991 to 2001, ﬁnding that the eﬃciency of eastern area is higher than that of central area and western area. Then they utilize the Tobit regression to conclude that ﬁscal decentralization, technology progress, economic scale, and regional diﬀerence can aﬀect environmental eﬃciency. Jiang et al. (2016) apply DEA and a structural equation model to examine the interaction among environmental eﬃciency, output eﬃciency, and proﬁt using the sample of 137 ﬁrms in the textile industry of China’s Jiangsu Province. They ﬁnd a negative eﬀect of environmental eﬃciency on proﬁt and a positive impact of proﬁt on environmental eﬃciency and they demonstrate that output eﬃciency could lower proﬁt while proﬁt is able to enhance output eﬃciency. With regard to environmental regulation, Wang et al. (2018a) investigate how the water quality regulations aﬀect ﬁrms’ emissions of chemical oxygen demand (COD) and productivity in a setting of China. Their result suggests that a 10% decrease in total COD emissions from the industrial sectors just need a 0.1% decrease in output values under the contemporary production technologies. Wang et al. (2018b) disengage the various functions of diﬀerent government levels in China and ﬁnd an inverted U-shaped rela- tionship between administrative hierarchical distance and corporate environmental actions. Despite the aforementioned research, an extremely important aspect of corporate green investment – to what extent is it eﬃcient in terms of reducing pollutant emis- sions? – is considerably unexplored. Our study attempts to answer this question by quantifying ﬁrm-level green investment eﬃciency and further identifying the direction of improvement for low-eﬃciency ﬁrms. In addition, it is not yet known whether environmental enforcement of local governments can inﬂuence the green investment eﬃciency of polluting ﬁrms. Therefore, we evaluate ﬁrm-level green investment eﬃ- ciency and analyze the association between local environmental enforcement and green investment eﬃciency. 3. Research design 3.1. Measuring corporate green investment eﬃciency 3.1.1. The deﬁnition of corporate green investments In order to measure GIE, we should ﬁrst understand what green investments are. However, there seems to be no general deﬁnition of green investments in prior litera- ture. From the perspective of macroeconomics, Eyraud et al. (2013)deﬁne green invest- ments as the inevitable public and private investments to reduce air pollutant emissions and greenhouse gas, without substantially decreasing the production and consumption of non-energy commodities. In the ﬁeld of enterprises’ micro-behaviors, Murillo-Luna, Garcés-Ayerbe, and Rivera-Torres (2008) mention that green investments are always related to morally charged problems linked with green management and corporate 68 Y. CHEN AND J. FENG environmentalism. Doval and Negulescu (2014) state that green investments could be essentially viewed as the expenses which companies made for a friendly impact on the environment. Ateş et al. (2012) consider green investments as the combination of corporate internal investments and external investments involved with the domains of environmental design, production, and logistics. Martin and Moser (2016) regard green investments as a special kind of CSR activity which aims at the reduction of carbon emissions. Voica et al. (2015) argue that green investments are fundamentally consid- ered to be the climate resilient or low-carbon investments made by ﬁrms in the areas of climate change, renewable energy and clean technologies. Since the deﬁnition of green investments varies among researchers, it is important to clarify how the concept is used in this paper. Remarkably, there is a substantial common intersection of the various deﬁnitions and concepts regarding sectors, goods, technologies, services, and processes (Inderst et al., 2012). Thus, we deﬁne corporate green investments as internal invest- ments in equipment, technologies, materials, energy and services that can prevent, control and reduce environmental pollution, produce environmental beneﬁts, and reduce environmental costs, with the goals of improving corporate environmental performance, developing green management and reducing environmental risks. Lundgren and Zhou (2017) propose that it would be helpful to analyze two types of environmental investments – prevention investments and treatment investments sepa- rately, and contend that these two kinds of investments are able to be deemed as proactive and reactive environmental investments. However, they fail to diﬀerentiate these two sorts of investments due to data limitation problems, instead, they mix them into one variable. Undoubtedly, the scope of green investments should be wider than environmental investments, but it is reasonable to classify green investments into prevention investments and treatment investments. Considering ﬁrm-level green invest- ments, we look through all the items of increased construction in progress, increased R&D expenditure, purchased ﬁxed assets and overhead expenses of polluting ﬁrms. If the items relate to investments that impact the production process and act to prevent pollution, we classify them into green prevention investments (GPI). Speciﬁcally, GPI include the use of cleaner energy, new and more eﬃcient advanced materials and less environmentally damaging input factors; the technical transformation of clean produc- tion; the investments in renewable technologies (including large hydroelectric projects); the R&D in energy-eﬃcient and green technologies; and the methods facilitating energy saving and resource recycling. For another, if the items are associated with investments that cope with already emitted pollutant emissions and do not inﬂuence the actual processes of production, such as the environmental treatment to low emissions, desul- phurization and dust removal, waste utilization and regeneration, and the maintenance of environmental protection equipment, we classify them into green treatment invest- ments (GTI). Thus, the total green investments (GI) are the aggregated amounts of GPI and GTI. See Table A1 of Appendix A for variable deﬁnitions. 3.1.2. SBM-DEA approach for measuring ﬁrm-level green investment eﬃciency DEA is a nonparametric method for measuring eﬃciency, productivity or performance of peer decision-making units (DMUs) with multiple inputs and outputs (Emrouznejad & Yang, 2017), which is introduced by Charnes, Cooper, and Rhodes (1978). DEA begins with the establishment of an ‘eﬃcient frontier’ consisting of a group of DMUs that CHINA JOURNAL OF ACCOUNTING STUDIES 69 demonstrate best practices and achieve an eﬃciency score of 1.00, and then assigns the other non-frontier DMUs speciﬁceﬃciency scores based on their distances to the eﬃcient frontier (Liu, Lu, Lu, & Lin, 2013). This method is capable of overcoming the interference of the subjective factors that traditional performance evaluation methods suﬀer when setting weights, and avoiding possible bias or errors in the model speciﬁcation. In this paper, we use pollutant emissions, which are undesirable, as proxies for environmental performance. The fewer pollutants emitted by enterprises, the better. The slack-based measure (SBM) approach in DEA is ﬁrst proposed by Tone (1997) and further improved by Tone (2001) to deal with undesirable factors as outputs. He deﬁnes the SBM model including undesirable outputs as follow: 1 i i¼1 m x ia min ρ ¼ þ b P P q s q s 1 1 2 r t 1 þ þ q þq r¼1 y t¼1 b 1 2 ra ra s.t. Xδ þ s ¼ x Yδ s ¼ y Bδ þ s ¼ b þ b δ; s ; s ; s 0 Among them, ρ measures the eﬃciency score of DMU ; x , y ; and b represent the a a a a actual input, desirable output, and undesirable output of DMU , respectively; X, Y, and B refer to the target values of input, desirable output, and undesirable output based on þ b the best practices, respectively; s , s , s denote slack values of input, desirable output, and undesirable output ofDMU , respectively. To be speciﬁc, s describes the excessive input, which is equal to the diﬀerence between actual input and best target input. s is the insuﬃcient desirable output, which is calculated as the diﬀerence between max- imum expected desirable output and actual desirable output. s is regarded as the undesirable output exorbitance, which captures the diﬀerence between actual undesir- able output and the minimum target of undesired output. Further, we are able to measure the input ineﬃciency and output ineﬃciency of DMU using these slack variables. The input ineﬃciency is the ratio of the absolute value of input’s slack variable to actual input, deﬁned as |s |/x . In the same manner, we measure the output ineﬃ- ciency by computing the ratio of the absolute value of output’s slack variable to actual þ b output, namely |s |/y or |s |/b . A higher value of these ratios could be interpreted as a a a higher degree of ineﬃciency of inputs or outputs. Environmental pollution is undesirable, thus, we apply this SBM-DEA approach of Tone (1997, 2001), which is capable of combining ﬁnancial information with environ- mental information and dealing with undesirable outputs, to measure ﬁrm-level green investment eﬃciency (GIE). While building the SBM-DEA model, we consider each sample ﬁrm to be a DMU and use two types of green investments – GPI and GTI – as actual inputs (x ). We obtain all the items of increased construction in progress (Fieldname: FN_Fn017B05), increased R&D expenditure (Fieldname: Fn02304), purchased 70 Y. CHEN AND J. FENG ﬁxed assets (Fieldname: Fn02006) and overhead expenses (Fieldname: FN05202) of Chinese listed companies in polluting industries from the GTA China Stock Market Financial Database – Statement Notes Database, and further classify them into GPI or GTI according to their deﬁnitions. We then hand-collect a variety of pollutant emission data of each polluting ﬁrm for actual undesirable outputs (b ) from ﬁrm annual reports, corporate social responsibility reports, sustainability reports, and environmental reports. Given that previous research on the relationship between investments in environmental dimensions and ﬁrm performance is inconsistent or inﬂuenced by certain moderators (Antonietti & Marzucchi, 2014; Bostian et al., 2016; Broberg, Marklund, Samakovlis, & Hammar, 2013; Jiang et al., 2016; Sengupta, 2012; Singal et al., 2014), we choose not to include ﬁrm performance as the actual desirable output (y ) in our models. Accordingly, the SBM-DEA model can establish an eﬃcient frontier, and calculate GIE (ρ), best target green investments (X) and target pollutant emissions (B) for each sample ﬁrm. The slack variables for green investments and pollutant emissions measure the excessive corpo- rate green investments (s ) and pollutant emission exorbitance (s ), respectively. In this study, we examine ﬁve output-oriented and variable returns to scale (VRS) SBM-DEA models that diﬀer in the number of undesirable outputs. As the main objective of green investments is to enhance corporate environmental performance, we use output- oriented models to ensure the minimization of pollutant emissions. Besides, the VRS models assume that as inputs change, outputs could change at a diﬀerent rate, which is more in line with the actual situation. A higher value of GIE indicates more eﬃcient green investments of the sample ﬁrm in terms of reducing undesirable outputs – pollutant emissions. See Table A1 of Appendix A for variable deﬁnitions. 3.2. The association between local environmental enforcement and corporate green investment eﬃciency Many researchers have identiﬁed that the government is the key driver for ﬁrms’ environmental practices (Delmas & Toﬀel, 2004; Henriques & Sadorsky, 1996; Porter & van der Linde, 1995; Stoever & Weche, 2018; Wang et al., 2018a; Zhao et al., 2015). In addition, governments operating at diﬀerent levels have alternate and even conﬂicting eﬀects on corporate environmental behaviors, and there is greater involvement of governments in developing countries (Wang et al., 2018b). China has adopted a decentralization regime of environmental regulation. The central government main- tains the authority over the planning, designing, and formulation of environmental policies, while local governments are mainly responsible for the enforcement of speciﬁc environmental policies (Zheng, 2007). Such environmental decentralization gives local governments considerable discretion in implementing the same environmental regula- tions (Zhang et al., 2018). Since the environmental enforcement of local governments is regular and its impact on local ﬁrms is more direct, we expect local environmental enforcement to be the main driver of corporate green investment eﬃciency. Several We hand-collect the emission data of polluting ﬁrms’ seven major pollutants – three for water pollution (COD, NH N, and Wastewater), three for air pollution (SO ,NO , and Soot), and one for soil pollution (Solid Waste). 2 X All output-oriented SBM-DEA models use two inputs – GPI and GTI, but the number of undesirable outputs – emissions of major pollutants – are diﬀerent across these ﬁve models due to the limitation of sample ﬁrms’ emission disclosure. See detail in Table 2. CHINA JOURNAL OF ACCOUNTING STUDIES 71 studies have revealed that the relationship between environmental regulation of gov- ernments and environmental behaviors of ﬁrms is nonlinear (Wang et al., 2018b;Yu et al., 2018), thus, we include the square term of local environmental enforcement in Equation (a). Further, we apply the Tobit regression to estimate Equation (a) because eﬃciency scores calculated by SBM-DEA models are censored for the above the max- imum value of GIE. GIE ¼ αþβ Enforcement þ β Enforcement þ β Size þ β SOE þ β Age þ β GIscale 1 2 3 4 5 6 þ β Leverage þ β OCF þ β Opportunity þ β ROA þ β Growth þ β LSR 7 8 9 10 11 12 þ β Duality þ β DAGE þ β PGDP þ β Structure þ β ENpressure 13 14 15 16 17 þ β ENsupervision þ Area fixed effects þ Industry fixed effects þ Year fixed effects þ ε (a) Enforcement is the proxy for local environmental enforcement, which is based on the Pollution Information Transparency Index (PITI) report issued by the Institute of Public and Environmental Aﬀairs (IPE) and the Natural Resources Defense Council (NRDC) on the IPE website (http://www.ipe.org.cn/reports/Reports_18336_1.html). The assessment standards of PITI cover ﬁve key dimensions: environmental supervision information; pollution source self-disclosure; interactive response; enterprise emission data; and information disclosure of environmental impact assessment. Its evaluation system fully takes into account the current severe environmental situation in China and the trend of reconstruction and improvement of environmental supervision regime. Therefore, PITI is the most compre- hensive and objective index to measure the actual situation of local governments in implementing environmental information disclosure policy (Shen & Feng, 2012). The higher the value of PITI is, the more transparent the pollution information is, suggesting that the environmental enforcement of this city is stricter (Yu et al., 2018). Since our focus is on the enforcement of environmental regulation, it is appropriate and feasible to use PITI as the proxy for local environmental enforcement. We match the city where the polluting ﬁrm is registered with PITI to get the enforcement score of local environmental regulation for each sample ﬁrm-year observation. Compared to other measures of environmental regula- tion, such as the state-level pollution abatement costs (Henderson & Millimet, 2007;Keller& Levinson, 2002) or the pollution discharge fees collected by the local governments (Levinson, 1996), PITI measures the environmental enforcement of local governments from various aspects, which can better reﬂect the comprehensive implementation of multi- dimensional environmental regulation by local governments. Nonetheless, we use alter- native measures of local environmental enforcement to conduct robustness tests, ensuring that our main result is not driven by the measure selection. We use Size, SOE, and Age to control for ﬁrm characteristics. Size is the natural logarithm of the ﬁrm’s total assets at the end of ﬁscal year t. The indicator variable SOE is 1 if the ﬁrm is state-owned during ﬁscal year t and 0 otherwise. Age is the number of years from the establishment of the ﬁrm to the sample year t. We also control for the factors that would aﬀect investment eﬃciency including GIscale, Leverage, OCF, Opportunity, ROA and Growth. Considering the continuous impact of green investment scale, we use lagged value for GIscale, which is equal to the total green investments of the ﬁrm during ﬁscal year t-1 scaled by total assets at the end of ﬁscal year t-1. Leverage is the total debt scaled by total assets, all measured at the end of ﬁscal year t. OCF is the 72 Y. CHEN AND J. FENG net cash ﬂow generated by the ﬁrm’s operating activities during ﬁscal year t scaled by total assets at the end of ﬁscal year t. Opportunity is the ratio of market value of equity to total assets at the end of ﬁscal year t. ROA is the ﬁrm’s return on assets for ﬁscal year t. Growth is the ﬁrm’s growth rate of operating income during ﬁscal year t. Moreover, we use LSR, Duality, and DAGE to control for the eﬀect of corporate governance. LSR is the ﬁrm’s largest shareholder rate for ﬁscal year t. The indicator variable Duality is 1 if the ﬁrm’s chair of the board is also the ﬁrm’s manager and 0 otherwise. DAGE is the average age of the ﬁrm’s directors. Further, we control for the external environment that sample ﬁrms are facing. PGDP is the regional GDP per capita of the province where sample ﬁrm is registered for ﬁscal year t. Structure is the provincial gross industrial output scaled by the regional GDP of the province where sample ﬁrm is registered for ﬁscal year t. ENpressure is the annual average PM2.5 concentration of the province where sample ﬁrm is registered for ﬁscal year t, and a higher value of it indicates ﬁrms that are located in this province are faced with higher pressure of environmental protection. The indi- cator variable ENsupervision is 1 if the ﬁrm is registered in the province where the central government conducted environmental protection supervision during ﬁscal year t and 0 otherwise. Finally, we include area ﬁxed eﬀects, industry ﬁxed eﬀects, and year ﬁxed eﬀects to ensure the generalization of our results across areas, industries and years. We summarize all our variable deﬁnitions in Appendix A. 4. Sample and descriptive statistics 4.1. Sample collection In December 2016, CSRC revised the information disclosure guidelines for publicly issuing securities, which requires Chinese listed ﬁrms that belong to NSMF program (key polluting ﬁrms) to mandatorily disclose detailed pollutants information in their annual reports. This pollutant disclosure regime switch setting provides available ﬁrm- level emission data in 2016 and 2017. According to the Directory of Industrial Classiﬁcations for Listed Firms Subject to Environmental Protection Supervisions issued by the Ministry of Ecology and Environment of the People’s Republic of China (MEEC) in 2008, we select listed ﬁrms in polluting industries from China Security Market and Accounting Research (CSMAR) database. After excluding ﬁrms that are under special treatment or delisted companies, we obtain 2004 sample ﬁrm-year observations. In order to acquire annual total emission data of each ﬁrm, we look through and hand- collect the emission data from ﬁrm annual reports, corporate social responsibility reports, sustainability reports, and environmental reports for the 2004 sample ﬁrm- years in 2016 and 2017. However, the majority of ﬁrms only have narrative or opaque statements about their pollutant emissions or do not disclose quantitative data of total emissions, which leads us to drop 1348 ﬁrm-years. Table 1 provides the descriptive statistics for the pollutant emission disclosure. We report the total number of polluting ﬁrms, the number of polluting ﬁrms with emission disclosure, and the proportion of disclosed ﬁrms in total polluting ﬁrms for key polluting ﬁrms and non-key polluting ﬁrms, respectively. For key polluting ﬁrms, which should mandatorily disclose annual As long as the sample ﬁrm discloses the total annual emissions of one major pollutant, the ﬁrm is deemed to have made the pollutant emission disclosure. CHINA JOURNAL OF ACCOUNTING STUDIES 73 Table 1. Descriptive statistics for pollutant emission disclosure of listed ﬁrms in polluting industries. Polluting ﬁrms with Polluting emission ﬁrms disclosure The proportion of disclosed Year N %Total N %Total ﬁrms in total polluting ﬁrms Key polluting ﬁrms 2016 325 33.71% 178 90.82% 54.77% (mandatory disclosure) 2017 549 52.79% 447 97.17% 81.42% Non-key polluting ﬁrms 2016 639 66.29% 18 9.18% 2.82% (voluntary disclosure) 2017 491 47.21% 13 2.83% 2.65% Total ﬁrms 2016 964 100% 196 100% 20.33% 2017 1040 100% 460 100% 44.23% Firm-year observations 2004 100% 656 100% 32.73% emission data, only 54.77% of them disclose such information in 2016. The situation has been improved in 2017 and 81.42% of key polluting ﬁrms disclose emission data. For non-key polluting ﬁrms, which are voluntary to disclose pollutant emissions, 2.82% and 2.65% of these ﬁrms choose to disclose related data in 2016 and 2017, respectively. The proportion of all disclosed sample ﬁrm-years in total polluting ﬁrm-years is only 32.73%, as a result, our initial test sample during 2016–2017 has 656 ﬁrm-year observations. The numbers suggest that environmental regulation and enforcement for polluting ﬁrms need to be strengthened and the requirements of environmental disclosure ought to be further enhanced. 4.2. Descriptive statistics We present descriptive statistics of variables used in the SBM-DEA models in Table 2.As for input variables, the means of GPI, GTI, and GI are greatly larger than their medians, implying that a part of sample ﬁrms have extremely large amounts of green invest- ments. However, some polluting ﬁrms even have no green investments. We calculate the standard deviation of GPI, GTI, and GI for 656 ﬁrm-year observations and ﬁnd the numbers are very large. The situation shows that managers’ green investment decisions are tremendously diﬀerent across sample ﬁrms. Table 2. Descriptive statistics of variables used in the SBM-DEA models. Variable N Mean Std. Dev. Median Min Max Inputs – Corporate green investments (CNY 10-thousand yuan) GPI 656 11,081.446 53,968.633 0.000 0.000 899,238.400 GTI 656 7952.651 75,115.371 300.964 0.000 1,554,073.000 GI 656 19,034.096 91,942.801 1026.736 0.000 1,554,073.000 Undesirable outputs – Corporate pollutant emissions (t) COD 518 417.636 2088.840 52.010 0.000 28,400.000 NH N 467 80.924 752.662 2.620 0.000 11,400.000 Wastewater 131 3,645,320.000 15,570,000.000 336,500.000 0.000 168,700,000.000 SO 476 233.345 135.056 234.500 1.000 468.000 NO 460 5939.395 54,370.720 224.057 0.016 1,140,781.000 Soot 378 188.212 108.316 188.500 1.000 374.000 Solid Waste 60 1,124,000.000 4,759,506.000 8268.328 1.000 35,100,000.000 74 Y. CHEN AND J. FENG On the other hand, Table 2 reports descriptive statistics of major pollutant emission data for 656 ﬁrm-years that disclose related information. Because CSRC has no speciﬁc requirements, such as which pollutants should be included and how many indicators need to be disclosed, the emission disclosures of each ﬁrm are slightly diﬀerent. We report seven major pollutants – three for water pollution (COD, NH N, and Wastewater), three for air pollution (SO ,NO , and Soot), and one for soil pollution (Solid Waste). 2 X Except for SO and Soot, the means of these pollutant emissions are all greater than their medians, suggesting that some ﬁrms have huge emissions. It is worth noting that only 131 sample ﬁrm-years disclose Wastewater discharges and 60 sample ﬁrm-years disclose Solid Waste emissions, thus, we choose the other ﬁve kinds of pollutant emissions as undesirable outputs in later SBM-DEA empirical examination. 5. Empirical results 5.1. Analyses of SBM-DEA results 5.1.1. Test of corporate green investment eﬃciency To examine the GIE of polluting ﬁrms, which should aim at reducing pollutant emissions as much as possible, we apply the output-oriented SBM-DEA approach to calculate the eﬃciency score. Particularly, we use VRS models, assuming that as inputs increase, undesir- able outputs decrease at a diﬀerent rate, to estimate GIE. Table 3 reports the primary SBM- DEA results. All models use two inputs – GPI and GTI, but the undesirable outputs are diﬀerent across these ﬁve models due to the limitation of sample ﬁrms’ emission disclosure. Model (1) through Model (3) consider both water pollution and air pollution outputs. More speciﬁcally, Model (1) has ﬁve undesirable outputs – COD, NH N, SO ,NO ,Soot,and uses 3 2 X Table 3. Corporate green investment eﬃciency – SBM-DEA results using diﬀerent outputs. Models using both water pollution and air pollution Model using only water pollution Model using only air pollution outputs outputs outputs Model Model Model (1) (2) (3) Model (4) Model (5) Obs 207 294 331 389 416 Mean 0.6090 0.6004 0.5622 0.5359 0.5377 Std. Dev. 0.1936 0.1857 0.1477 0.1176 0.1177 Median 0.5083 0.5067 0.5030 0.5001 0.5004 Min 0.5000 0.5000 0.5000 0.5000 0.5000 Max 1.0000 1.0000 1.0000 1.0000 1.0000 N. eﬃcient 40 50 32 22 23 %eﬃcient 19.32% 17.01% 9.67% 5.66% 5.53% Notes: All models are output-oriented and VRS models based on the SBM approach using two inputs – GPI and GTI. The major diﬀerences among these models are their outputs. Model (1) has ﬁve undesirable outputs – COD, NH N, SO , 3 2 NO , Soot, and uses a sample consisting of ﬁrm-years that disclose these ﬁve pollutant emissions at the same time. Model (2) has four undesirable outputs – COD, NH N, SO ,NO , and its sample is formed from ﬁrm-years that disclose 3 2 X these four pollutant emissions in the meantime. Model (3) uses COD and SO as undesirable outputs, and its sample is constitutive of ﬁrm-years that disclose COD and SO emissions at the same time. Model (4) has only two outputs of water pollution – COD and NH N, and its sample includes ﬁrm-years that disclose these two pollutant emissions in the meantime. Model (5) uses only two outputs of air pollution – SO and NO , and its sample is comprised of ﬁrm-years 2 X that disclose these two pollutant emissions at the same time. We exclude ﬁrm-years with zero total amounts of GPI and GTI for all samples used in diﬀerent SBM-DEA models. CHINA JOURNAL OF ACCOUNTING STUDIES 75 a sample consisting of ﬁrm-years that disclose these ﬁve pollutant emissions at the same time. In Table 3, Model (1) presents that 19.32% of polluting ﬁrms are considered to be eﬃcient with an eﬃciency score equal to 1.00, and the average GIE is 0.6090. For Model (2), we select four outputs – COD, NH N, SO ,NO , and its sample is formed from ﬁrm-years that 3 2 X disclose these four pollutant emissions in the meantime. The average GIE is 0.6004 and 17.01% of polluting ﬁrms are considered to be eﬃcient in Model (2) of Table 3.Wenextuse COD and SO as undesirable outputs in Model (3), and its sample is constitutive of ﬁrm- years that disclose COD and SO emissions at the same time. The result of Model (3) shows that the proportion of eﬃcient polluting ﬁrms is 9.67% and the average GIE is 0.5622. Taken together, these SBM-DEA results demonstrate that the green investment eﬃciency of polluting companies is overall low, and highlight that managers should recognize the importance of identifying and reducing the low-value allocation of limited resource. Next, we attempt to test GIE, which aims at reducing water pollution and air pollution separately, and use Model (4) and Model (5) to conduct the SBM-DEA tests. Speciﬁcally, Model (4) has only two outputs of water pollution – COD and NH N, and its sample includes ﬁrm-years that disclose these two pollutant emissions in the meantime; Model (5) uses only two outputs of air pollution – SO and NO , and its sample is comprised of 2 X ﬁrm-years that disclose these two pollutant emissions at the same time. Model (4) in Table 3 reports an average GIE of 0.5359, with 5.66% of the sample ﬁrms on the eﬃciency frontier. By comparison, Model (5) presents that the average GIE is 0.5377, and only 5.53% of polluting ﬁrms are considered to be eﬃcient, which are not much diﬀerent from the results of Model (4), suggesting the eﬃciency of corporate green investments is still low. However, we caution that these ﬁndings of Model (4) and Model (5) might have some limitations since we cannot distinguish which green investments are used to cope with water pollution and which green investments are used to control air pollution from our data. Based on the above results, we conclude that it is more appropriate to include both water pollution and air pollution outputs in SBM-DEA models. In addition, considering the necessity of suﬃcient observations, we focus our subsequent analyses on Model (3). Polluting ﬁrms are often aﬀected by certain industry policies or regional factors. Hence, GIE could be aggregated into clusters according to those two dimensions. Table 4 summarizes the GIE across industries and areas based on SBM-DEA model (3). We only report results for industries that have at least ten sample ﬁrm-years. The average GIE across industries ranges from a low of 0.5007 (coal mining and washing) to a high of 0.6129 (pharmaceutical manufacturing). Meanwhile, we can observe a greater variance in the number of polluting ﬁrms that are regarded to be eﬃcient, with none of the ﬁrms in coal mining and washing, paper-making and paper products being eﬃcient, compared to 16.95% of ﬁrms in pharmaceutical manufacturing being on the eﬃciency frontier. Furthermore, we provide the GIE classiﬁed by area, from a low of 0.5597 (central region) to a high of 0.5796 (northeastern region). The lowest proportion of eﬃcient ﬁrms is the western region (8.43%) and the highest proportion of eﬃcient ﬁrms is the eastern region (10.12%). These variations in GIE across industries and areas suggest that we should control for industry ﬁxed eﬀects and area ﬁxed eﬀects in second- stage regression analyses. 76 Y. CHEN AND J. FENG Table 4. Corporate green investment eﬃciency classiﬁed by industry and area based on Model (3). Std. % Obs Mean Dev. Median Min Max N. eﬃcient eﬃcient Sample by Industry B06. Coal mining and washing 14 0.5007 0.0007 0.5003 0.5000 0.5023 0 0.00% B09. Non-ferrous metals mining and 10 0.5548 0.1569 0.5013 0.5001 1.0000 1 10.00% dressing C15. Alcohol and wine manufacturing 13 0.5802 0.1351 0.5237 0.5009 1.0000 1 7.69% C22. Paper-making and paper products 19 0.5026 0.0049 0.5004 0.5000 0.5162 0 0.00% C26. Chemical raw materials and chemical 102 0.5585 0.1429 0.5033 0.5001 1.0000 9 8.82% products manufacturing C27. Pharmaceutical manufacturing 59 0.6129 0.1836 0.5179 0.5003 1.0000 10 16.95% C31. Ferrous metal smelting and rolling 31 0.5542 0.1513 0.5002 0.5000 1.0000 3 9.68% processing C32. Non-ferrous metal smelting and 25 0.5670 0.1635 0.5060 0.5000 1.0000 3 12.00% rolling processing Sample by Area Eastern Region 168 0.5621 0.1504 0.5033 0.5000 1.0000 17 10.12% Central Region 60 0.5597 0.1500 0.5013 0.5001 1.0000 6 10.00% Western Region 83 0.5599 0.1413 0.5037 0.5001 1.0000 7 8.43% Northeastern Region 20 0.5796 0.1539 0.5048 0.5001 1.0000 2 10.00% All industries are categorized by the China Securities Regulatory Commission (CSRC) and only eﬃciencies for polluting industries with ten or more ﬁrm-years are reported. 5.1.2. Test of input ineﬃciency and output ineﬃciency The results in Table 3 show that the GIE of polluting ﬁrms is low. To further illustrate the reason for their ineﬃciency, we apply non-oriented and VRS models based on SBM to compare the input ineﬃciency and output ineﬃciency in diﬀerent models. We measure the input/output ineﬃciency by computing the ratio of the absolute value of input/ output’s slack variable to the actual input/output. The slack variable for input is exces- sive green investments, which is equal to the diﬀerence between actual input and target input that captures the best. And the slack variable for undesirable output is emission exorbitance, which is calculated as the diﬀerence between actual output and best target output. In Table 5, Model (6) through Model (8) are all non-oriented and VRS models based on the SBM approach using GI as input, but they diﬀer in the number of undesirable outputs. Model (6) has ﬁve undesirable outputs – COD, NH N, SO ,NO , Soot, and uses 3 2 X a sample comprising of ﬁrm-years that disclose these ﬁve pollutant emissions at the same time. The mean input ineﬃciency score of Model (6) is 0.8003, which is larger than the mean output ineﬃciency score calculated as 0.5946. For Model (7), we use four outputs – COD, NH N, SO ,NO , and its sample is made up of ﬁrm-years that disclose 3 2 X these four pollutant emissions in the meantime. Similarly, Model (7) reports that the mean input ineﬃciency score (0.8368) is larger than the mean output ineﬃciency score (0.6557). Moreover, we select COD and SO as undesirable outputs of Model (8), and its sample is composed of ﬁrm-years that disclose COD and SO emissions at the same time. The result of Model (8) exhibits that mean input ineﬃciency score (0.9002) is also larger than the mean output ineﬃciency score (0.7481). Further, we conduct T-tests to examine the diﬀerence between the mean of input ineﬃciency and output ineﬃciency for these models separately, and the t-statistics are all statistically signiﬁcant at 1% level in all models. Thus, we conclude that a large proportion of the ineﬃciency observed in polluting ﬁrms is attributable to input ineﬃciency, suggesting that the problem of CHINA JOURNAL OF ACCOUNTING STUDIES 77 Table 5. Input ineﬃciency and output ineﬃciency based on non-oriented SBM-DEA models. Obs Mean Std. Dev. Median Min Max Test of diﬀerence Model (6) Input Ineﬃciency 207 0.8003 0.3653 0.9915 0 1 14.84*** Output Ineﬃciency 207 0.5946 0.3198 0.6814 0 0.9965 Model (7) . Input Ineﬃciency 294 0.8368 0.3353 0.9949 0 1 15.64*** Output Ineﬃciency 294 0.6557 0.3158 0.7297 0 0.9999 Model (8) Input Ineﬃciency 331 0.9002 0.2563 0.9963 0 1 12.11*** Output Ineﬃciency 331 0.7481 0.2995 0.9055 0 0.9999 Notes: Model (6) through Model (8) are all non-oriented and VRS models based on the SBM approach using GI as input, but they diﬀer in the number of undesirable outputs. Model (6) has ﬁve undesirable outputs – COD, NH N, SO ,NO , 3 2 X Soot, and uses a sample comprising of ﬁrm-years that disclose these ﬁve pollutant emissions at the same time. Model (7) has four outputs – COD, NH N, SO ,NO , and its sample is made up of ﬁrm-years that disclose these four pollutant 3 2 X emissions in the meantime. Model (8) uses COD and SO as undesirable outputs, and its sample is composed of ﬁrm- years that disclose COD and SO emissions at the same time. We conduct T-tests to examine the diﬀerence between the mean of input ineﬃciency and output ineﬃciency, and *** indicates statistical signiﬁcance at the 1% level. excessive green investments is prominent and managers neglect the eﬃcient allocation and value-creating use of limited resources. 5.1.3. Analyses of excessive corporate green investments By using non-oriented SBM-DEA models in section 5.1.2, we ﬁnd that the dominating reason for polluting ﬁrms’ low GIE is that managers aimlessly invest a large amount of money in environmental dimensions. To further analyze the diﬀerent types of excessive green invest- ments, we use slack variables for GPI and GTI in output-oriented and VRS Model (1) through Model (3) as excessive prevention investments (GPIslack) and excessive treatment investments (GTIslack), respectively. To examine the diﬀerence between the mean of GPIslack and GTIslack, we conduct T-tests in all three models in Table 6. However, the t-statistics for all models are insigniﬁcant, indicating that there is no signiﬁcant diﬀerence between GPIslack and GTIslack. Overall, masses of polluting ﬁrms excessively input both green prevention investments and green treatment investments without considering the input-output eﬃciency. Table 6. Analyses of excessive corporate green investments. Obs Mean Std. Dev. Median Min Max Test of diﬀerence Model (1) GPIslack 207 3782.621 17,590.44 0 0 172,984.5 −1.14 GTIslack 207 10,094.88 77,073.08 139.811 0 1,084,328 Model (2) GPIslack 294 6604.772 37,340.18 0 0 404,856.8 −0.37 GTIslack 294 8222.277 64,890.81 151.231 0 1,084,328 Model (3) GPIslack 331 7008.761 38,205.35 0 0 404,856.8 −0.90 GTIslack 331 12,561.97 104,856.4 330.988 0 1,553,757 Notes: All models are output-oriented and VRS models based on the SBM approach using two inputs – GPI and GTI. The major diﬀerences among these models are their outputs. Model (1) has ﬁve undesirable outputs – COD, NH N, SO , 3 2 NO , Soot, and uses a sample consisting of ﬁrm-years that disclose these ﬁve pollutant emissions at the same time. Model (2) has four undesirable outputs – COD, NH N, SO ,NO , and its sample is formed from ﬁrm-years that disclose 3 2 X these four pollutant emissions in the meantime. Model (3) uses COD and SO as undesirable outputs, and its sample is constitutive of ﬁrm-years that disclose COD and SO emissions at the same time. We conduct T-tests to examine the diﬀerence between the mean of GPIslack and GTIslack. 78 Y. CHEN AND J. FENG 5.2. Tobit regression In the second stage analysis, the Tobit regression is applied to examine the association between environmental enforcement of local governments and corporate green invest- ment eﬃciency. The sample for regression estimation is the same as the sample used in SBM-DEA model (3), which is comprised of ﬁrm-years that disclose COD and SO emissions at the same time. In Table 7, we provide descriptive statistics of the main variables used in the Tobit regression. GIE is calculated by SBM-DEA model (3). The mean of Enforcement is 56.831, which is smaller than the median (59.300), implying that a part of sample ﬁrms are facing weak environmental enforcement of local governments. Besides, the standard deviation of Enforcement is 13.578, demonstrating that local environmental enforcement is quite diﬀerent across sample ﬁrms and local governments have greater discretion in implementing the same environmental regulations. We observe reasonable variation in all control variables. Table 8 reports the pairwise correlations of main variables included in the Tobit regression. We ﬁnd positive Pearson correlation and negative Spearman correlation between Enforcement and GIE, probably because we do not control for other factors and omit the square term of Enforcement. These correlations show that there are no serious multicollinearity issues among variables. 5.2.1. The eﬀect of local environmental enforcement on corporate green investment eﬃciency Table 9 reports the results for testing the eﬀect of local environmental enforcement on corporate green investment eﬃciency. In Column 1, we only control for the internal factors including ﬁrm characteristics, ﬁnancial performance, previous green investment scale, investment opportunity, and corporate governance. The coeﬃcient on Enforcement is signiﬁcantly positive and the coeﬃcient on the square term of Enforcement is signiﬁcantly Table 7. Descriptive statistics of main variables used in Tobit regression. Variable N Mean Std. Dev. Median Min Max GIE 331 0.562 0.148 0.503 0.500 1.000 Enforcement 331 56.831 13.578 59.300 24.600 79.600 Size 331 22.913 1.455 22.818 20.151 28.509 SOE 331 0.492 0.501 0.000 0.000 1.000 Age 331 18.453 4.815 18.000 7.000 34.000 GIscale 331 0.008 0.017 0.002 0.000 0.157 Leverage 331 0.431 0.214 0.426 0.047 1.229 OCF 331 0.069 0.060 0.068 −0.108 0.243 Opportunity 331 1.875 1.559 1.388 0.113 8.607 ROA 331 0.052 0.061 0.045 −0.250 0.340 Growth 331 0.314 1.247 0.183 −0.544 21.886 LSR 331 0.372 0.157 0.359 0.096 0.891 Duality 331 0.254 0.436 0.000 0.000 1.000 DAGE 331 52.116 3.079 52.111 42.385 62.111 PGDP 331 66,937.370 27,566.900 60,199.000 27,643.000 128,994.100 Structure 331 0.349 0.072 0.380 0.078 0.421 ENpressure 331 49.463 13.899 48.200 18.600 81.300 ENsupervision 331 0.502 0.501 1.000 0.000 1.000 We also exclude ﬁrm-years that have no green investments during the sample period. Our ﬁnal sample for Tobit regression has 331 ﬁrm-years, including 96 observations in 2016 and 235 observations in 2017. CHINA JOURNAL OF ACCOUNTING STUDIES 79 Table 8. Pairwise correlations of main variables included in Tobit regression. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 1.GIE −0.01 −0.60* −0.38* −0.12* −0.28* −0.47* −0.10 0.61* 0.21* −0.11* −0.19* 0.17* −0.19* −0.05 0.05 −0.07 −0.01 2.Enforcement 0.05 0.01 −0.08 −0.13* −0.07 −0.13* 0.03 0.07 0.18* 0.04 −0.03 0.02 0.12* 0.67* 0.23* 0.26* 0.09 3.Size −0.26* 0.04 0.42* 0.13* 0.09 0.61* 0.06 −0.74* −0.20* 0.16* 0.33* −0.18* 0.36* 0.04 −0.12* 0.16* 0.02 4.SOE −0.21* −0.07 0.43* 0.22* 0.13* 0.38* −0.01 −0.47* −0.36* −0.11* 0.21* −0.27* 0.23* −0.08 −0.11 0.16* −0.02 5.Age −0.06 −0.13* 0.09 0.20* 0.07 0.21* −0.07 −0.25* −0.20* −0.09 −0.16* −0.02 −0.02 −0.13* 0.00 −0.05 0.03 6.GIscale −0.09 0.05 −0.04 0.05 0.02 0.18* 0.08 −0.20* −0.10 0.13* −0.08 −0.03 −0.14* −0.04 0.07 0.02 0.09 7.Leverage −0.27* −0.15* 0.53* 0.39* 0.21* 0.01 −0.11 −0.73* −0.45* 0.13* 0.10 −0.11* 0.09 −0.10 −0.08 0.14* −0.01 8.OCF −0.03 0.02 0.06 −0.03 −0.04 −0.04 −0.13* 0.13* 0.49* 0.13* 0.14* 0.08 0.02 0.01 0.04 0.00 −0.00 9.Opportunity 0.31* 0.05 −0.59* −0.37* −0.17* −0.06 −0.57* 0.16* 0.50* 0.00 −0.19* 0.18* −0.27* 0.02 0.06 −0.19* −0.00 10.ROA 0.06 0.15* −0.10 −0.26* −0.17* −0.04 −0.42* 0.46* 0.35* 0.30* 0.02 0.15* −0.05 0.19* 0.12* −0.07 0.03 11.Growth −0.01 0.05 0.06 −0.09 −0.03 −0.02 0.08 −0.01 −0.02 0.06 0.07 −0.05 −0.06 −0.01 0.07 −0.02 0.08 12.LSR −0.04 −0.02 0.40* 0.24* −0.18* −0.08 0.09 0.15* −0.11* 0.05 0.07 −0.12* 0.15* 0.04 −0.13* 0.09 −0.06 13.Duality 0.12* 0.02 −0.20* −0.27* −0.02 −0.02 −0.12* 0.08 0.15* 0.10 0.08 −0.14* −0.05 −0.02 −0.04 −0.12* −0.04 14.DAGE −0.04 0.12* 0.39* 0.23* −0.04 −0.11 0.06 0.04 −0.23* −0.00 0.03 0.21* −0.03 0.16* −0.05 0.27* −0.02 15.PGDP −0.02 0.65* 0.12* −0.06 −0.15* −0.00 −0.10 −0.01 −0.05 0.15* −0.05 0.12* −0.01 0.19* 0.21* 0.24* −0.04 16.Structure −0.02 0.09 −0.21* −0.14* 0.09 0.11 −0.07 0.05 0.06 0.06 0.00 −0.24* −0.05 −0.17* −0.03 0.29* −0.13* 17.ENpressure 0.01 0.28* 0.23* 0.17* −0.05 0.01 0.13* −0.03 −0.12* −0.08 0.03 0.13* −0.12* 0.28* 0.23* 0.06 −0.05 18.ENsupervision −0.02 0.07 0.02 −0.02 0.01 0.03 0.00 −0.00 0.00 0.02 −0.05 −0.04 −0.04 0.00 −0.07 −0.02 −0.05 Note: We present the pairwise correlations of main variables included in second-stage Tobit regression, using the sample of SBM-DEA model (3), and report Pearson correlations in the lower triangle and Spearman correlations in the upper triangle. *indicates that the correlation coeﬃcients are statistically signiﬁcant at 5% level. 80 Y. CHEN AND J. FENG Table 9. Tobit regression analysis – the eﬀect of local environmental enforcement on corporate green investment eﬃciency. Dependent Variable = GIE Large- Full Sample SOEs Non-SOEs scale Small-scale (1) (2) (3) (4) (5) (6) (7) Enforcement 0.0089** 0.0139*** 0.0010 0.0038 0.0254*** 0.0065 0.0239*** (2.21) (2.95) (1.22) (0.78) (3.07) (1.39) (2.76) Enforcement2 −0.0001* −0.0001*** −0.0000 −0.0002*** −0.0000 −0.0002** (−1.93) (−2.75) (−0.56) (−2.97) (−1.19) (−2.49) Size −0.0111 −0.0137 −0.0143 −0.0041 −0.0273 (−1.04) (−1.28) (−1.32) (−0.49) (−1.17) SOE −0.0377* −0.0450** −0.0423** 0.0035 −0.0792** (−1.91) (−2.32) (−2.22) (0.18) (−2.38) Age 0.0010 0.0015 0.0015 0.0056** −0.0026 0.0040* −0.0019 (0.47) (0.67) (0.68) (2.09) (−0.86) (1.71) (−0.55) GIscale −0.5435 −0.4450 −0.5317* −0.2926 −1.6747** −0.3272 −0.6312 (−1.63) (−1.34) (−1.70) (−1.01) (−2.45) (−1.32) (−0.94) Leverage −0.0948 −0.1000 −0.0974 −0.1050 −0.0815 −0.1087 −0.1419 (−1.49) (−1.61) (−1.55) (−1.30) (−0.63) (−1.61) (−1.61) OCF −0.2061 −0.1803 −0.1530 −0.2249 −0.2196 −0.0563 −0.3424 (−1.25) (−1.11) (−0.95) (−1.37) (−0.82) (−0.40) (−1.26) Opportunity 0.0176** 0.0158* 0.0170** 0.0152 0.0092 0.0196 0.0104 (2.01) (1.84) (1.97) (1.37) (0.66) (1.14) (0.91) ROA −0.1441 −0.1232 −0.1408 0.0852 −0.2606 −0.0434 −0.1217 (−0.54) (−0.50) (−0.56) (0.23) (−0.70) (−0.21) (−0.32) Growth −0.0030 −0.0023 −0.0023 −0.0181 −0.0047 −0.0009 0.0088 (−0.70) (−0.53) (−0.54) (−0.50) (−0.80) (−0.37) (0.14) LSR 0.0499 0.0330 0.0244 0.0876 −0.0516 −0.0070 0.0743 (0.73) (0.48) (0.35) (1.47) (−0.37) (−0.13) (0.51) Duality 0.0188 0.0144 0.0162 0.0080 0.0062 0.0164 0.0068 (0.80) (0.63) (0.70) (0.36) (0.19) (0.71) (0.17) DAGE 0.0025 0.0014 0.0010 −0.0066* 0.0051 −0.0062 0.0073 (0.65) (0.36) (0.27) (−1.78) (0.86) (−1.34) (1.26) PGDP −0.0000 −0.0000 0.0000 −0.0000 0.0000 −0.0000 (−0.47) (−0.28) (0.90) (−0.96) (0.36) (−0.40) Structure −0.3373* −0.1821 −0.0540 −0.5292 −0.0254 −0.6325 (−1.88) (−1.07) (−0.40) (−1.64) (−0.32) (−1.65) ENpressure 0.0011 0.0007 0.0002 0.0026 0.0012* 0.0020 (1.33) (0.90) (0.39) (1.56) (1.73) (1.13) ENsupervision −0.0031 −0.0092 0.0107 −0.0205 −0.0022 −0.0077 (−0.19) (−0.57) (0.75) (−0.65) (−0.15) (−0.24) Intercept 0.4707* 0.5907** 0.8852*** 0.7319*** 0.6254 0.5453* −0.0681 (1.89) (2.27) (3.38) (2.72) (1.21) (1.96) (−0.17) Area/Industry/Year ﬁxed eﬀects YES YES YES YES YES YES YES F 2.03 1.77 1.72 0.82 1.64 0.56 1.71 Observations 331 331 331 163 168 165 166 Notes: We report the Tobit estimations with robust standard errors clustered at the ﬁrm level and present t-statistics in parentheses. ***, **, and * indicate statistical signiﬁcance at the 1%, 5%, and 10% levels, respectively. negative. In Column 2, we further control for the external environment that sample ﬁrms are faced with. Similarly, we ﬁnd that the coeﬃcient on Enforcement is signiﬁcantly positive at 1% level, and the coeﬃcient on the square term of Enforcement is signiﬁcantly negative at 1% level. Contrastively, we estimate Equation (a) without the square term of Enforcement and Column 3 shows that the coeﬃcient on Enforcement is positive but insigniﬁcant. Taken together, we conclude that the local environmental enforcement has an inverted U-shaped eﬀect on green investment eﬃciency of polluting ﬁrms. Beyond that, we interpret these CHINA JOURNAL OF ACCOUNTING STUDIES 81 signiﬁcantly positive coeﬃcients on Enforcement as indicating that the current intensity of local environmental enforcement in China lies on the left side of this inverted U-shaped curve, which means that the increasingly stringent environmental enforcement of local governments is conducive to improving the GIE of polluting ﬁrms. These ﬁndings provide evidence that Chinese governments should further optimize environmental supervision methods and strengthen environmental enforcement to enhance the eﬃciency of corpo- rate green investments. Next, we divide sample ﬁrms into two groups based on whether they are state-owned and report the results in Column 4 and Column 5 of Table 9. For SOEs, Column 4 shows that the coeﬃcients on Enforcement and the square term of Enforcement are statistically insigniﬁcant. However, for Non-SOEs, we ﬁnd the coeﬃcient on Enforcement is positive and the coeﬃcient on the square term of Enforcement is negative, both are statistically signiﬁcant at 1% level. Considering all of this evidence, it suggests that the inverted U-shaped relationship between local environmental enforcement and the GIE of pollut- ing ﬁrms is only pronounced in the Non-SOE group. These results are reasonable because the corresponding penalties of local environmental enforcement and environ- mental reputation have a greater impact on non-state-owned enterprises. At the same time, Non-SOEs have more limited resources to invest in environmental dimensions, thus, they would pay more attention to improving the eﬃciency of green investments. Furthermore, we group our sample ﬁrms into two types based upon their size. To be speciﬁc, we classify the ﬁrms with above sample median value of Size as ‘Large-scale’ ﬁrms and ‘Small-scale’ ﬁrms otherwise. The results in Column 6 and Column 7 of Table 9 suggest that the local environmental enforcement has an inverted U-shaped eﬀect on GIE only when the ﬁrms belong to the Small-scale group. The main reason for these results is that the coordination costs of small-scale enterprises are relatively low, which is more favorable to green management and eﬃcient allocation of resources. Consequently, the promotion of local environmental enforcement on green investment eﬃciency is more obvious. 5.2.2. Supplementary analyses: local environmental enforcement and excessive corporate green investments From the ﬁrst stage analyses of SBM-DEA results, we ﬁnd that the green investment eﬃciency of polluting companies is overall low primarily due to excessive green invest- ments. This essentially means that polluting ﬁrms have to focus on solving the problem of redundant green investments if they intend to enhance their green investment eﬃciency. Since we have documented that local environmental enforcement exerts a signiﬁcant inverted U-shaped impact on the eﬃciency of corporate green investments in Section 5.2.1, it is necessary for local governments to further consider promoting polluting ﬁrms’ green investment eﬃciency by restraining their excessive green invest- ments. Therefore, we conduct supplementary analyses to examine whether increasingly strict local environmental enforcement is accompanied with a decrease in redundancy of corporate green investments, which would oﬀer a useful reference for the Chinese government’s current environmental regulatory reform. The slack variables for GPI and GTI calculated by SBM-DEA model (3) measure the excessive prevention investments (GPIslack) and excessive treatment investments (GTIslack), respectively. As we ﬁnd that there is no signiﬁcant diﬀerence between 82 Y. CHEN AND J. FENG GPIslack and GTIslack in Section 5.1.3, we use the total amounts of GPIslack and GTIslack, namely GIslack, as the dependent variable. Because the values for GIslack of sample ﬁrms that have no excessive green investments are all zero, which are censored data, we also apply the Tobit regression to estimate Equation (b) that is speciﬁed as follows: LogðÞ 1 þ GIslack¼ αþβ Enforcement þ β Enforcement þ β Size þ β SOE þ β Age 1 2 3 4 5 þβ GIscale þ β Leverage þ β OCF þ β Opportunity þ β ROA 6 7 8 9 10 þ β Growth þ β LSR þ β Duality þ β DAGE þ β PGDP 11 12 13 14 15 þ β Structure þ β ENpressure þ β ENsupervision þ Area fixed effects 16 17 18 þ Industry fixed effects þ Year fixed effects þ ε (b) where GIslack represents the total excessive green investments. All control variables are similar to the variables used in Equation (a). We estimate Equation (b) separately for the full sample and subsamples. For the full sample, Column 1 and Column 2 of Table 10 report that the coeﬃcients on Enforcement are signiﬁcantly negative and the coeﬃcients on the square term of Enforcement are signiﬁcantly positive. By comparison, we estimate Equation (b) without the square term of Enforcement and Column 3 demonstrates that the coeﬃcient on Enforcement is negative but insigniﬁcant. These results suggest that there is a U-shaped eﬀect of local environmental enforcement on excessive green investments. In terms of the signiﬁcantly negative coeﬃcients on Enforcement in Column 1 and Column 2, we conclude that the current intensity of local environmental enforcement in China lies on the left side of this U-shaped curve, which means that the stricter environmental enforcement of local governments can signiﬁcantly restrain excessive green investments of polluting ﬁrms. Therefore, we believe that curbing corporate green investment redundancy is an eﬀective regulatory approach for local governments to enhance the green investment eﬃciency of polluting enterprises. For our subsamples, in Column 4 and Column 5 of Table 10,we ﬁnd the U-shaped relationship between local environmental enforcement and excessive green investments is only pronounced in Non-SOEs. Besides, Column 6 and Column 7 exhibit that local environmental enforcement has a statistically signiﬁcant U-shaped eﬀect only on the small-scale ﬁrms’ excessive green investments. These ﬁndings are consistent with our preceding results that the inverted U-shaped relationship between local environmental enforcement and GIE is only statistically signiﬁcant in non-state-owned enterprises and small-scale enterprises, indicating that local governments could improve GIE of these ﬁrms by restraining their excessive green investments. As for SOEs and large-scale ﬁrms, local governments should make more use of market-based means, such as environ- mental taxes and tradable emission permit, encourage managers to actively take corpo- rate social responsibilities instead of passively catering to government regulation, and guide the majority of stakeholders to play a supervisory role. 5.2.3. Robustness tests In this section, we conduct a series of robustness tests to assure that our primary ﬁndings on the inverted U-shaped relationship between local environmental enforce- ment and corporate green investment eﬃciency are robust to the alternative sample, CHINA JOURNAL OF ACCOUNTING STUDIES 83 Table 10. Local environmental enforcement and excess corporate green investments. Dependent Variable = Log(1+ GIslack) Large- Full Sample SOEs Non-SOEs scale Small-scale (1) (2) (3) (4) (5) (6) (7) Enforcement −0.4279*** −0.5778*** −0.0468 −0.3507 −1.0122*** −0.2439 −1.1292*** (−2.74) (−3.37) (−1.46) (−1.31) (−3.51) (−0.98) (−3.65) Enforcement2 0.0037** 0.0050*** 0.0028 0.0088*** 0.0021 0.0095*** (2.51) (3.13) (1.11) (3.30) (0.90) (3.34) Size 0.9718*** 0.9734*** 0.9823*** 1.0770*** 1.2102* (3.02) (3.02) (3.02) (2.96) (1.79) SOE 0.4591 0.6890 0.5950 −0.6765 1.7216* (0.64) (0.96) (0.82) (−0.66) (1.67) Age −0.0825 −0.0880 −0.0930 −0.0446 −0.0677 −0.1058 −0.0693 (−1.23) (−1.31) (−1.37) (−0.43) (−0.74) (−1.11) (−0.72) GIscale 24.0560 24.4617 27.5132 8.5194 60.4735** 21.1233 28.7209 (1.10) (1.10) (1.29) (0.34) (2.01) (0.66) (0.87) Leverage 2.9308 2.9096 2.8723 2.3695 2.2691 3.8302 1.6468 (1.37) (1.39) (1.35) (0.92) (0.56) (1.18) (0.60) OCF 15.1716*** 15.7160*** 14.4933** 8.7828 20.2882* 14.1440* 20.6909** (2.59) (2.68) (2.46) (1.32) (1.92) (1.97) (2.04) Opportunity −1.0431*** −1.0350*** −1.0979*** −0.8152* −0.9577** −0.7394 −1.4690*** (−3.22) (−3.37) (−3.55) (−1.80) (−2.11) (−1.22) (−4.12) ROA 14.1512* 13.0407* 14.4387* 14.5701* 7.1876 0.8787 14.5125 (1.87) (1.75) (1.88) (1.70) (0.52) (0.07) (1.28) Growth 0.1226 0.1491 0.1471 0.6663 0.2256 0.0327 0.9519 (1.11) (1.35) (1.29) (0.58) (1.36) (0.26) (0.67) LSR −2.6394 −2.6600 −2.4163 −3.3735 −2.2050 −0.3148 −2.4276 (−1.26) (−1.26) (−1.13) (−1.41) (−0.56) (−0.13) (−0.60) Duality 0.6198 0.7697 0.7585 −0.3403 1.8099* 0.0849 2.3558** (0.85) (1.06) (1.04) (−0.28) (1.97) (0.09) (2.07) DAGE −0.0731 −0.0798 −0.0660 −0.0013 −0.0781 0.1458 −0.2050 (−0.63) (−0.68) (−0.56) (−0.01) (−0.46) (0.90) (−1.27) PGDP 0.0001** 0.0000** 0.0001* 0.0001* 0.0000 0.0001*** (2.58) (2.21) (1.87) (1.95) (0.60) (2.92) Structure 13.1153** 5.6017 8.3163 18.5583* 0.9766 23.9084** (2.49) (1.16) (1.20) (1.82) (0.13) (1.98) ENpressure −0.0042 0.0094 −0.0139 0.0108 −0.0168 0.0354 (−0.17) (0.39) (−0.44) (0.24) (−0.58) (0.76) ENsupervision 0.5080 0.7660 0.2845 0.8671 0.5543 0.6204 (0.93) (1.37) (0.42) (0.97) (0.86) (0.63) Intercept 1.5907 −3.6222 −14.7038 −11.8404 −3.8917 7.9823 27.0331** (0.17) (−0.37) (−1.60) (−0.97) (−0.22) (0.64) (2.23) Area/Industry/Year YES YES YES YES YES YES YES ﬁxed eﬀects F 10.49 9.79 9.80 6.48 5.46 2.50 5.37 Observations 331 331 331 163 168 165 166 Notes: We report the Tobit estimations with robust standard errors clustered at the ﬁrm level and present t-statistics in parentheses. ***, **, and * indicate statistical signiﬁcance at the 1%, 5%, and 10% levels, respectively. additional control variables, and alternative measures of local environmental enforcement. In Column 1 of Table 11, we select an alternative sample that is used for the preceding SBM-DEA model (2) and rerun the Tobit regression to estimate Equation (a). The result illustrates that local environmental enforcement has a statistically signiﬁcant inverted U-shaped impact on corporate green investment eﬃciency at 1% level, sug- gesting that our ﬁndings are robust to this alternative sample. 84 Y. CHEN AND J. FENG Table 11. Robustness tests. Dependent Variable = GIE (1) (2) (3) (4) Enforcement 0.0152*** 0.0091** (2.92) (2.15) Enforcement2 −0.0001*** −0.0001** (−2.76) (−2.00) LPAI 0.1478** (1.99) LPAI2 −0.0911*** (−2.64) NPLES 0.0311*** (2.71) NPLES2 −0.0021*** (−3.56) Size −0.0057 −0.0125 −0.0119 −0.0113 (−0.55) (−1.21) (−1.11) (−1.08) SOE −0.0355* −0.0256 −0.0184 −0.0228 (−1.82) (−1.39) (−1.00) (−1.30) Age 0.0010 0.0019 0.0017 0.0022 (0.44) (0.91) (0.87) (1.07) GIscale −0.3164 −0.5223* −0.7090** −0.6188** (−1.01) (−1.69) (−2.11) (−1.99) Leverage −0.0748 −0.1060* −0.1024* −0.1045 (−1.20) (−1.73) (−1.67) (−1.62) OCF −0.1566 −0.1938 −0.1887 −0.1416 (−0.91) (−1.25) (−1.21) (−0.94) Opportunity 0.0129 0.0138 0.0170** 0.0143* (1.41) (1.63) (2.00) (1.70) ROA 0.1212 −0.1670 −0.1708 −0.1421 (0.59) (−0.71) (−0.73) (−0.60) Growth −0.0014 0.0003 0.0010 0.0009 (−0.31) (0.08) (0.26) (0.22) LSR −0.0224 0.0404 0.0212 0.0320 (−0.32) (0.60) (0.32) (0.47) Duality 0.0055 −0.0066 −0.0113 −0.0094 (0.26) (−0.31) (−0.54) (−0.44) DAGE −0.0016 0.0029 0.0027 0.0034 (−0.40) (0.76) (0.70) (0.85) PGDP −0.0000 −0.0000 0.0000 0.0000 (−0.78) (−0.23) (0.36) (0.95) Structure −0.4974** −0.1596 0.0404 0.0244 (−2.51) (−0.89) (0.26) (0.16) ENpressure 0.0013* 0.0013* 0.0009 0.0019** (1.73) (1.67) (1.07) (2.27) ENsupervision 0.0074 −0.0037 −0.0046 0.0013 (0.41) (−0.23) (−0.29) (0.08) BDsize 0.0024 0.0013 0.0009 (0.58) (0.32) (0.22) Female 0.2165*** 0.2709** 0.2890** (2.86) (2.15) (2.28) EXSR 0.2526** 0.2304*** 0.2201*** (2.00) (3.02) (2.90) ENLAW 0.0104 0.0160 0.0794*** (0.94) (1.46) (3.55) ENpenalty 0.0479 0.0575 0.0486 (1.33) (1.61) (1.33) Intercept 0.5684** 0.4388* 0.5670** 0.3520 (2.10) (1.68) (2.32) (1.35) Area/Industry/Year ﬁxed eﬀects YES YES YES YES F 1.64 2.27 2.24 2.78 Observations 294 331 331 331 Notes: We report the Tobit estimations with robust standard errors clustered at the ﬁrm level and present t-statistics in parentheses. ***, **, and * indicate statistical signiﬁcance at the 1%, 5%, and 10% levels, respectively. CHINA JOURNAL OF ACCOUNTING STUDIES 85 Next, we include extra internal and external control variables in the Tobit regression and report the result in Column 2 of Table 11. Among these additional control variables, we use BDsize, Female, and EXSR to control for internal ﬁrm-level factors. BDsize is the number of directors on the board of the ﬁrm for ﬁscal year t. Female is the proportion of the ﬁrm’s female directors to the total number of directors for ﬁscal year t. EXSR is the ratio of equity shares held by executives to total equity shares of the ﬁrm at the end of ﬁscal year t. In addition, we use ENLAW and ENpenalty to further control for the external environment of sample ﬁrms. ENLAW is the sum of local environmental laws and regulations scaled by local industrial output for ﬁscal year t-1. The indicator variable ENpenalty is 1 if the ﬁrm has been punished for environmental violations, etc. in previous years and 0 otherwise. We ﬁnd that the coeﬃcient on Enforcement is signiﬁcantly positive and the coeﬃcient on the square term of Enforcement is signiﬁcantly negative, which is consistent with our preceding result and further conﬁrm the inverted U-shaped relationship between local environmental enforcement and green investment eﬃciency of polluting ﬁrms. Moreover, to address the concern that our primary ﬁndings might result from the measure of local environmental enforcement and omitted correlated variables, we use two alternative measures – LPAI and NPLES – to capture the stringency of local environ- mental regulation, and include the above-mentioned extra control variables to rerun the Tobit regression. Following Keller and Levinson (2002) and Henderson and Millimet (2007), we use LPAI, which is local industrial pollution abatement investments scaled by the industrial sector’s contribution to the regional GDP of the province where sample ﬁrm is registered for ﬁscal year t, to measure environmental regulation of local govern- ments. Provinces with larger values of LPAI have stricter environmental regulations. We rerun the Tobit regression to estimate Equation (a) with LPAI and the square term of LPAI as the primary independent variables of interest and report the result in Column 3 of Table 11. Similarly, we ﬁnd the coeﬃcient on LPAI is signiﬁcantly positive and the coeﬃcient on the square term of LPAI is signiﬁcantly negative, demonstrating a signiﬁcant inverted U-shaped relationship between the stringency of local environ- mental regulation and corporate green investment eﬃciency. In addition, following Bu, Liu, Wagner, and Yu (2013), we measure the environmental regulation of local govern- ments with NPLES, the number of enforcement personnel in local environmental super- vision scaled by the environmental investments of local governments. A higher value of NPLES indicates ﬁrms that are located in this province are facing stronger environmental regulation of local governments. NPLES is lagged by one year when we conduct the Tobit regression. Column 4 of Table 11 reports that the coeﬃcient on NPLES is positive and the coeﬃcient on the square term of NPLES is negative, both are statistically signiﬁcant at 1% level, which still indicates that local environmental regulation has an inverted U-shaped eﬀect on GIE of polluting ﬁrms. Thus, our primary results are robust to these alternative measures of local environmental enforcement and extra control variables. 6. Conclusion Corporate green investment eﬃciency plays a leading role in achieving sustainable development and maximizing social value for polluting ﬁrms. We assess ﬁrm-level 86 Y. CHEN AND J. FENG green investment eﬃciency and explore the association between local environmental enforcement and green investment eﬃciency. Applying SBM-DEA approach with green investments as inputs and pollutant emissions as undesirable outputs in the ﬁrst stage, we quantify and evaluate green investment eﬃciency of Chinese listed companies in polluting industries in 2016 and 2017, ﬁnding that the corporate green investment eﬃciency is overall low, primarily due to excessive green investments. The results indicate that managers may only invest extensively in environmental dimensions for compliance reasons and neglect the eﬃcient allocation of resources to reduce pollutant emissions, which is a waste of resources to some extent. In the second stage, we conduct the Tobit regression to examine the relationship between local environmental enforcement and green investment eﬃciency of polluting ﬁrms. We ﬁrst ﬁnd that local environmental enforcement has an inverted U-shaped eﬀect on corporate green investment eﬃciency. Notably, this eﬀect is only statistically signiﬁcant in non-state-owned enterprises and small-scale enterprises. Furthermore, we document that there exists a U-shaped eﬀect of local environmental enforcement on excessive green investments, and this eﬀect is only pronounced in Non-SOEs and small- scale ﬁrms as well. Our results highlight that managers should recognize the importance of identifying and reducing the low-value allocation of limited resources, and indicate that local governments should optimize corporate green investment eﬃciency through diﬀeren- tiated environmental regulation. Regarding Non-SOEs and small-scale ﬁrms, local gov- ernments could adopt prudent environmental regulation to improve their green investment eﬃciency by restraining excessive green investments. As for SOEs and large- scale ﬁrms, local governments should make more use of market-based means, such as environmental taxes and tradable emission permit, encourage managers to actively take corporate social responsibilities instead of passively catering to government regulation, and guide the majority of stakeholders to play a supervisory role. Our ﬁndings should be of interest to policymakers, managers, and stakeholders involved in the regulation of environmental pollution. Acknowledgments We appreciate the helpful comments and suggestions of the anonymous referee and conference participants at the 23rd Annual Conference of Chinese Finance. We acknowledge ﬁnancial support from the Fundamental Research Funds for the Central Universities (No. JBK1807077). Disclosure statement No potential conﬂict of interest was reported by the authors. Funding This work was supported by the Fundamental Research Funds for the Central Universities [JBK1807077]. CHINA JOURNAL OF ACCOUNTING STUDIES 87 Data availability All data are available from public sources. References Antonietti, R., & Marzucchi, A. (2014). Green tangible investment strategies and export perfor- mance: A ﬁrm-level investigation. Ecological Economics, 108, 150–161. Ateş, M.A., Bloemhof, J., van Raaij, E.M., & Wynstra, F. (2012). Proactive environmental strategy in a supply chain context: The mediating role of investments. International Journal of Production Research, 50(4), 1079–1095. Bahn, O., Chesney, M., & Gheyssens, J. (2012). The eﬀect of proactive adaptation on green investment. Environmental Science & Policy, 18(4), 9–24. Bostian, M., Färe, R., Grosskopf, S., & Lundgren, T. (2016). Environmental investment and ﬁrm performance: A network approach. Energy Economics, 57, 243–255. Broberg, T., Marklund, P.-O., Samakovlis, E., & Hammar, H. (2013). Testing the Porter hypothesis: The eﬀects of environmental investments on eﬃciency in Swedish industry. Journal of Productivity Analysis, 40(1), 43–56. Bu, M., Liu, Z., Wagner, M., & Yu, X. (2013). Corporate social responsibility and the pollution haven hypothesis: Evidence from multinationals’ investment decision in China. Asia-Paciﬁc Journal of Accounting & Economics, 20(1), 85–99. Burnett, R.D., & Hansen, D.R. (2008). Ecoeﬃciency: Deﬁning a role for environmental cost management. Accounting Organizations & Society, 33(6), 551–581. Carroll, A.B. (1991). The pyramid of corporate social responsibility: Toward the moral management of organizational stakeholders. Business Horizons, 34(4), 39–48. Chang, Y.T., Zhang, N., Danao, D., & Zhang, N. (2013). Environmental eﬃciency analysis of transportation system in China: A non-radial DEA approach. Energy Policy, 58(9), 277–283. Charnes, A., Cooper, W.W., & Rhodes, E. (1978). Measuring the eﬃciency of decision making units. European Journal of Operational Research, 2(6), 429–444. Cho, C.H., Guidry, R.P., Hageman, A.M., & Patten, D.M. (2012). Do actions speak louder than words? An empirical investigation of corporate environmental reputation. Accounting, Organizations and Society, 37(1), 14–25. Costa-Campi, M.T., García-Quevedo, J., & Martínez-Ros, E. (2017). What are the determinants of investment in environmental R&D? Energy Policy, 104, 455–465. Delmas, M., & Toﬀel, M.W. (2004). Stakeholders and environmental management practices: An institutional framework. Business Strategy and the Environment, 13(4), 209–222. Doval, E., & Negulescu, O. (2014). A model of green investments approach. Procedia Economics and Finance, 15, 847–852. Emrouznejad,A.,&Yang,G.L.(2017). A survey and analysis of the ﬁrst 40 years of scholarly literature in DEA: 1978–2016.Socio-Economic Planning Sciences.doi:10.1016/j. seps.2017.01.008 Eyraud, L., Clements, B., & Wane, A. (2013). Green investment: Trends and determinants. Energy Policy, 60, 852–865. Figge, F., & Hahn, T. (2013). Value drivers of corporate eco-eﬃciency: Management accounting information for the eﬃcient use of environmental resources. Management Accounting Research, 24(4), 387–400. Henderson, D.J., & Millimet, D.L. (2007). Pollution abatement costs and foreign direct investment inﬂows to u.s. states: a nonparametric reassessment. Review of Economics & Statistics, 89(1), 178–183. Henriques, I., & Sadorsky, P. (1996). The determinants of an environmentally responsive ﬁrm: an empirical approach. Journal of Environmental Economics and Management, 30(3), 381–395. Hua, Z., Bian, Y., & Liang, L. (2007). Eco-eﬃciency analysis of paper mills along the Huai River: An extended DEA approach. Omega, 35(5), 578–587. 88 Y. CHEN AND J. FENG Huang, X.B., & Watson, L. (2015). Corporate social responsibility research in accounting. Journal of Accounting Literature, 34,1–16. Inderst, G., Kaminker, C., & Stewart, F. 2012.Deﬁning and measuring green investments: implica- tions for institutional investors’ asset allocations. OECD Working Papers on Finance, Insurance and Private Pensions No. 24, OECD Publishing. doi:10.1787/5k9312twnn44-en Jiang, L., Folmer, H., & Bu, M. (2016). Interaction between output eﬃciency and environmental eﬃciency: Evidence from the textile industry in Jiangsu Province, China. Journal of Cleaner Production, 113, 123–132. Karásek, J., & Pavlica, J. (2016). Green investment scheme: experience and results in the czech republic. Energy Policy, 90, 121–130. Keller, W., & Levinson, A. (2002). Pollution abatement costs and foreign direct investment inﬂows to U.S. states. Review of Economics & Statistics, 84(4), 691–703. Kim, E.H. (2013). Deregulation and diﬀerentiation: Incumbent investment in green technologies. Strategic Management Journal, 34(10), 1162–1185. Kim, K.T., Lee, D.J., Park, S.J., Zhang, Y., & Sultanov, A. (2015). Measuring the eﬃciency of the investment for renewable energy in Korea using data envelopment analysis. Renewable & Sustainable Energy Reviews, 47, 694–702. Levinson, A. (1996). Environmental regulations and manufacturers’ location choices: Evidence from the Census of Manufactures. Journal of Public Economics, 62(1), 5–29. Li, H., Fang, K., Yang, W., Wang, D., & Hong, X. (2013). Regional environmental eﬃciency evaluation in China: Analysis based on the Super-SBM model with undesirable outputs. Mathematical and Computer Modelling, 58(5–6), 1018–1031. Liu, J.S., Lu, L.Y.Y., Lu, W.-M., & Lin, B.J.Y. (2013). Data envelopment analysis 1978–2010: A citation- based literature survey. Omega, 41(1), 3–15. Lundgren, T., & Zhou, W. (2017). Firm performance and the role of environmental management. Journal of Environmental Management, 203(Pt 1), 330–341. Maggioni, D., & Santangelo, G.D. (2017). Local environmental non-proﬁt organizations and the green investment strategies of family ﬁrms. Ecological Economics, 138, 126–138. Martin, P.R., & Moser, D.V. (2016). Managers’ green investment disclosures and investors’ reaction. Journal of Accounting and Economics, 61(1), 239–254. Maxwell, J.W., & Decker, C.S. (2006). Voluntary environmental investment and responsive regulation. Environmental and Resource Economics, 33(4), 425–439. Murillo-Luna, J.L., Garcés-Ayerbe, C., & Rivera-Torres, P. (2008). Why do patterns of environmental response diﬀer? A stakeholders’ pressure approach. Strategic Management Journal, 29(11), 1225–1240. Pearce, D.W., Markandya, A., & Barbier, E. (1989). Blueprint for a green economy. London, United Kingdom: Routledge. Porter, M.E., & van der Linde, C. (1995). Toward a new conception of the environment-competitiveness relationship. Journal of Economic Perspectives, 9(4), 97–118. Schaltenbrand, B., Foerstl, K., Azadegan, A., & Lindeman, K. (2016). See what we want to see? The eﬀects of managerial experience on corporate green investments. Journal of Business Ethics. doi:10.1007/s10551-016-3191-x Sengupta, A. (2012). Investment in cleaner technology and signaling distortions in a market with green consumers. Journal of Environmental Economics and Management, 64(3), 468–480. Shen, H., & Feng, J. (2012). Media monitoring, government supervision, and corporate environ- mental disclosure. Accounting Research, 20(02), 72–78+97. in Chinese. Singal, M., Parsa, H.G., Segarraoña, M., Jang, S., Chen, R., & Singh, A.J. (2014). The link between ﬁrm ﬁnancial performance and investment in sustainability initiatives. Cornell Hospitality Quarterly, 55(1), 19–30. Song, M., & Zhou, Y. (2016). Quantitative analysis of foreign trade and environmental eﬃciency in China. Emerging Markets Finance and Trade, 52(7), 1647–1660. Song, Y., Yao, S., Yu, D., & Shen, Y. (2017). Risky multi-criteria group decision making on green capacity investment projects based on supply chain. Journal of Business Economics & Management, 18(3), 355–372. CHINA JOURNAL OF ACCOUNTING STUDIES 89 Stoever, J., & Weche, J.P. (2018). Environmental regulation and sustainable competitiveness: evaluating the role of ﬁrm-level green investments in the context of the porter hypothesis. Environmental and Resource Economics, 70(2), 429–455. Tone, K. (1997). Several algorithms to determine multipliers for use in cone-ratio envelopment approaches to eﬃciency evaluations in DEA. In H. Amman, B. Rustem & A. Whinston Computational Approaches to Economic Problems(pp. 91–109). Boston, MA: Springer. Tone, K. (2001). A slacks-based measure of eﬃciency in data envelopment analysis. European Journal of Operational Research, 130(3), 498–509. Voica, M.C., Panait, M., & Radulescu, I. (2015). Green investments – between necessity, ﬁscal constraints and proﬁt. Procedia Economics and Finance, 22,72–79. Wang, C., Wu, J., & Zhang, B. (2018a). Environmental regulation, emissions and productivity: Evidence from Chinese COD-emitting manufacturers. Journal of Environmental Economics and Management, 92,54–73. Wang, R., Wijen, F., & Heugens, P.P.M.A.R. (2018b). Government’s green grip: Multifaceted state inﬂuence on corporate environmental actions in China. Strategic Management Journal, 39(2), 403–428. Yu, Y., Huang, J., & Luo, N. (2018). Can more environmental information disclosure lead to higher eco-eﬃciency? Evidence from China. Sustainability, 10(2), 528. Zhang, B., Chen, X., & Guo, H. (2018). Does central supervision enhance local environmental enforcement? Quasi-experimental evidence from China. Journal of Public Economics, 164, 70–90. Zhao, X., Zhao, Y., Zeng, S., & Zhang, S. (2015). Corporate behavior and competitiveness: Impact of environmental regulation on Chinese ﬁrms. Journal of Cleaner Production, 86, 311–322. Zheng, Y. (2007). De Facto Federalism in China: Reforms and dynamics of central-local relations. Singapore: World Scientiﬁc Publishing. 90 Y. CHEN AND J. FENG Appendices Appendix A Table A1. Variable deﬁnitions. Variable Description Data Source Inputs – Corporate green investments by ﬁrm i during ﬁscal year t: GI = the total amounts of corporate green China Security Market and Accounting Research prevention investments and green treatment (CSMAR) database investments. (We deﬁne corporate green investments as internal investments in equipment, technologies, materials, energy and services that can prevent, control and reduce environmental pollution, produce environmental beneﬁts, and reduce environmental costs, with the goals of improving corporate environmental performance, developing green management and reducing environmental risks.) GPI = the amounts of items that are categorized as China Security Market and Accounting Research green prevention investments from the (CSMAR) database increased construction in progress, the increase in R&D expenditure and purchased ﬁxed assets. (Items are related to investments that aﬀect the production process and act to prevent pollution.) GTI = the amounts of items that are categorized as China Security Market and Accounting Research green treatment investments from the (CSMAR) database increased construction in progress, the increase in R&D expenditure, purchased ﬁxed assets and overhead expenses. (Items are related to investments that deal with already emitted pollution and do not aﬀect the actual production processes.) Undesirable outputs – Corporate pollutant emissions during ﬁscal year t of ﬁrm i: COD = annual COD emissions. The emission data is hand collected form ﬁrm N = annual NH N emissions. NH annual reports, corporate social responsibility 3 3 Wastewater = annual Wastewater discharges. reports, sustainability reports, and SO = annual SO emissions. environmental reports 2 2 NO = annual NO emissions. X X Soot = annual Soot emissions. Solid Waste = annual Solid Waste emissions. Corporate green investment eﬃciency for ﬁscal year t of ﬁrm i: GIE = the measure of ﬁrm-level green investment Results from SBM-DEA models eﬃciency calculated by SBM-DEA models. A higher value of GIE indicates more eﬃcient green investments of the sample ﬁrm in terms of reducing undesirable outputs (pollutant emissions). Input = the ratio of the absolute value of input’s slack Ineﬃciency variable to actual input. Output = the ratio of the absolute value of output’s slack Ineﬃciency variable to actual output. GPIslack = the slack variable of green prevention investments calculated by SBM-DEA models, which is equal to the diﬀerence between actual GPI and best target GPI. GTIslack = the slack variable of green treatment investments calculated by SBM-DEA models, which is equal to the diﬀerence between actual GTI and best target GTI. GIslack = the total amounts of GPIslack and GTIslack. (Continued) CHINA JOURNAL OF ACCOUNTING STUDIES 91 Table A1. (Continued). Variable Description Data Source Environmental enforcement of local governments for ﬁscal year t: Enforcement = the measure of local environmental We hand-collect the data from PITI reports issued enforcement which is based on the 2016–2017 by Institute of Public and Environmental and 2017–2018 Pollution Information Aﬀairs (IPE) and Natural Resources Defense Transparency Index (PITI). We match the city Council (NRDC). where the polluting ﬁrm is registered with PITI to get the enforcement score of local environmental regulation for each sample ﬁrm- year observation. A higher value of Enforcement indicates ﬁrms that are located in this city are facing stricter environmental enforcement of local governments. Control variables: Size = the natural logarithm of the ﬁrm’s total assets at the end of ﬁscal year t. SOE = 1 if the ﬁrm is state-owned during ﬁscal year t and 0 otherwise. Age = the number of years from the establishment of China Security Market and Accounting Research the ﬁrm to the sample year t. (CSMAR) database GIscale = the total green investments of the ﬁrm during ﬁscal year t-1 scaled by total assets at the end of ﬁscal year t-1. Leverage = the total debt scaled by total assets, all measured at the end of ﬁscal year t. OCF = the net cash ﬂow generated by the ﬁrm’s operating activities during ﬁscal year t scaled by total assets at the end of ﬁscal year t. Opportunity = the ratio of market value of equity to total assets at the end of ﬁscal year t. ROA = the ﬁrm’s return on assets during ﬁscal year t. Growth = the ﬁrm’s growth rate of operating income during ﬁscal year t. LSR =the ﬁrm’s largest shareholder rate for ﬁscal year t. Duality = 1 if the ﬁrm’s chair of the board is also the ﬁrm’s manager and 0 otherwise. DAGE = the average age of the ﬁrm’s directors. PGDP = the regional GDP per capita of the province China Statistical Yearbook compiled by National where sample ﬁrm is registered for ﬁscal year t. Bureau of Statistics of China Structure = the provincial gross industrial output scaled by the regional GDP of the province where sample ﬁrm is registered for ﬁscal year t. ENpressure = the annual average PM2.5 concentration of the GREENPEACE – the international environmental province where sample ﬁrm is registered for protection organization ﬁscal year t-1. A higher value of ENpressure indicates ﬁrms that are located in this province are facing higher pressure of environmental protection. ENsupervision = 1 if the ﬁrm is registered in the province where The website of Ministry of Ecology and the central government conducted Environment of the People’s Republic of China environmental protection supervisions during (MEEC) ﬁscal year t and 0 otherwise. 92 Y. CHEN AND J. FENG Appendix B CNY 100 million 9575.5 9539.0 9219.8 9037.2 8806.3 8253.5 7612.2 7114.0 5258.4 4937.0 3668.8 2779.5 2565.2 2057.5 1750.1 1456.5 1166.7 1062.0 Year 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Figure B1. Histogram of China’s environmental investments from 2000 to 2017. %GDP 1.86 2.00 1.57 1.54 1.53 1.52 1.49 1.45 1.39 1.38 1.50 1.29 1.29 1.28 1.28 1.24 1.21 1.15 1.06 1.02 1.00 0.50 0.00 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Year Figure B2. The proportion of China’s environmental investments in GDP from 2000 to 2017. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png China Journal of Accounting Studies Taylor & Francis http://www.deepdyve.com/lp/taylor-francis/do-corporate-green-investments-improve-environmental-performance-YNMZ1F0IHH

Do corporate green investments improve environmental performance? Evidence from the perspective of efficiency

Chen, Yutao; Feng, Jian

China Journal of Accounting Studies , Volume 7 (1): 31 – Jan 2, 2019

Download PDF

Share Full Text for Free

31 pages

Do corporate green investments improve environmental performance? Evidence from the perspective of efficiency

Abstract

By linking green investments to corporate environmental performance from the perspective of efficiency, this paper quantifies and evaluates firm-level green investment efficiency, with SBM-DEA approach and hand-collected emission data of Chinese listed companies in polluting industries. We find that corporate green investment efficiency is overall low, primarily due to excessive green investments, suggesting that managers only invest extensively in environmental dimensions without considering...

Loading next page...

Publisher: Taylor & Francis
ISSN: 2169-7221
eISSN: 2169-7213
DOI: 10.1080/21697213.2019.1625578
Publisher site: See Article on Publisher Site

Abstract

CHINA JOURNAL OF ACCOUNTING STUDIES 2019, VOL. 7, NO. 1, 62–92 https://doi.org/10.1080/21697213.2019.1625578 ARTICLE Do corporate green investments improve environmental performance? Evidence from the perspective of eﬃciency Yutao Chen and Jian Feng School of Accounting, Southwestern University of Finance and Economics, Chengdu, China ABSTRACT KEYWORDS Corporate green investment By linking green investments to corporate environmental perfor- eﬃciency; green mance from the perspective of eﬃciency, this paper quantiﬁes and investments; environmental evaluates ﬁrm-level green investment eﬃciency, with SBM-DEA performance; local approach and hand-collected emission data of Chinese listed environmental enforcement; companies in polluting industries. We ﬁnd that corporate green SBM-DEA investment eﬃciency is overall low, primarily due to excessive green investments, suggesting that managers only invest exten- sively in environmental dimensions without considering the eﬃ- cient allocation and value-creating use of limited resources. Further, we conduct the Tobit regression and demonstrate that local environmental enforcement has an inverted U-shaped eﬀect on green investment eﬃciency of polluting ﬁrms. Notably, this eﬀect is only statistically signiﬁcant in Non-SOEs and small-scale enterprises. Moreover, we document a U-shaped relationship between local environmental enforcement and excessive green investments, and this relationship is also only pronounced in Non- SOEs and small-scale ﬁrms. Our ﬁndings indicate that local govern- ments should optimize corporate green investment eﬃciency through diﬀerentiated environmental regulation. 1. Introduction Enterprises are major consumers of resources and major producers of environmental problems. Due to the increasing concern of environmental issues, polluting ﬁrms are facing more and more stringent environmental regulation. The key to solving the contradiction between economic growth and environmental protection lies in incorpor- ating environmental factors into corporate investment decision-making process (Pearce, Markandya, & Barbier, 1989). Green investments, a particular type of corporate social responsibility (CSR) activity that involves the allocation of ﬁnancial and intangible resources of ﬁrms to transform environmental strategies and objectives into corporate actions and higher environmental performance (Ateş, Bloemhof, van Raaij, & Wynstra, 2012; Martin & Moser, 2016), play a vital role in achieving sustainable development and maximizing social value for enterprises. What green investments should essentially aim at is the reduction of environmental pollution. However, in reality, managerial incentives CONTACT Yutao Chen yutao.chen@hotmail.com School of Accounting, Southwestern University of Finance and Economics, 555 Liutai Avenue, Wenjiang District, Chengdu 611130, China Paper accepted by Kangtao Ye. © 2019 Accounting Society of China CHINA JOURNAL OF ACCOUNTING STUDIES 63 for green investments include regulatory preemption, green images creating and pro- duction cost savings (Maxwell & Decker, 2006), and managers usually utilize qualitative environmental disclosure as the channel to manage investors’ impressions with the purpose of mediating the eﬀect of poor environmental performance on corporate reputation (Cho, Guidry, Hageman, & Patten, 2012). Therefore, we call into question whether ﬁrm-level green investments eﬀectively improve corporate environmental performance. The resources of the enterprise are limited, and the green investments could not generate direct economic beneﬁts. To some extent, green investments transfer ﬁrm resources to other outside stakeholders, which leads to the tension between traditional performance objectives and societal value objectives. Under the stakeholder theory, ﬁrms need to strike a balance between shareholder interests and the interests of non- shareholder stakeholders (Carroll, 1991). On the one hand, shareholders are reluctant to bear the opportunity costs of green investments; on the other hand, the majority of stakeholders hope that polluting ﬁrms would invest a lot of money to reduce environ- mental pollution. Thus, it is crucial for polluting ﬁrms to optimize the input-output eﬃciency of green investments. In this paper, by linking green investments to corporate environmental performance from the perspective of eﬃciency, we quantify and conduct an objective and comprehensive assessment of corporate green investment eﬃciency (GIE). Furthermore, there is a consensus among researchers that government is the major determinant of corporate environmental practices (Delmas & Toﬀel, 2004; Henriques & Sadorsky, 1996; Porter & van der Linde, 1995; Stoever & Weche, 2018; Wang, Wu, & Zhang, 2018a; Zhao, Zhao, Zeng, & Zhang, 2015). China has adopted a decentralization regime of environmental regulation. The central government main- tains the authority over the planning, designing, and formulation of environmental policies, while local governments are mainly responsible for the enforcement of speciﬁc environmental policies (Zheng, 2007). Such environmental decentralization gives local governments considerable discretion in implementing the same environmental regula- tions (Zhang, Chen, & Guo, 2018). Thus, we are also interested in the eﬀect of local environmental enforcement on ﬁrm-level green investment eﬃciency. Previous research has primarily concentrated on green investments rather than green investment eﬃciency. A growing body of literature has investigated the determinants of green investment behaviors and strategies (Ateş et al., 2012; Bahn, Chesney, & Gheyssens, 2012; Costa-Campi, García-Quevedo, & Martínez-Ros, 2017;Eyraud,Clements,&Wane, 2013; Kim, 2013; Maggioni & Santangelo, 2017;Schaltenbrand,Foerstl,Azadegan,&Lindeman, 2016; Song, Yao, Yu, & Shen, 2017). Several studies reveal the trends and schemes of green investments (Eyraud et al., 2013; Karásek & Pavlica, 2016), and examine investors’ response to green investments (Martin & Moser, 2016). Another scream of literature explores the relationship between green investments and ﬁrm performance, including environmental performance, productivity, and export performance (Antonietti & Marzucchi, 2014;Bostian, Färe, Grosskopf, & Lundgren, 2016;Lundgren & Zhou, 2017). There is a paucity of empirical research to directly combine corporate green investments with environmental performance from the point of eﬃciency. Most researchers mainly focus on environmental aspects and assess eco-eﬃciency (Burnett & Hansen, 2008; Figge & Hahn, 2013;Hua,Bian,&Liang, 2007; Yu,Huang,&Luo, 2018)orenvironmental eﬃciency (Chang, Zhang, Danao, & Zhang, 2013; Jiang, Folmer, & Bu, 2016; Li, Fang, Yang, Wang, & Hong, 2013; Song & Zhou, 2016). As for 64 Y. CHEN AND J. FENG environmental regulation, recent studies have examined the association of environmental regulation with productivity (Wang et al., 2018a) or corporate environmental actions (Wang, Wijen, & Heugens, 2018b). It is not yet known whether the enforcement of local environ- mental regulation can aﬀect green investment eﬃciency of polluting ﬁrms. In this study, we call researchers’ attention to the green investment eﬃciency by (1) quantifying ﬁrm-level green investment eﬃciency with nonparametric eﬃciency evalua- tion and (2) exploring the association between local environmental enforcement and corporate green investment eﬃciency. We choose Chinese listed companies in polluting industries as our sample for two reasons. First, it is more direct and accurate to use pollutant emissions as proxies for environmental performance, and the availability of emission data is made possible by the pollutant disclosure regime switch in China. In December 2016, China Securities Regulatory Commission (CSRC) revised the information disclosure guidelines for publicly issuing securities, which requires listed companies that belong to National Specially Monitored Firms (NSMF) program to mandatorily disclose detailed pollutants information in their annual reports. Second, to the best of our knowledge, there is seldom empirical evidence for ﬁrm-level GIE of Chinese listed companies that are crucial components of emerging capital markets. Developing coun- tries are facing severe environmental challenges, however, prior research has only to a limited extent explored how governments inﬂuence corporate environmental practices in emerging economies. In the ﬁrst stage, in order to evaluate green investment eﬃciency of polluting ﬁrms, we hand-collect ﬁrm-level emission data and apply Data Envelopment Analysis (DEA) based on slack-based measure (SBM) approach, with greeninvestmentsasinputsand pollutant emis- sions as undesirable outputs, to examine GIE of Chinese listed polluting ﬁrms in 2016 and 2017. DEA is a nonparametric method for measuring the eﬃciency of peer decision-making units (DMUs) with multiple inputs and outputs (Emrouznejad & Yang, 2017). This method is appro- priate to associate diﬀerent types of green investments with varieties of pollutant emissions, and is able to overcome the interference of the subjective factors that traditional performance evaluation methods suﬀer when setting weights and avoid possible bias or errors in the model speciﬁcation. The SBM approach in DEA is ﬁrst proposed by Tone (1997) and further improved by Tone (2001) to deal with undesirable factors as outputs, which allows us to use pollutant emissions as undesirable outputs in SBM-DEA tests. By in-depth analyses of diﬀerent models, weﬁnd that the corporate green investment eﬃciency is overall low, primarily due to excessive green investments. The results indicate that managers only invest extensively in environmen- tal dimensions for compliance reasons, neglecting the eﬃcient allocation and value-creating use of limited resources. In the second stage analysis, we conduct the Tobit regression to examine the relationship between local environmental enforcement and green investment eﬃciency of polluting ﬁrms. We ﬁrst ﬁnd that local environmental enforcement has an inverted U-shaped eﬀect on green investment eﬃciency. Notably, this eﬀect is only statistically signiﬁcant in non-state- owned enterprises and small-scale enterprises. Further, we document that there exists a U-shaped relationship between local environmental enforcement and excessive green investments, and this relationship is only pronounced in Non-SOEs and small-scale ﬁrms as well. Our research provides several contributions. First, our study extends green ﬁnance and environmental accounting literature by demonstrating the importance of green CHINA JOURNAL OF ACCOUNTING STUDIES 65 investment eﬃciency. Previous literature focuses almost exclusively on green investments (Antonietti & Marzucchi, 2014;Ateş et al., 2012; Bahn et al., 2012;Bostian et al., 2016;Doval & Negulescu, 2014; Eyraud et al., 2013; Inderst, Kaminker, & Stewart, 2012; Karásek & Pavlica, 2016;Kim, 2013; Maggioni & Santangelo, 2017; Schaltenbrand et al., 2016; Song et al., 2017; Voica, Panait, & Radulescu, 2015). Nevertheless, we quantify ﬁrm-level green investment eﬃciency and provide the initial evidence, which suggests the main problem of low- eﬃciency ﬁrms is that managers ignore the eﬃcient allocation and value-creating use of resources in terms of reducing pollutant emissions. Second, our study contributes to CSR literature by dealing with the integration of environmental performance into corporate green investment decisions from the perspective of eﬃciency. The environment is asigniﬁcant element of CSR (Huang & Watson, 2015), which means our research should be based on stakeholder theory. Our study adds value to this strand of literature in that we show an eﬀective way to help resolve the conﬂict of interests between shareholders and non-shareholder stakeholders, which is optimizing corporate green investment eﬃciency. Third, our study is related to research that examines the eﬀect of environmental regulation on corporate behaviors and environmental practices. We shed new light on how the local environmental enforcement aﬀects green investment eﬃciency of polluting ﬁrms in emer- ging economies, which helps to pursue the path of environmental regulatory reform to a greater depth and width and furthers our understanding of how to confront environ- mental challenges more eﬀectively. We believe our ﬁndings have implications for managers and policymakers. Our results suggest that managers ought to recognize the importance of identifying and reducing the low-value allocation of limited resource, and indicate that local governments should optimize corporate green investment eﬃciency through diﬀerentiated environmental regulation. Regarding Non-SOEs and small-scale ﬁrms, local governments could adopt prudent environmental regulation to improve their green investment eﬃciency by restraining excessive green investments. As for SOEs and large-scale ﬁrms, local govern- ments should make more use of market-based means, such as environmental taxes and tradable emission permit, encourage managers to actively take corporate social respon- sibilities instead of passively catering to government regulation, and guide the majority of stakeholders to play a supervisory role. The rest of this paper proceeds as follows. Section 2 provides background information about our setting and discusses related research. Section 3 presents the research design. Section 4 describes the sample and provides descriptive statistics. Section 5 provides the SBM-DEA tests and the Tobit regression results. Section 6 concludes. 2. Background and related research 2.1. Background As China’s economic development has entered a new normal situation, the tolerance of the ecological environment has reached or close to the upper limit. The traditional way of economic growth through excessively consuming resources has been unable to continue. China urgently needs to seek a new path of green economy transformation to solve the dilemma between environmental protection and economic growth. In recent years, China has tremendously invested in environmental governance. The total 66 Y. CHEN AND J. FENG amounts of environmental investments increased by eight times from 2000 to 2017 (Figure B1, Appendix B), however, the average proportion of annual environmental investments in GDP is about 1.36% (Figure B2, Appendix B), which is relatively low as for improving environmental conditions. Meanwhile, China has continuously strength- ened environmental regulation in recent years. The central government has conducted environmental protection supervision since January 2016 and has extended to 31 provincial-level administrative zones by the end of 2017. The supervision teams talked with tens of thousands of local politicians and inspected quite a lot of ﬁrms. As a result, these inspections held local governments and oﬃcials accountable for misconduct and ineﬃciency, and tremendous polluting ﬁrms were rectiﬁed and ﬁned. Regarding listed companies, China Securities Regulatory Commission (CSRC) revised the information disclosure guidelines for publicly issuing securities in December 2016, which requires listed companies in polluting industries to mandatorily disclose detailed pollutants information, such as the name of major pollutants, emission concentration and total amounts, excessive discharge, etc., in their annual reports. This regime switch setting of pollutant disclosure provides available emission data that we need as proxies for environmental performance and further quantify ﬁrm-level green investment eﬃciency. 2.2. Related research Prior literature focuses almost exclusively on green investments rather than green investment eﬃciency, including the determinants of green investment behaviors and strategies (Ateş et al., 2012; Bahn et al., 2012; Costa-Campi et al., 2017; Eyraud et al., 2013; Kim, 2013; Maggioni & Santangelo, 2017; Schaltenbrand et al., 2016; Song et al., 2017), the trends and schemes of green investments (Eyraud et al., 2013; Karásek & Pavlica, 2016), and investors’ reaction to green investments(Martin & Moser, 2016). Although there is research evaluating the green investment eﬃciency of the govern- ment (Kim, Lee, Park, Zhang, & Sultanov, 2015), researchers seldom examine ﬁrm-level green investment eﬃciency. Many researchers have primarily concentrated on environ- mental aspects and assess eco-eﬃciency (Burnett & Hansen, 2008; Figge & Hahn, 2013; Hua et al., 2007; Yu et al., 2018) or environmental eﬃciency (Chang et al., 2013; Jiang et al., 2016; Li et al., 2013; Song & Zhou, 2016). Another scream of literature explores the interactions among environmental investments, productivity, energy eﬃciency, and environmental performance (Bostian et al., 2016; Lundgren & Zhou, 2017). There is a paucity of empirical research to directly combine corporate green investments with environmental performance from the perspective of eﬃciency. As for environmental regulation, recent studies have examined the association of environmental regulation with productivity (Wang et al., 2018a) or corporate environmental actions (Wang et al., 2018b). However, there has been little discussion about the correlation between the enforcement of environmental regulation and corporate green investment eﬃciency. We highlight a few related papers and discuss how our study advances the literature. Kim et al. (2015) analyze the eﬃciency of the government’s green investments in three major new and renewable energy (NRE) sources in Korea by applying DEA, and they argue that strategic selection and focused investment help accomplish the policy objectives with fewer resources and budget. Lundgren and Zhou (2017) use DEA to calculate the Malmquist ﬁrm performance indexes that include productivity, energy CHINA JOURNAL OF ACCOUNTING STUDIES 67 eﬃciency, and environmental performance, and then apply a panel vector auto- regression (VAR) method to explore the dynamic and causal relationship between the three dimensions of ﬁrm performance and green investments, ﬁnding that improving energy eﬃciency is able to obtain various advantages. Some research applies DEA to test environmental eﬃciency. Li et al. (2013) use the Super-SBM model with undesirable outputs to calculate regional environmental eﬃ- ciency in China from 1991 to 2001, ﬁnding that the eﬃciency of eastern area is higher than that of central area and western area. Then they utilize the Tobit regression to conclude that ﬁscal decentralization, technology progress, economic scale, and regional diﬀerence can aﬀect environmental eﬃciency. Jiang et al. (2016) apply DEA and a structural equation model to examine the interaction among environmental eﬃciency, output eﬃciency, and proﬁt using the sample of 137 ﬁrms in the textile industry of China’s Jiangsu Province. They ﬁnd a negative eﬀect of environmental eﬃciency on proﬁt and a positive impact of proﬁt on environmental eﬃciency and they demonstrate that output eﬃciency could lower proﬁt while proﬁt is able to enhance output eﬃciency. With regard to environmental regulation, Wang et al. (2018a) investigate how the water quality regulations aﬀect ﬁrms’ emissions of chemical oxygen demand (COD) and productivity in a setting of China. Their result suggests that a 10% decrease in total COD emissions from the industrial sectors just need a 0.1% decrease in output values under the contemporary production technologies. Wang et al. (2018b) disengage the various functions of diﬀerent government levels in China and ﬁnd an inverted U-shaped rela- tionship between administrative hierarchical distance and corporate environmental actions. Despite the aforementioned research, an extremely important aspect of corporate green investment – to what extent is it eﬃcient in terms of reducing pollutant emis- sions? – is considerably unexplored. Our study attempts to answer this question by quantifying ﬁrm-level green investment eﬃciency and further identifying the direction of improvement for low-eﬃciency ﬁrms. In addition, it is not yet known whether environmental enforcement of local governments can inﬂuence the green investment eﬃciency of polluting ﬁrms. Therefore, we evaluate ﬁrm-level green investment eﬃ- ciency and analyze the association between local environmental enforcement and green investment eﬃciency. 3. Research design 3.1. Measuring corporate green investment eﬃciency 3.1.1. The deﬁnition of corporate green investments In order to measure GIE, we should ﬁrst understand what green investments are. However, there seems to be no general deﬁnition of green investments in prior litera- ture. From the perspective of macroeconomics, Eyraud et al. (2013)deﬁne green invest- ments as the inevitable public and private investments to reduce air pollutant emissions and greenhouse gas, without substantially decreasing the production and consumption of non-energy commodities. In the ﬁeld of enterprises’ micro-behaviors, Murillo-Luna, Garcés-Ayerbe, and Rivera-Torres (2008) mention that green investments are always related to morally charged problems linked with green management and corporate 68 Y. CHEN AND J. FENG environmentalism. Doval and Negulescu (2014) state that green investments could be essentially viewed as the expenses which companies made for a friendly impact on the environment. Ateş et al. (2012) consider green investments as the combination of corporate internal investments and external investments involved with the domains of environmental design, production, and logistics. Martin and Moser (2016) regard green investments as a special kind of CSR activity which aims at the reduction of carbon emissions. Voica et al. (2015) argue that green investments are fundamentally consid- ered to be the climate resilient or low-carbon investments made by ﬁrms in the areas of climate change, renewable energy and clean technologies. Since the deﬁnition of green investments varies among researchers, it is important to clarify how the concept is used in this paper. Remarkably, there is a substantial common intersection of the various deﬁnitions and concepts regarding sectors, goods, technologies, services, and processes (Inderst et al., 2012). Thus, we deﬁne corporate green investments as internal invest- ments in equipment, technologies, materials, energy and services that can prevent, control and reduce environmental pollution, produce environmental beneﬁts, and reduce environmental costs, with the goals of improving corporate environmental performance, developing green management and reducing environmental risks. Lundgren and Zhou (2017) propose that it would be helpful to analyze two types of environmental investments – prevention investments and treatment investments sepa- rately, and contend that these two kinds of investments are able to be deemed as proactive and reactive environmental investments. However, they fail to diﬀerentiate these two sorts of investments due to data limitation problems, instead, they mix them into one variable. Undoubtedly, the scope of green investments should be wider than environmental investments, but it is reasonable to classify green investments into prevention investments and treatment investments. Considering ﬁrm-level green invest- ments, we look through all the items of increased construction in progress, increased R&D expenditure, purchased ﬁxed assets and overhead expenses of polluting ﬁrms. If the items relate to investments that impact the production process and act to prevent pollution, we classify them into green prevention investments (GPI). Speciﬁcally, GPI include the use of cleaner energy, new and more eﬃcient advanced materials and less environmentally damaging input factors; the technical transformation of clean produc- tion; the investments in renewable technologies (including large hydroelectric projects); the R&D in energy-eﬃcient and green technologies; and the methods facilitating energy saving and resource recycling. For another, if the items are associated with investments that cope with already emitted pollutant emissions and do not inﬂuence the actual processes of production, such as the environmental treatment to low emissions, desul- phurization and dust removal, waste utilization and regeneration, and the maintenance of environmental protection equipment, we classify them into green treatment invest- ments (GTI). Thus, the total green investments (GI) are the aggregated amounts of GPI and GTI. See Table A1 of Appendix A for variable deﬁnitions. 3.1.2. SBM-DEA approach for measuring ﬁrm-level green investment eﬃciency DEA is a nonparametric method for measuring eﬃciency, productivity or performance of peer decision-making units (DMUs) with multiple inputs and outputs (Emrouznejad & Yang, 2017), which is introduced by Charnes, Cooper, and Rhodes (1978). DEA begins with the establishment of an ‘eﬃcient frontier’ consisting of a group of DMUs that CHINA JOURNAL OF ACCOUNTING STUDIES 69 demonstrate best practices and achieve an eﬃciency score of 1.00, and then assigns the other non-frontier DMUs speciﬁceﬃciency scores based on their distances to the eﬃcient frontier (Liu, Lu, Lu, & Lin, 2013). This method is capable of overcoming the interference of the subjective factors that traditional performance evaluation methods suﬀer when setting weights, and avoiding possible bias or errors in the model speciﬁcation. In this paper, we use pollutant emissions, which are undesirable, as proxies for environmental performance. The fewer pollutants emitted by enterprises, the better. The slack-based measure (SBM) approach in DEA is ﬁrst proposed by Tone (1997) and further improved by Tone (2001) to deal with undesirable factors as outputs. He deﬁnes the SBM model including undesirable outputs as follow: 1 i i¼1 m x ia min ρ ¼ þ b P P q s q s 1 1 2 r t 1 þ þ q þq r¼1 y t¼1 b 1 2 ra ra s.t. Xδ þ s ¼ x Yδ s ¼ y Bδ þ s ¼ b þ b δ; s ; s ; s 0 Among them, ρ measures the eﬃciency score of DMU ; x , y ; and b represent the a a a a actual input, desirable output, and undesirable output of DMU , respectively; X, Y, and B refer to the target values of input, desirable output, and undesirable output based on þ b the best practices, respectively; s , s , s denote slack values of input, desirable output, and undesirable output ofDMU , respectively. To be speciﬁc, s describes the excessive input, which is equal to the diﬀerence between actual input and best target input. s is the insuﬃcient desirable output, which is calculated as the diﬀerence between max- imum expected desirable output and actual desirable output. s is regarded as the undesirable output exorbitance, which captures the diﬀerence between actual undesir- able output and the minimum target of undesired output. Further, we are able to measure the input ineﬃciency and output ineﬃciency of DMU using these slack variables. The input ineﬃciency is the ratio of the absolute value of input’s slack variable to actual input, deﬁned as |s |/x . In the same manner, we measure the output ineﬃ- ciency by computing the ratio of the absolute value of output’s slack variable to actual þ b output, namely |s |/y or |s |/b . A higher value of these ratios could be interpreted as a a a higher degree of ineﬃciency of inputs or outputs. Environmental pollution is undesirable, thus, we apply this SBM-DEA approach of Tone (1997, 2001), which is capable of combining ﬁnancial information with environ- mental information and dealing with undesirable outputs, to measure ﬁrm-level green investment eﬃciency (GIE). While building the SBM-DEA model, we consider each sample ﬁrm to be a DMU and use two types of green investments – GPI and GTI – as actual inputs (x ). We obtain all the items of increased construction in progress (Fieldname: FN_Fn017B05), increased R&D expenditure (Fieldname: Fn02304), purchased 70 Y. CHEN AND J. FENG ﬁxed assets (Fieldname: Fn02006) and overhead expenses (Fieldname: FN05202) of Chinese listed companies in polluting industries from the GTA China Stock Market Financial Database – Statement Notes Database, and further classify them into GPI or GTI according to their deﬁnitions. We then hand-collect a variety of pollutant emission data of each polluting ﬁrm for actual undesirable outputs (b ) from ﬁrm annual reports, corporate social responsibility reports, sustainability reports, and environmental reports. Given that previous research on the relationship between investments in environmental dimensions and ﬁrm performance is inconsistent or inﬂuenced by certain moderators (Antonietti & Marzucchi, 2014; Bostian et al., 2016; Broberg, Marklund, Samakovlis, & Hammar, 2013; Jiang et al., 2016; Sengupta, 2012; Singal et al., 2014), we choose not to include ﬁrm performance as the actual desirable output (y ) in our models. Accordingly, the SBM-DEA model can establish an eﬃcient frontier, and calculate GIE (ρ), best target green investments (X) and target pollutant emissions (B) for each sample ﬁrm. The slack variables for green investments and pollutant emissions measure the excessive corpo- rate green investments (s ) and pollutant emission exorbitance (s ), respectively. In this study, we examine ﬁve output-oriented and variable returns to scale (VRS) SBM-DEA models that diﬀer in the number of undesirable outputs. As the main objective of green investments is to enhance corporate environmental performance, we use output- oriented models to ensure the minimization of pollutant emissions. Besides, the VRS models assume that as inputs change, outputs could change at a diﬀerent rate, which is more in line with the actual situation. A higher value of GIE indicates more eﬃcient green investments of the sample ﬁrm in terms of reducing undesirable outputs – pollutant emissions. See Table A1 of Appendix A for variable deﬁnitions. 3.2. The association between local environmental enforcement and corporate green investment eﬃciency Many researchers have identiﬁed that the government is the key driver for ﬁrms’ environmental practices (Delmas & Toﬀel, 2004; Henriques & Sadorsky, 1996; Porter & van der Linde, 1995; Stoever & Weche, 2018; Wang et al., 2018a; Zhao et al., 2015). In addition, governments operating at diﬀerent levels have alternate and even conﬂicting eﬀects on corporate environmental behaviors, and there is greater involvement of governments in developing countries (Wang et al., 2018b). China has adopted a decentralization regime of environmental regulation. The central government main- tains the authority over the planning, designing, and formulation of environmental policies, while local governments are mainly responsible for the enforcement of speciﬁc environmental policies (Zheng, 2007). Such environmental decentralization gives local governments considerable discretion in implementing the same environmental regula- tions (Zhang et al., 2018). Since the environmental enforcement of local governments is regular and its impact on local ﬁrms is more direct, we expect local environmental enforcement to be the main driver of corporate green investment eﬃciency. Several We hand-collect the emission data of polluting ﬁrms’ seven major pollutants – three for water pollution (COD, NH N, and Wastewater), three for air pollution (SO ,NO , and Soot), and one for soil pollution (Solid Waste). 2 X All output-oriented SBM-DEA models use two inputs – GPI and GTI, but the number of undesirable outputs – emissions of major pollutants – are diﬀerent across these ﬁve models due to the limitation of sample ﬁrms’ emission disclosure. See detail in Table 2. CHINA JOURNAL OF ACCOUNTING STUDIES 71 studies have revealed that the relationship between environmental regulation of gov- ernments and environmental behaviors of ﬁrms is nonlinear (Wang et al., 2018b;Yu et al., 2018), thus, we include the square term of local environmental enforcement in Equation (a). Further, we apply the Tobit regression to estimate Equation (a) because eﬃciency scores calculated by SBM-DEA models are censored for the above the max- imum value of GIE. GIE ¼ αþβ Enforcement þ β Enforcement þ β Size þ β SOE þ β Age þ β GIscale 1 2 3 4 5 6 þ β Leverage þ β OCF þ β Opportunity þ β ROA þ β Growth þ β LSR 7 8 9 10 11 12 þ β Duality þ β DAGE þ β PGDP þ β Structure þ β ENpressure 13 14 15 16 17 þ β ENsupervision þ Area fixed effects þ Industry fixed effects þ Year fixed effects þ ε (a) Enforcement is the proxy for local environmental enforcement, which is based on the Pollution Information Transparency Index (PITI) report issued by the Institute of Public and Environmental Aﬀairs (IPE) and the Natural Resources Defense Council (NRDC) on the IPE website (http://www.ipe.org.cn/reports/Reports_18336_1.html). The assessment standards of PITI cover ﬁve key dimensions: environmental supervision information; pollution source self-disclosure; interactive response; enterprise emission data; and information disclosure of environmental impact assessment. Its evaluation system fully takes into account the current severe environmental situation in China and the trend of reconstruction and improvement of environmental supervision regime. Therefore, PITI is the most compre- hensive and objective index to measure the actual situation of local governments in implementing environmental information disclosure policy (Shen & Feng, 2012). The higher the value of PITI is, the more transparent the pollution information is, suggesting that the environmental enforcement of this city is stricter (Yu et al., 2018). Since our focus is on the enforcement of environmental regulation, it is appropriate and feasible to use PITI as the proxy for local environmental enforcement. We match the city where the polluting ﬁrm is registered with PITI to get the enforcement score of local environmental regulation for each sample ﬁrm-year observation. Compared to other measures of environmental regula- tion, such as the state-level pollution abatement costs (Henderson & Millimet, 2007;Keller& Levinson, 2002) or the pollution discharge fees collected by the local governments (Levinson, 1996), PITI measures the environmental enforcement of local governments from various aspects, which can better reﬂect the comprehensive implementation of multi- dimensional environmental regulation by local governments. Nonetheless, we use alter- native measures of local environmental enforcement to conduct robustness tests, ensuring that our main result is not driven by the measure selection. We use Size, SOE, and Age to control for ﬁrm characteristics. Size is the natural logarithm of the ﬁrm’s total assets at the end of ﬁscal year t. The indicator variable SOE is 1 if the ﬁrm is state-owned during ﬁscal year t and 0 otherwise. Age is the number of years from the establishment of the ﬁrm to the sample year t. We also control for the factors that would aﬀect investment eﬃciency including GIscale, Leverage, OCF, Opportunity, ROA and Growth. Considering the continuous impact of green investment scale, we use lagged value for GIscale, which is equal to the total green investments of the ﬁrm during ﬁscal year t-1 scaled by total assets at the end of ﬁscal year t-1. Leverage is the total debt scaled by total assets, all measured at the end of ﬁscal year t. OCF is the 72 Y. CHEN AND J. FENG net cash ﬂow generated by the ﬁrm’s operating activities during ﬁscal year t scaled by total assets at the end of ﬁscal year t. Opportunity is the ratio of market value of equity to total assets at the end of ﬁscal year t. ROA is the ﬁrm’s return on assets for ﬁscal year t. Growth is the ﬁrm’s growth rate of operating income during ﬁscal year t. Moreover, we use LSR, Duality, and DAGE to control for the eﬀect of corporate governance. LSR is the ﬁrm’s largest shareholder rate for ﬁscal year t. The indicator variable Duality is 1 if the ﬁrm’s chair of the board is also the ﬁrm’s manager and 0 otherwise. DAGE is the average age of the ﬁrm’s directors. Further, we control for the external environment that sample ﬁrms are facing. PGDP is the regional GDP per capita of the province where sample ﬁrm is registered for ﬁscal year t. Structure is the provincial gross industrial output scaled by the regional GDP of the province where sample ﬁrm is registered for ﬁscal year t. ENpressure is the annual average PM2.5 concentration of the province where sample ﬁrm is registered for ﬁscal year t, and a higher value of it indicates ﬁrms that are located in this province are faced with higher pressure of environmental protection. The indi- cator variable ENsupervision is 1 if the ﬁrm is registered in the province where the central government conducted environmental protection supervision during ﬁscal year t and 0 otherwise. Finally, we include area ﬁxed eﬀects, industry ﬁxed eﬀects, and year ﬁxed eﬀects to ensure the generalization of our results across areas, industries and years. We summarize all our variable deﬁnitions in Appendix A. 4. Sample and descriptive statistics 4.1. Sample collection In December 2016, CSRC revised the information disclosure guidelines for publicly issuing securities, which requires Chinese listed ﬁrms that belong to NSMF program (key polluting ﬁrms) to mandatorily disclose detailed pollutants information in their annual reports. This pollutant disclosure regime switch setting provides available ﬁrm- level emission data in 2016 and 2017. According to the Directory of Industrial Classiﬁcations for Listed Firms Subject to Environmental Protection Supervisions issued by the Ministry of Ecology and Environment of the People’s Republic of China (MEEC) in 2008, we select listed ﬁrms in polluting industries from China Security Market and Accounting Research (CSMAR) database. After excluding ﬁrms that are under special treatment or delisted companies, we obtain 2004 sample ﬁrm-year observations. In order to acquire annual total emission data of each ﬁrm, we look through and hand- collect the emission data from ﬁrm annual reports, corporate social responsibility reports, sustainability reports, and environmental reports for the 2004 sample ﬁrm- years in 2016 and 2017. However, the majority of ﬁrms only have narrative or opaque statements about their pollutant emissions or do not disclose quantitative data of total emissions, which leads us to drop 1348 ﬁrm-years. Table 1 provides the descriptive statistics for the pollutant emission disclosure. We report the total number of polluting ﬁrms, the number of polluting ﬁrms with emission disclosure, and the proportion of disclosed ﬁrms in total polluting ﬁrms for key polluting ﬁrms and non-key polluting ﬁrms, respectively. For key polluting ﬁrms, which should mandatorily disclose annual As long as the sample ﬁrm discloses the total annual emissions of one major pollutant, the ﬁrm is deemed to have made the pollutant emission disclosure. CHINA JOURNAL OF ACCOUNTING STUDIES 73 Table 1. Descriptive statistics for pollutant emission disclosure of listed ﬁrms in polluting industries. Polluting ﬁrms with Polluting emission ﬁrms disclosure The proportion of disclosed Year N %Total N %Total ﬁrms in total polluting ﬁrms Key polluting ﬁrms 2016 325 33.71% 178 90.82% 54.77% (mandatory disclosure) 2017 549 52.79% 447 97.17% 81.42% Non-key polluting ﬁrms 2016 639 66.29% 18 9.18% 2.82% (voluntary disclosure) 2017 491 47.21% 13 2.83% 2.65% Total ﬁrms 2016 964 100% 196 100% 20.33% 2017 1040 100% 460 100% 44.23% Firm-year observations 2004 100% 656 100% 32.73% emission data, only 54.77% of them disclose such information in 2016. The situation has been improved in 2017 and 81.42% of key polluting ﬁrms disclose emission data. For non-key polluting ﬁrms, which are voluntary to disclose pollutant emissions, 2.82% and 2.65% of these ﬁrms choose to disclose related data in 2016 and 2017, respectively. The proportion of all disclosed sample ﬁrm-years in total polluting ﬁrm-years is only 32.73%, as a result, our initial test sample during 2016–2017 has 656 ﬁrm-year observations. The numbers suggest that environmental regulation and enforcement for polluting ﬁrms need to be strengthened and the requirements of environmental disclosure ought to be further enhanced. 4.2. Descriptive statistics We present descriptive statistics of variables used in the SBM-DEA models in Table 2.As for input variables, the means of GPI, GTI, and GI are greatly larger than their medians, implying that a part of sample ﬁrms have extremely large amounts of green invest- ments. However, some polluting ﬁrms even have no green investments. We calculate the standard deviation of GPI, GTI, and GI for 656 ﬁrm-year observations and ﬁnd the numbers are very large. The situation shows that managers’ green investment decisions are tremendously diﬀerent across sample ﬁrms. Table 2. Descriptive statistics of variables used in the SBM-DEA models. Variable N Mean Std. Dev. Median Min Max Inputs – Corporate green investments (CNY 10-thousand yuan) GPI 656 11,081.446 53,968.633 0.000 0.000 899,238.400 GTI 656 7952.651 75,115.371 300.964 0.000 1,554,073.000 GI 656 19,034.096 91,942.801 1026.736 0.000 1,554,073.000 Undesirable outputs – Corporate pollutant emissions (t) COD 518 417.636 2088.840 52.010 0.000 28,400.000 NH N 467 80.924 752.662 2.620 0.000 11,400.000 Wastewater 131 3,645,320.000 15,570,000.000 336,500.000 0.000 168,700,000.000 SO 476 233.345 135.056 234.500 1.000 468.000 NO 460 5939.395 54,370.720 224.057 0.016 1,140,781.000 Soot 378 188.212 108.316 188.500 1.000 374.000 Solid Waste 60 1,124,000.000 4,759,506.000 8268.328 1.000 35,100,000.000 74 Y. CHEN AND J. FENG On the other hand, Table 2 reports descriptive statistics of major pollutant emission data for 656 ﬁrm-years that disclose related information. Because CSRC has no speciﬁc requirements, such as which pollutants should be included and how many indicators need to be disclosed, the emission disclosures of each ﬁrm are slightly diﬀerent. We report seven major pollutants – three for water pollution (COD, NH N, and Wastewater), three for air pollution (SO ,NO , and Soot), and one for soil pollution (Solid Waste). 2 X Except for SO and Soot, the means of these pollutant emissions are all greater than their medians, suggesting that some ﬁrms have huge emissions. It is worth noting that only 131 sample ﬁrm-years disclose Wastewater discharges and 60 sample ﬁrm-years disclose Solid Waste emissions, thus, we choose the other ﬁve kinds of pollutant emissions as undesirable outputs in later SBM-DEA empirical examination. 5. Empirical results 5.1. Analyses of SBM-DEA results 5.1.1. Test of corporate green investment eﬃciency To examine the GIE of polluting ﬁrms, which should aim at reducing pollutant emissions as much as possible, we apply the output-oriented SBM-DEA approach to calculate the eﬃciency score. Particularly, we use VRS models, assuming that as inputs increase, undesir- able outputs decrease at a diﬀerent rate, to estimate GIE. Table 3 reports the primary SBM- DEA results. All models use two inputs – GPI and GTI, but the undesirable outputs are diﬀerent across these ﬁve models due to the limitation of sample ﬁrms’ emission disclosure. Model (1) through Model (3) consider both water pollution and air pollution outputs. More speciﬁcally, Model (1) has ﬁve undesirable outputs – COD, NH N, SO ,NO ,Soot,and uses 3 2 X Table 3. Corporate green investment eﬃciency – SBM-DEA results using diﬀerent outputs. Models using both water pollution and air pollution Model using only water pollution Model using only air pollution outputs outputs outputs Model Model Model (1) (2) (3) Model (4) Model (5) Obs 207 294 331 389 416 Mean 0.6090 0.6004 0.5622 0.5359 0.5377 Std. Dev. 0.1936 0.1857 0.1477 0.1176 0.1177 Median 0.5083 0.5067 0.5030 0.5001 0.5004 Min 0.5000 0.5000 0.5000 0.5000 0.5000 Max 1.0000 1.0000 1.0000 1.0000 1.0000 N. eﬃcient 40 50 32 22 23 %eﬃcient 19.32% 17.01% 9.67% 5.66% 5.53% Notes: All models are output-oriented and VRS models based on the SBM approach using two inputs – GPI and GTI. The major diﬀerences among these models are their outputs. Model (1) has ﬁve undesirable outputs – COD, NH N, SO , 3 2 NO , Soot, and uses a sample consisting of ﬁrm-years that disclose these ﬁve pollutant emissions at the same time. Model (2) has four undesirable outputs – COD, NH N, SO ,NO , and its sample is formed from ﬁrm-years that disclose 3 2 X these four pollutant emissions in the meantime. Model (3) uses COD and SO as undesirable outputs, and its sample is constitutive of ﬁrm-years that disclose COD and SO emissions at the same time. Model (4) has only two outputs of water pollution – COD and NH N, and its sample includes ﬁrm-years that disclose these two pollutant emissions in the meantime. Model (5) uses only two outputs of air pollution – SO and NO , and its sample is comprised of ﬁrm-years 2 X that disclose these two pollutant emissions at the same time. We exclude ﬁrm-years with zero total amounts of GPI and GTI for all samples used in diﬀerent SBM-DEA models. CHINA JOURNAL OF ACCOUNTING STUDIES 75 a sample consisting of ﬁrm-years that disclose these ﬁve pollutant emissions at the same time. In Table 3, Model (1) presents that 19.32% of polluting ﬁrms are considered to be eﬃcient with an eﬃciency score equal to 1.00, and the average GIE is 0.6090. For Model (2), we select four outputs – COD, NH N, SO ,NO , and its sample is formed from ﬁrm-years that 3 2 X disclose these four pollutant emissions in the meantime. The average GIE is 0.6004 and 17.01% of polluting ﬁrms are considered to be eﬃcient in Model (2) of Table 3.Wenextuse COD and SO as undesirable outputs in Model (3), and its sample is constitutive of ﬁrm- years that disclose COD and SO emissions at the same time. The result of Model (3) shows that the proportion of eﬃcient polluting ﬁrms is 9.67% and the average GIE is 0.5622. Taken together, these SBM-DEA results demonstrate that the green investment eﬃciency of polluting companies is overall low, and highlight that managers should recognize the importance of identifying and reducing the low-value allocation of limited resource. Next, we attempt to test GIE, which aims at reducing water pollution and air pollution separately, and use Model (4) and Model (5) to conduct the SBM-DEA tests. Speciﬁcally, Model (4) has only two outputs of water pollution – COD and NH N, and its sample includes ﬁrm-years that disclose these two pollutant emissions in the meantime; Model (5) uses only two outputs of air pollution – SO and NO , and its sample is comprised of 2 X ﬁrm-years that disclose these two pollutant emissions at the same time. Model (4) in Table 3 reports an average GIE of 0.5359, with 5.66% of the sample ﬁrms on the eﬃciency frontier. By comparison, Model (5) presents that the average GIE is 0.5377, and only 5.53% of polluting ﬁrms are considered to be eﬃcient, which are not much diﬀerent from the results of Model (4), suggesting the eﬃciency of corporate green investments is still low. However, we caution that these ﬁndings of Model (4) and Model (5) might have some limitations since we cannot distinguish which green investments are used to cope with water pollution and which green investments are used to control air pollution from our data. Based on the above results, we conclude that it is more appropriate to include both water pollution and air pollution outputs in SBM-DEA models. In addition, considering the necessity of suﬃcient observations, we focus our subsequent analyses on Model (3). Polluting ﬁrms are often aﬀected by certain industry policies or regional factors. Hence, GIE could be aggregated into clusters according to those two dimensions. Table 4 summarizes the GIE across industries and areas based on SBM-DEA model (3). We only report results for industries that have at least ten sample ﬁrm-years. The average GIE across industries ranges from a low of 0.5007 (coal mining and washing) to a high of 0.6129 (pharmaceutical manufacturing). Meanwhile, we can observe a greater variance in the number of polluting ﬁrms that are regarded to be eﬃcient, with none of the ﬁrms in coal mining and washing, paper-making and paper products being eﬃcient, compared to 16.95% of ﬁrms in pharmaceutical manufacturing being on the eﬃciency frontier. Furthermore, we provide the GIE classiﬁed by area, from a low of 0.5597 (central region) to a high of 0.5796 (northeastern region). The lowest proportion of eﬃcient ﬁrms is the western region (8.43%) and the highest proportion of eﬃcient ﬁrms is the eastern region (10.12%). These variations in GIE across industries and areas suggest that we should control for industry ﬁxed eﬀects and area ﬁxed eﬀects in second- stage regression analyses. 76 Y. CHEN AND J. FENG Table 4. Corporate green investment eﬃciency classiﬁed by industry and area based on Model (3). Std. % Obs Mean Dev. Median Min Max N. eﬃcient eﬃcient Sample by Industry B06. Coal mining and washing 14 0.5007 0.0007 0.5003 0.5000 0.5023 0 0.00% B09. Non-ferrous metals mining and 10 0.5548 0.1569 0.5013 0.5001 1.0000 1 10.00% dressing C15. Alcohol and wine manufacturing 13 0.5802 0.1351 0.5237 0.5009 1.0000 1 7.69% C22. Paper-making and paper products 19 0.5026 0.0049 0.5004 0.5000 0.5162 0 0.00% C26. Chemical raw materials and chemical 102 0.5585 0.1429 0.5033 0.5001 1.0000 9 8.82% products manufacturing C27. Pharmaceutical manufacturing 59 0.6129 0.1836 0.5179 0.5003 1.0000 10 16.95% C31. Ferrous metal smelting and rolling 31 0.5542 0.1513 0.5002 0.5000 1.0000 3 9.68% processing C32. Non-ferrous metal smelting and 25 0.5670 0.1635 0.5060 0.5000 1.0000 3 12.00% rolling processing Sample by Area Eastern Region 168 0.5621 0.1504 0.5033 0.5000 1.0000 17 10.12% Central Region 60 0.5597 0.1500 0.5013 0.5001 1.0000 6 10.00% Western Region 83 0.5599 0.1413 0.5037 0.5001 1.0000 7 8.43% Northeastern Region 20 0.5796 0.1539 0.5048 0.5001 1.0000 2 10.00% All industries are categorized by the China Securities Regulatory Commission (CSRC) and only eﬃciencies for polluting industries with ten or more ﬁrm-years are reported. 5.1.2. Test of input ineﬃciency and output ineﬃciency The results in Table 3 show that the GIE of polluting ﬁrms is low. To further illustrate the reason for their ineﬃciency, we apply non-oriented and VRS models based on SBM to compare the input ineﬃciency and output ineﬃciency in diﬀerent models. We measure the input/output ineﬃciency by computing the ratio of the absolute value of input/ output’s slack variable to the actual input/output. The slack variable for input is exces- sive green investments, which is equal to the diﬀerence between actual input and target input that captures the best. And the slack variable for undesirable output is emission exorbitance, which is calculated as the diﬀerence between actual output and best target output. In Table 5, Model (6) through Model (8) are all non-oriented and VRS models based on the SBM approach using GI as input, but they diﬀer in the number of undesirable outputs. Model (6) has ﬁve undesirable outputs – COD, NH N, SO ,NO , Soot, and uses 3 2 X a sample comprising of ﬁrm-years that disclose these ﬁve pollutant emissions at the same time. The mean input ineﬃciency score of Model (6) is 0.8003, which is larger than the mean output ineﬃciency score calculated as 0.5946. For Model (7), we use four outputs – COD, NH N, SO ,NO , and its sample is made up of ﬁrm-years that disclose 3 2 X these four pollutant emissions in the meantime. Similarly, Model (7) reports that the mean input ineﬃciency score (0.8368) is larger than the mean output ineﬃciency score (0.6557). Moreover, we select COD and SO as undesirable outputs of Model (8), and its sample is composed of ﬁrm-years that disclose COD and SO emissions at the same time. The result of Model (8) exhibits that mean input ineﬃciency score (0.9002) is also larger than the mean output ineﬃciency score (0.7481). Further, we conduct T-tests to examine the diﬀerence between the mean of input ineﬃciency and output ineﬃciency for these models separately, and the t-statistics are all statistically signiﬁcant at 1% level in all models. Thus, we conclude that a large proportion of the ineﬃciency observed in polluting ﬁrms is attributable to input ineﬃciency, suggesting that the problem of CHINA JOURNAL OF ACCOUNTING STUDIES 77 Table 5. Input ineﬃciency and output ineﬃciency based on non-oriented SBM-DEA models. Obs Mean Std. Dev. Median Min Max Test of diﬀerence Model (6) Input Ineﬃciency 207 0.8003 0.3653 0.9915 0 1 14.84*** Output Ineﬃciency 207 0.5946 0.3198 0.6814 0 0.9965 Model (7) . Input Ineﬃciency 294 0.8368 0.3353 0.9949 0 1 15.64*** Output Ineﬃciency 294 0.6557 0.3158 0.7297 0 0.9999 Model (8) Input Ineﬃciency 331 0.9002 0.2563 0.9963 0 1 12.11*** Output Ineﬃciency 331 0.7481 0.2995 0.9055 0 0.9999 Notes: Model (6) through Model (8) are all non-oriented and VRS models based on the SBM approach using GI as input, but they diﬀer in the number of undesirable outputs. Model (6) has ﬁve undesirable outputs – COD, NH N, SO ,NO , 3 2 X Soot, and uses a sample comprising of ﬁrm-years that disclose these ﬁve pollutant emissions at the same time. Model (7) has four outputs – COD, NH N, SO ,NO , and its sample is made up of ﬁrm-years that disclose these four pollutant 3 2 X emissions in the meantime. Model (8) uses COD and SO as undesirable outputs, and its sample is composed of ﬁrm- years that disclose COD and SO emissions at the same time. We conduct T-tests to examine the diﬀerence between the mean of input ineﬃciency and output ineﬃciency, and *** indicates statistical signiﬁcance at the 1% level. excessive green investments is prominent and managers neglect the eﬃcient allocation and value-creating use of limited resources. 5.1.3. Analyses of excessive corporate green investments By using non-oriented SBM-DEA models in section 5.1.2, we ﬁnd that the dominating reason for polluting ﬁrms’ low GIE is that managers aimlessly invest a large amount of money in environmental dimensions. To further analyze the diﬀerent types of excessive green invest- ments, we use slack variables for GPI and GTI in output-oriented and VRS Model (1) through Model (3) as excessive prevention investments (GPIslack) and excessive treatment investments (GTIslack), respectively. To examine the diﬀerence between the mean of GPIslack and GTIslack, we conduct T-tests in all three models in Table 6. However, the t-statistics for all models are insigniﬁcant, indicating that there is no signiﬁcant diﬀerence between GPIslack and GTIslack. Overall, masses of polluting ﬁrms excessively input both green prevention investments and green treatment investments without considering the input-output eﬃciency. Table 6. Analyses of excessive corporate green investments. Obs Mean Std. Dev. Median Min Max Test of diﬀerence Model (1) GPIslack 207 3782.621 17,590.44 0 0 172,984.5 −1.14 GTIslack 207 10,094.88 77,073.08 139.811 0 1,084,328 Model (2) GPIslack 294 6604.772 37,340.18 0 0 404,856.8 −0.37 GTIslack 294 8222.277 64,890.81 151.231 0 1,084,328 Model (3) GPIslack 331 7008.761 38,205.35 0 0 404,856.8 −0.90 GTIslack 331 12,561.97 104,856.4 330.988 0 1,553,757 Notes: All models are output-oriented and VRS models based on the SBM approach using two inputs – GPI and GTI. The major diﬀerences among these models are their outputs. Model (1) has ﬁve undesirable outputs – COD, NH N, SO , 3 2 NO , Soot, and uses a sample consisting of ﬁrm-years that disclose these ﬁve pollutant emissions at the same time. Model (2) has four undesirable outputs – COD, NH N, SO ,NO , and its sample is formed from ﬁrm-years that disclose 3 2 X these four pollutant emissions in the meantime. Model (3) uses COD and SO as undesirable outputs, and its sample is constitutive of ﬁrm-years that disclose COD and SO emissions at the same time. We conduct T-tests to examine the diﬀerence between the mean of GPIslack and GTIslack. 78 Y. CHEN AND J. FENG 5.2. Tobit regression In the second stage analysis, the Tobit regression is applied to examine the association between environmental enforcement of local governments and corporate green invest- ment eﬃciency. The sample for regression estimation is the same as the sample used in SBM-DEA model (3), which is comprised of ﬁrm-years that disclose COD and SO emissions at the same time. In Table 7, we provide descriptive statistics of the main variables used in the Tobit regression. GIE is calculated by SBM-DEA model (3). The mean of Enforcement is 56.831, which is smaller than the median (59.300), implying that a part of sample ﬁrms are facing weak environmental enforcement of local governments. Besides, the standard deviation of Enforcement is 13.578, demonstrating that local environmental enforcement is quite diﬀerent across sample ﬁrms and local governments have greater discretion in implementing the same environmental regulations. We observe reasonable variation in all control variables. Table 8 reports the pairwise correlations of main variables included in the Tobit regression. We ﬁnd positive Pearson correlation and negative Spearman correlation between Enforcement and GIE, probably because we do not control for other factors and omit the square term of Enforcement. These correlations show that there are no serious multicollinearity issues among variables. 5.2.1. The eﬀect of local environmental enforcement on corporate green investment eﬃciency Table 9 reports the results for testing the eﬀect of local environmental enforcement on corporate green investment eﬃciency. In Column 1, we only control for the internal factors including ﬁrm characteristics, ﬁnancial performance, previous green investment scale, investment opportunity, and corporate governance. The coeﬃcient on Enforcement is signiﬁcantly positive and the coeﬃcient on the square term of Enforcement is signiﬁcantly Table 7. Descriptive statistics of main variables used in Tobit regression. Variable N Mean Std. Dev. Median Min Max GIE 331 0.562 0.148 0.503 0.500 1.000 Enforcement 331 56.831 13.578 59.300 24.600 79.600 Size 331 22.913 1.455 22.818 20.151 28.509 SOE 331 0.492 0.501 0.000 0.000 1.000 Age 331 18.453 4.815 18.000 7.000 34.000 GIscale 331 0.008 0.017 0.002 0.000 0.157 Leverage 331 0.431 0.214 0.426 0.047 1.229 OCF 331 0.069 0.060 0.068 −0.108 0.243 Opportunity 331 1.875 1.559 1.388 0.113 8.607 ROA 331 0.052 0.061 0.045 −0.250 0.340 Growth 331 0.314 1.247 0.183 −0.544 21.886 LSR 331 0.372 0.157 0.359 0.096 0.891 Duality 331 0.254 0.436 0.000 0.000 1.000 DAGE 331 52.116 3.079 52.111 42.385 62.111 PGDP 331 66,937.370 27,566.900 60,199.000 27,643.000 128,994.100 Structure 331 0.349 0.072 0.380 0.078 0.421 ENpressure 331 49.463 13.899 48.200 18.600 81.300 ENsupervision 331 0.502 0.501 1.000 0.000 1.000 We also exclude ﬁrm-years that have no green investments during the sample period. Our ﬁnal sample for Tobit regression has 331 ﬁrm-years, including 96 observations in 2016 and 235 observations in 2017. CHINA JOURNAL OF ACCOUNTING STUDIES 79 Table 8. Pairwise correlations of main variables included in Tobit regression. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 1.GIE −0.01 −0.60* −0.38* −0.12* −0.28* −0.47* −0.10 0.61* 0.21* −0.11* −0.19* 0.17* −0.19* −0.05 0.05 −0.07 −0.01 2.Enforcement 0.05 0.01 −0.08 −0.13* −0.07 −0.13* 0.03 0.07 0.18* 0.04 −0.03 0.02 0.12* 0.67* 0.23* 0.26* 0.09 3.Size −0.26* 0.04 0.42* 0.13* 0.09 0.61* 0.06 −0.74* −0.20* 0.16* 0.33* −0.18* 0.36* 0.04 −0.12* 0.16* 0.02 4.SOE −0.21* −0.07 0.43* 0.22* 0.13* 0.38* −0.01 −0.47* −0.36* −0.11* 0.21* −0.27* 0.23* −0.08 −0.11 0.16* −0.02 5.Age −0.06 −0.13* 0.09 0.20* 0.07 0.21* −0.07 −0.25* −0.20* −0.09 −0.16* −0.02 −0.02 −0.13* 0.00 −0.05 0.03 6.GIscale −0.09 0.05 −0.04 0.05 0.02 0.18* 0.08 −0.20* −0.10 0.13* −0.08 −0.03 −0.14* −0.04 0.07 0.02 0.09 7.Leverage −0.27* −0.15* 0.53* 0.39* 0.21* 0.01 −0.11 −0.73* −0.45* 0.13* 0.10 −0.11* 0.09 −0.10 −0.08 0.14* −0.01 8.OCF −0.03 0.02 0.06 −0.03 −0.04 −0.04 −0.13* 0.13* 0.49* 0.13* 0.14* 0.08 0.02 0.01 0.04 0.00 −0.00 9.Opportunity 0.31* 0.05 −0.59* −0.37* −0.17* −0.06 −0.57* 0.16* 0.50* 0.00 −0.19* 0.18* −0.27* 0.02 0.06 −0.19* −0.00 10.ROA 0.06 0.15* −0.10 −0.26* −0.17* −0.04 −0.42* 0.46* 0.35* 0.30* 0.02 0.15* −0.05 0.19* 0.12* −0.07 0.03 11.Growth −0.01 0.05 0.06 −0.09 −0.03 −0.02 0.08 −0.01 −0.02 0.06 0.07 −0.05 −0.06 −0.01 0.07 −0.02 0.08 12.LSR −0.04 −0.02 0.40* 0.24* −0.18* −0.08 0.09 0.15* −0.11* 0.05 0.07 −0.12* 0.15* 0.04 −0.13* 0.09 −0.06 13.Duality 0.12* 0.02 −0.20* −0.27* −0.02 −0.02 −0.12* 0.08 0.15* 0.10 0.08 −0.14* −0.05 −0.02 −0.04 −0.12* −0.04 14.DAGE −0.04 0.12* 0.39* 0.23* −0.04 −0.11 0.06 0.04 −0.23* −0.00 0.03 0.21* −0.03 0.16* −0.05 0.27* −0.02 15.PGDP −0.02 0.65* 0.12* −0.06 −0.15* −0.00 −0.10 −0.01 −0.05 0.15* −0.05 0.12* −0.01 0.19* 0.21* 0.24* −0.04 16.Structure −0.02 0.09 −0.21* −0.14* 0.09 0.11 −0.07 0.05 0.06 0.06 0.00 −0.24* −0.05 −0.17* −0.03 0.29* −0.13* 17.ENpressure 0.01 0.28* 0.23* 0.17* −0.05 0.01 0.13* −0.03 −0.12* −0.08 0.03 0.13* −0.12* 0.28* 0.23* 0.06 −0.05 18.ENsupervision −0.02 0.07 0.02 −0.02 0.01 0.03 0.00 −0.00 0.00 0.02 −0.05 −0.04 −0.04 0.00 −0.07 −0.02 −0.05 Note: We present the pairwise correlations of main variables included in second-stage Tobit regression, using the sample of SBM-DEA model (3), and report Pearson correlations in the lower triangle and Spearman correlations in the upper triangle. *indicates that the correlation coeﬃcients are statistically signiﬁcant at 5% level. 80 Y. CHEN AND J. FENG Table 9. Tobit regression analysis – the eﬀect of local environmental enforcement on corporate green investment eﬃciency. Dependent Variable = GIE Large- Full Sample SOEs Non-SOEs scale Small-scale (1) (2) (3) (4) (5) (6) (7) Enforcement 0.0089** 0.0139*** 0.0010 0.0038 0.0254*** 0.0065 0.0239*** (2.21) (2.95) (1.22) (0.78) (3.07) (1.39) (2.76) Enforcement2 −0.0001* −0.0001*** −0.0000 −0.0002*** −0.0000 −0.0002** (−1.93) (−2.75) (−0.56) (−2.97) (−1.19) (−2.49) Size −0.0111 −0.0137 −0.0143 −0.0041 −0.0273 (−1.04) (−1.28) (−1.32) (−0.49) (−1.17) SOE −0.0377* −0.0450** −0.0423** 0.0035 −0.0792** (−1.91) (−2.32) (−2.22) (0.18) (−2.38) Age 0.0010 0.0015 0.0015 0.0056** −0.0026 0.0040* −0.0019 (0.47) (0.67) (0.68) (2.09) (−0.86) (1.71) (−0.55) GIscale −0.5435 −0.4450 −0.5317* −0.2926 −1.6747** −0.3272 −0.6312 (−1.63) (−1.34) (−1.70) (−1.01) (−2.45) (−1.32) (−0.94) Leverage −0.0948 −0.1000 −0.0974 −0.1050 −0.0815 −0.1087 −0.1419 (−1.49) (−1.61) (−1.55) (−1.30) (−0.63) (−1.61) (−1.61) OCF −0.2061 −0.1803 −0.1530 −0.2249 −0.2196 −0.0563 −0.3424 (−1.25) (−1.11) (−0.95) (−1.37) (−0.82) (−0.40) (−1.26) Opportunity 0.0176** 0.0158* 0.0170** 0.0152 0.0092 0.0196 0.0104 (2.01) (1.84) (1.97) (1.37) (0.66) (1.14) (0.91) ROA −0.1441 −0.1232 −0.1408 0.0852 −0.2606 −0.0434 −0.1217 (−0.54) (−0.50) (−0.56) (0.23) (−0.70) (−0.21) (−0.32) Growth −0.0030 −0.0023 −0.0023 −0.0181 −0.0047 −0.0009 0.0088 (−0.70) (−0.53) (−0.54) (−0.50) (−0.80) (−0.37) (0.14) LSR 0.0499 0.0330 0.0244 0.0876 −0.0516 −0.0070 0.0743 (0.73) (0.48) (0.35) (1.47) (−0.37) (−0.13) (0.51) Duality 0.0188 0.0144 0.0162 0.0080 0.0062 0.0164 0.0068 (0.80) (0.63) (0.70) (0.36) (0.19) (0.71) (0.17) DAGE 0.0025 0.0014 0.0010 −0.0066* 0.0051 −0.0062 0.0073 (0.65) (0.36) (0.27) (−1.78) (0.86) (−1.34) (1.26) PGDP −0.0000 −0.0000 0.0000 −0.0000 0.0000 −0.0000 (−0.47) (−0.28) (0.90) (−0.96) (0.36) (−0.40) Structure −0.3373* −0.1821 −0.0540 −0.5292 −0.0254 −0.6325 (−1.88) (−1.07) (−0.40) (−1.64) (−0.32) (−1.65) ENpressure 0.0011 0.0007 0.0002 0.0026 0.0012* 0.0020 (1.33) (0.90) (0.39) (1.56) (1.73) (1.13) ENsupervision −0.0031 −0.0092 0.0107 −0.0205 −0.0022 −0.0077 (−0.19) (−0.57) (0.75) (−0.65) (−0.15) (−0.24) Intercept 0.4707* 0.5907** 0.8852*** 0.7319*** 0.6254 0.5453* −0.0681 (1.89) (2.27) (3.38) (2.72) (1.21) (1.96) (−0.17) Area/Industry/Year ﬁxed eﬀects YES YES YES YES YES YES YES F 2.03 1.77 1.72 0.82 1.64 0.56 1.71 Observations 331 331 331 163 168 165 166 Notes: We report the Tobit estimations with robust standard errors clustered at the ﬁrm level and present t-statistics in parentheses. ***, **, and * indicate statistical signiﬁcance at the 1%, 5%, and 10% levels, respectively. negative. In Column 2, we further control for the external environment that sample ﬁrms are faced with. Similarly, we ﬁnd that the coeﬃcient on Enforcement is signiﬁcantly positive at 1% level, and the coeﬃcient on the square term of Enforcement is signiﬁcantly negative at 1% level. Contrastively, we estimate Equation (a) without the square term of Enforcement and Column 3 shows that the coeﬃcient on Enforcement is positive but insigniﬁcant. Taken together, we conclude that the local environmental enforcement has an inverted U-shaped eﬀect on green investment eﬃciency of polluting ﬁrms. Beyond that, we interpret these CHINA JOURNAL OF ACCOUNTING STUDIES 81 signiﬁcantly positive coeﬃcients on Enforcement as indicating that the current intensity of local environmental enforcement in China lies on the left side of this inverted U-shaped curve, which means that the increasingly stringent environmental enforcement of local governments is conducive to improving the GIE of polluting ﬁrms. These ﬁndings provide evidence that Chinese governments should further optimize environmental supervision methods and strengthen environmental enforcement to enhance the eﬃciency of corpo- rate green investments. Next, we divide sample ﬁrms into two groups based on whether they are state-owned and report the results in Column 4 and Column 5 of Table 9. For SOEs, Column 4 shows that the coeﬃcients on Enforcement and the square term of Enforcement are statistically insigniﬁcant. However, for Non-SOEs, we ﬁnd the coeﬃcient on Enforcement is positive and the coeﬃcient on the square term of Enforcement is negative, both are statistically signiﬁcant at 1% level. Considering all of this evidence, it suggests that the inverted U-shaped relationship between local environmental enforcement and the GIE of pollut- ing ﬁrms is only pronounced in the Non-SOE group. These results are reasonable because the corresponding penalties of local environmental enforcement and environ- mental reputation have a greater impact on non-state-owned enterprises. At the same time, Non-SOEs have more limited resources to invest in environmental dimensions, thus, they would pay more attention to improving the eﬃciency of green investments. Furthermore, we group our sample ﬁrms into two types based upon their size. To be speciﬁc, we classify the ﬁrms with above sample median value of Size as ‘Large-scale’ ﬁrms and ‘Small-scale’ ﬁrms otherwise. The results in Column 6 and Column 7 of Table 9 suggest that the local environmental enforcement has an inverted U-shaped eﬀect on GIE only when the ﬁrms belong to the Small-scale group. The main reason for these results is that the coordination costs of small-scale enterprises are relatively low, which is more favorable to green management and eﬃcient allocation of resources. Consequently, the promotion of local environmental enforcement on green investment eﬃciency is more obvious. 5.2.2. Supplementary analyses: local environmental enforcement and excessive corporate green investments From the ﬁrst stage analyses of SBM-DEA results, we ﬁnd that the green investment eﬃciency of polluting companies is overall low primarily due to excessive green invest- ments. This essentially means that polluting ﬁrms have to focus on solving the problem of redundant green investments if they intend to enhance their green investment eﬃciency. Since we have documented that local environmental enforcement exerts a signiﬁcant inverted U-shaped impact on the eﬃciency of corporate green investments in Section 5.2.1, it is necessary for local governments to further consider promoting polluting ﬁrms’ green investment eﬃciency by restraining their excessive green invest- ments. Therefore, we conduct supplementary analyses to examine whether increasingly strict local environmental enforcement is accompanied with a decrease in redundancy of corporate green investments, which would oﬀer a useful reference for the Chinese government’s current environmental regulatory reform. The slack variables for GPI and GTI calculated by SBM-DEA model (3) measure the excessive prevention investments (GPIslack) and excessive treatment investments (GTIslack), respectively. As we ﬁnd that there is no signiﬁcant diﬀerence between 82 Y. CHEN AND J. FENG GPIslack and GTIslack in Section 5.1.3, we use the total amounts of GPIslack and GTIslack, namely GIslack, as the dependent variable. Because the values for GIslack of sample ﬁrms that have no excessive green investments are all zero, which are censored data, we also apply the Tobit regression to estimate Equation (b) that is speciﬁed as follows: LogðÞ 1 þ GIslack¼ αþβ Enforcement þ β Enforcement þ β Size þ β SOE þ β Age 1 2 3 4 5 þβ GIscale þ β Leverage þ β OCF þ β Opportunity þ β ROA 6 7 8 9 10 þ β Growth þ β LSR þ β Duality þ β DAGE þ β PGDP 11 12 13 14 15 þ β Structure þ β ENpressure þ β ENsupervision þ Area fixed effects 16 17 18 þ Industry fixed effects þ Year fixed effects þ ε (b) where GIslack represents the total excessive green investments. All control variables are similar to the variables used in Equation (a). We estimate Equation (b) separately for the full sample and subsamples. For the full sample, Column 1 and Column 2 of Table 10 report that the coeﬃcients on Enforcement are signiﬁcantly negative and the coeﬃcients on the square term of Enforcement are signiﬁcantly positive. By comparison, we estimate Equation (b) without the square term of Enforcement and Column 3 demonstrates that the coeﬃcient on Enforcement is negative but insigniﬁcant. These results suggest that there is a U-shaped eﬀect of local environmental enforcement on excessive green investments. In terms of the signiﬁcantly negative coeﬃcients on Enforcement in Column 1 and Column 2, we conclude that the current intensity of local environmental enforcement in China lies on the left side of this U-shaped curve, which means that the stricter environmental enforcement of local governments can signiﬁcantly restrain excessive green investments of polluting ﬁrms. Therefore, we believe that curbing corporate green investment redundancy is an eﬀective regulatory approach for local governments to enhance the green investment eﬃciency of polluting enterprises. For our subsamples, in Column 4 and Column 5 of Table 10,we ﬁnd the U-shaped relationship between local environmental enforcement and excessive green investments is only pronounced in Non-SOEs. Besides, Column 6 and Column 7 exhibit that local environmental enforcement has a statistically signiﬁcant U-shaped eﬀect only on the small-scale ﬁrms’ excessive green investments. These ﬁndings are consistent with our preceding results that the inverted U-shaped relationship between local environmental enforcement and GIE is only statistically signiﬁcant in non-state-owned enterprises and small-scale enterprises, indicating that local governments could improve GIE of these ﬁrms by restraining their excessive green investments. As for SOEs and large-scale ﬁrms, local governments should make more use of market-based means, such as environ- mental taxes and tradable emission permit, encourage managers to actively take corpo- rate social responsibilities instead of passively catering to government regulation, and guide the majority of stakeholders to play a supervisory role. 5.2.3. Robustness tests In this section, we conduct a series of robustness tests to assure that our primary ﬁndings on the inverted U-shaped relationship between local environmental enforce- ment and corporate green investment eﬃciency are robust to the alternative sample, CHINA JOURNAL OF ACCOUNTING STUDIES 83 Table 10. Local environmental enforcement and excess corporate green investments. Dependent Variable = Log(1+ GIslack) Large- Full Sample SOEs Non-SOEs scale Small-scale (1) (2) (3) (4) (5) (6) (7) Enforcement −0.4279*** −0.5778*** −0.0468 −0.3507 −1.0122*** −0.2439 −1.1292*** (−2.74) (−3.37) (−1.46) (−1.31) (−3.51) (−0.98) (−3.65) Enforcement2 0.0037** 0.0050*** 0.0028 0.0088*** 0.0021 0.0095*** (2.51) (3.13) (1.11) (3.30) (0.90) (3.34) Size 0.9718*** 0.9734*** 0.9823*** 1.0770*** 1.2102* (3.02) (3.02) (3.02) (2.96) (1.79) SOE 0.4591 0.6890 0.5950 −0.6765 1.7216* (0.64) (0.96) (0.82) (−0.66) (1.67) Age −0.0825 −0.0880 −0.0930 −0.0446 −0.0677 −0.1058 −0.0693 (−1.23) (−1.31) (−1.37) (−0.43) (−0.74) (−1.11) (−0.72) GIscale 24.0560 24.4617 27.5132 8.5194 60.4735** 21.1233 28.7209 (1.10) (1.10) (1.29) (0.34) (2.01) (0.66) (0.87) Leverage 2.9308 2.9096 2.8723 2.3695 2.2691 3.8302 1.6468 (1.37) (1.39) (1.35) (0.92) (0.56) (1.18) (0.60) OCF 15.1716*** 15.7160*** 14.4933** 8.7828 20.2882* 14.1440* 20.6909** (2.59) (2.68) (2.46) (1.32) (1.92) (1.97) (2.04) Opportunity −1.0431*** −1.0350*** −1.0979*** −0.8152* −0.9577** −0.7394 −1.4690*** (−3.22) (−3.37) (−3.55) (−1.80) (−2.11) (−1.22) (−4.12) ROA 14.1512* 13.0407* 14.4387* 14.5701* 7.1876 0.8787 14.5125 (1.87) (1.75) (1.88) (1.70) (0.52) (0.07) (1.28) Growth 0.1226 0.1491 0.1471 0.6663 0.2256 0.0327 0.9519 (1.11) (1.35) (1.29) (0.58) (1.36) (0.26) (0.67) LSR −2.6394 −2.6600 −2.4163 −3.3735 −2.2050 −0.3148 −2.4276 (−1.26) (−1.26) (−1.13) (−1.41) (−0.56) (−0.13) (−0.60) Duality 0.6198 0.7697 0.7585 −0.3403 1.8099* 0.0849 2.3558** (0.85) (1.06) (1.04) (−0.28) (1.97) (0.09) (2.07) DAGE −0.0731 −0.0798 −0.0660 −0.0013 −0.0781 0.1458 −0.2050 (−0.63) (−0.68) (−0.56) (−0.01) (−0.46) (0.90) (−1.27) PGDP 0.0001** 0.0000** 0.0001* 0.0001* 0.0000 0.0001*** (2.58) (2.21) (1.87) (1.95) (0.60) (2.92) Structure 13.1153** 5.6017 8.3163 18.5583* 0.9766 23.9084** (2.49) (1.16) (1.20) (1.82) (0.13) (1.98) ENpressure −0.0042 0.0094 −0.0139 0.0108 −0.0168 0.0354 (−0.17) (0.39) (−0.44) (0.24) (−0.58) (0.76) ENsupervision 0.5080 0.7660 0.2845 0.8671 0.5543 0.6204 (0.93) (1.37) (0.42) (0.97) (0.86) (0.63) Intercept 1.5907 −3.6222 −14.7038 −11.8404 −3.8917 7.9823 27.0331** (0.17) (−0.37) (−1.60) (−0.97) (−0.22) (0.64) (2.23) Area/Industry/Year YES YES YES YES YES YES YES ﬁxed eﬀects F 10.49 9.79 9.80 6.48 5.46 2.50 5.37 Observations 331 331 331 163 168 165 166 Notes: We report the Tobit estimations with robust standard errors clustered at the ﬁrm level and present t-statistics in parentheses. ***, **, and * indicate statistical signiﬁcance at the 1%, 5%, and 10% levels, respectively. additional control variables, and alternative measures of local environmental enforcement. In Column 1 of Table 11, we select an alternative sample that is used for the preceding SBM-DEA model (2) and rerun the Tobit regression to estimate Equation (a). The result illustrates that local environmental enforcement has a statistically signiﬁcant inverted U-shaped impact on corporate green investment eﬃciency at 1% level, sug- gesting that our ﬁndings are robust to this alternative sample. 84 Y. CHEN AND J. FENG Table 11. Robustness tests. Dependent Variable = GIE (1) (2) (3) (4) Enforcement 0.0152*** 0.0091** (2.92) (2.15) Enforcement2 −0.0001*** −0.0001** (−2.76) (−2.00) LPAI 0.1478** (1.99) LPAI2 −0.0911*** (−2.64) NPLES 0.0311*** (2.71) NPLES2 −0.0021*** (−3.56) Size −0.0057 −0.0125 −0.0119 −0.0113 (−0.55) (−1.21) (−1.11) (−1.08) SOE −0.0355* −0.0256 −0.0184 −0.0228 (−1.82) (−1.39) (−1.00) (−1.30) Age 0.0010 0.0019 0.0017 0.0022 (0.44) (0.91) (0.87) (1.07) GIscale −0.3164 −0.5223* −0.7090** −0.6188** (−1.01) (−1.69) (−2.11) (−1.99) Leverage −0.0748 −0.1060* −0.1024* −0.1045 (−1.20) (−1.73) (−1.67) (−1.62) OCF −0.1566 −0.1938 −0.1887 −0.1416 (−0.91) (−1.25) (−1.21) (−0.94) Opportunity 0.0129 0.0138 0.0170** 0.0143* (1.41) (1.63) (2.00) (1.70) ROA 0.1212 −0.1670 −0.1708 −0.1421 (0.59) (−0.71) (−0.73) (−0.60) Growth −0.0014 0.0003 0.0010 0.0009 (−0.31) (0.08) (0.26) (0.22) LSR −0.0224 0.0404 0.0212 0.0320 (−0.32) (0.60) (0.32) (0.47) Duality 0.0055 −0.0066 −0.0113 −0.0094 (0.26) (−0.31) (−0.54) (−0.44) DAGE −0.0016 0.0029 0.0027 0.0034 (−0.40) (0.76) (0.70) (0.85) PGDP −0.0000 −0.0000 0.0000 0.0000 (−0.78) (−0.23) (0.36) (0.95) Structure −0.4974** −0.1596 0.0404 0.0244 (−2.51) (−0.89) (0.26) (0.16) ENpressure 0.0013* 0.0013* 0.0009 0.0019** (1.73) (1.67) (1.07) (2.27) ENsupervision 0.0074 −0.0037 −0.0046 0.0013 (0.41) (−0.23) (−0.29) (0.08) BDsize 0.0024 0.0013 0.0009 (0.58) (0.32) (0.22) Female 0.2165*** 0.2709** 0.2890** (2.86) (2.15) (2.28) EXSR 0.2526** 0.2304*** 0.2201*** (2.00) (3.02) (2.90) ENLAW 0.0104 0.0160 0.0794*** (0.94) (1.46) (3.55) ENpenalty 0.0479 0.0575 0.0486 (1.33) (1.61) (1.33) Intercept 0.5684** 0.4388* 0.5670** 0.3520 (2.10) (1.68) (2.32) (1.35) Area/Industry/Year ﬁxed eﬀects YES YES YES YES F 1.64 2.27 2.24 2.78 Observations 294 331 331 331 Notes: We report the Tobit estimations with robust standard errors clustered at the ﬁrm level and present t-statistics in parentheses. ***, **, and * indicate statistical signiﬁcance at the 1%, 5%, and 10% levels, respectively. CHINA JOURNAL OF ACCOUNTING STUDIES 85 Next, we include extra internal and external control variables in the Tobit regression and report the result in Column 2 of Table 11. Among these additional control variables, we use BDsize, Female, and EXSR to control for internal ﬁrm-level factors. BDsize is the number of directors on the board of the ﬁrm for ﬁscal year t. Female is the proportion of the ﬁrm’s female directors to the total number of directors for ﬁscal year t. EXSR is the ratio of equity shares held by executives to total equity shares of the ﬁrm at the end of ﬁscal year t. In addition, we use ENLAW and ENpenalty to further control for the external environment of sample ﬁrms. ENLAW is the sum of local environmental laws and regulations scaled by local industrial output for ﬁscal year t-1. The indicator variable ENpenalty is 1 if the ﬁrm has been punished for environmental violations, etc. in previous years and 0 otherwise. We ﬁnd that the coeﬃcient on Enforcement is signiﬁcantly positive and the coeﬃcient on the square term of Enforcement is signiﬁcantly negative, which is consistent with our preceding result and further conﬁrm the inverted U-shaped relationship between local environmental enforcement and green investment eﬃciency of polluting ﬁrms. Moreover, to address the concern that our primary ﬁndings might result from the measure of local environmental enforcement and omitted correlated variables, we use two alternative measures – LPAI and NPLES – to capture the stringency of local environ- mental regulation, and include the above-mentioned extra control variables to rerun the Tobit regression. Following Keller and Levinson (2002) and Henderson and Millimet (2007), we use LPAI, which is local industrial pollution abatement investments scaled by the industrial sector’s contribution to the regional GDP of the province where sample ﬁrm is registered for ﬁscal year t, to measure environmental regulation of local govern- ments. Provinces with larger values of LPAI have stricter environmental regulations. We rerun the Tobit regression to estimate Equation (a) with LPAI and the square term of LPAI as the primary independent variables of interest and report the result in Column 3 of Table 11. Similarly, we ﬁnd the coeﬃcient on LPAI is signiﬁcantly positive and the coeﬃcient on the square term of LPAI is signiﬁcantly negative, demonstrating a signiﬁcant inverted U-shaped relationship between the stringency of local environ- mental regulation and corporate green investment eﬃciency. In addition, following Bu, Liu, Wagner, and Yu (2013), we measure the environmental regulation of local govern- ments with NPLES, the number of enforcement personnel in local environmental super- vision scaled by the environmental investments of local governments. A higher value of NPLES indicates ﬁrms that are located in this province are facing stronger environmental regulation of local governments. NPLES is lagged by one year when we conduct the Tobit regression. Column 4 of Table 11 reports that the coeﬃcient on NPLES is positive and the coeﬃcient on the square term of NPLES is negative, both are statistically signiﬁcant at 1% level, which still indicates that local environmental regulation has an inverted U-shaped eﬀect on GIE of polluting ﬁrms. Thus, our primary results are robust to these alternative measures of local environmental enforcement and extra control variables. 6. Conclusion Corporate green investment eﬃciency plays a leading role in achieving sustainable development and maximizing social value for polluting ﬁrms. We assess ﬁrm-level 86 Y. CHEN AND J. FENG green investment eﬃciency and explore the association between local environmental enforcement and green investment eﬃciency. Applying SBM-DEA approach with green investments as inputs and pollutant emissions as undesirable outputs in the ﬁrst stage, we quantify and evaluate green investment eﬃciency of Chinese listed companies in polluting industries in 2016 and 2017, ﬁnding that the corporate green investment eﬃciency is overall low, primarily due to excessive green investments. The results indicate that managers may only invest extensively in environmental dimensions for compliance reasons and neglect the eﬃcient allocation of resources to reduce pollutant emissions, which is a waste of resources to some extent. In the second stage, we conduct the Tobit regression to examine the relationship between local environmental enforcement and green investment eﬃciency of polluting ﬁrms. We ﬁrst ﬁnd that local environmental enforcement has an inverted U-shaped eﬀect on corporate green investment eﬃciency. Notably, this eﬀect is only statistically signiﬁcant in non-state-owned enterprises and small-scale enterprises. Furthermore, we document that there exists a U-shaped eﬀect of local environmental enforcement on excessive green investments, and this eﬀect is only pronounced in Non-SOEs and small- scale ﬁrms as well. Our results highlight that managers should recognize the importance of identifying and reducing the low-value allocation of limited resources, and indicate that local governments should optimize corporate green investment eﬃciency through diﬀeren- tiated environmental regulation. Regarding Non-SOEs and small-scale ﬁrms, local gov- ernments could adopt prudent environmental regulation to improve their green investment eﬃciency by restraining excessive green investments. As for SOEs and large- scale ﬁrms, local governments should make more use of market-based means, such as environmental taxes and tradable emission permit, encourage managers to actively take corporate social responsibilities instead of passively catering to government regulation, and guide the majority of stakeholders to play a supervisory role. Our ﬁndings should be of interest to policymakers, managers, and stakeholders involved in the regulation of environmental pollution. Acknowledgments We appreciate the helpful comments and suggestions of the anonymous referee and conference participants at the 23rd Annual Conference of Chinese Finance. We acknowledge ﬁnancial support from the Fundamental Research Funds for the Central Universities (No. JBK1807077). Disclosure statement No potential conﬂict of interest was reported by the authors. Funding This work was supported by the Fundamental Research Funds for the Central Universities [JBK1807077]. CHINA JOURNAL OF ACCOUNTING STUDIES 87 Data availability All data are available from public sources. References Antonietti, R., & Marzucchi, A. (2014). Green tangible investment strategies and export perfor- mance: A ﬁrm-level investigation. Ecological Economics, 108, 150–161. Ateş, M.A., Bloemhof, J., van Raaij, E.M., & Wynstra, F. (2012). Proactive environmental strategy in a supply chain context: The mediating role of investments. International Journal of Production Research, 50(4), 1079–1095. Bahn, O., Chesney, M., & Gheyssens, J. (2012). The eﬀect of proactive adaptation on green investment. Environmental Science & Policy, 18(4), 9–24. Bostian, M., Färe, R., Grosskopf, S., & Lundgren, T. (2016). Environmental investment and ﬁrm performance: A network approach. Energy Economics, 57, 243–255. Broberg, T., Marklund, P.-O., Samakovlis, E., & Hammar, H. (2013). Testing the Porter hypothesis: The eﬀects of environmental investments on eﬃciency in Swedish industry. Journal of Productivity Analysis, 40(1), 43–56. Bu, M., Liu, Z., Wagner, M., & Yu, X. (2013). Corporate social responsibility and the pollution haven hypothesis: Evidence from multinationals’ investment decision in China. Asia-Paciﬁc Journal of Accounting & Economics, 20(1), 85–99. Burnett, R.D., & Hansen, D.R. (2008). Ecoeﬃciency: Deﬁning a role for environmental cost management. Accounting Organizations & Society, 33(6), 551–581. Carroll, A.B. (1991). The pyramid of corporate social responsibility: Toward the moral management of organizational stakeholders. Business Horizons, 34(4), 39–48. Chang, Y.T., Zhang, N., Danao, D., & Zhang, N. (2013). Environmental eﬃciency analysis of transportation system in China: A non-radial DEA approach. Energy Policy, 58(9), 277–283. Charnes, A., Cooper, W.W., & Rhodes, E. (1978). Measuring the eﬃciency of decision making units. European Journal of Operational Research, 2(6), 429–444. Cho, C.H., Guidry, R.P., Hageman, A.M., & Patten, D.M. (2012). Do actions speak louder than words? An empirical investigation of corporate environmental reputation. Accounting, Organizations and Society, 37(1), 14–25. Costa-Campi, M.T., García-Quevedo, J., & Martínez-Ros, E. (2017). What are the determinants of investment in environmental R&D? Energy Policy, 104, 455–465. Delmas, M., & Toﬀel, M.W. (2004). Stakeholders and environmental management practices: An institutional framework. Business Strategy and the Environment, 13(4), 209–222. Doval, E., & Negulescu, O. (2014). A model of green investments approach. Procedia Economics and Finance, 15, 847–852. Emrouznejad,A.,&Yang,G.L.(2017). A survey and analysis of the ﬁrst 40 years of scholarly literature in DEA: 1978–2016.Socio-Economic Planning Sciences.doi:10.1016/j. seps.2017.01.008 Eyraud, L., Clements, B., & Wane, A. (2013). Green investment: Trends and determinants. Energy Policy, 60, 852–865. Figge, F., & Hahn, T. (2013). Value drivers of corporate eco-eﬃciency: Management accounting information for the eﬃcient use of environmental resources. Management Accounting Research, 24(4), 387–400. Henderson, D.J., & Millimet, D.L. (2007). Pollution abatement costs and foreign direct investment inﬂows to u.s. states: a nonparametric reassessment. Review of Economics & Statistics, 89(1), 178–183. Henriques, I., & Sadorsky, P. (1996). The determinants of an environmentally responsive ﬁrm: an empirical approach. Journal of Environmental Economics and Management, 30(3), 381–395. Hua, Z., Bian, Y., & Liang, L. (2007). Eco-eﬃciency analysis of paper mills along the Huai River: An extended DEA approach. Omega, 35(5), 578–587. 88 Y. CHEN AND J. FENG Huang, X.B., & Watson, L. (2015). Corporate social responsibility research in accounting. Journal of Accounting Literature, 34,1–16. Inderst, G., Kaminker, C., & Stewart, F. 2012.Deﬁning and measuring green investments: implica- tions for institutional investors’ asset allocations. OECD Working Papers on Finance, Insurance and Private Pensions No. 24, OECD Publishing. doi:10.1787/5k9312twnn44-en Jiang, L., Folmer, H., & Bu, M. (2016). Interaction between output eﬃciency and environmental eﬃciency: Evidence from the textile industry in Jiangsu Province, China. Journal of Cleaner Production, 113, 123–132. Karásek, J., & Pavlica, J. (2016). Green investment scheme: experience and results in the czech republic. Energy Policy, 90, 121–130. Keller, W., & Levinson, A. (2002). Pollution abatement costs and foreign direct investment inﬂows to U.S. states. Review of Economics & Statistics, 84(4), 691–703. Kim, E.H. (2013). Deregulation and diﬀerentiation: Incumbent investment in green technologies. Strategic Management Journal, 34(10), 1162–1185. Kim, K.T., Lee, D.J., Park, S.J., Zhang, Y., & Sultanov, A. (2015). Measuring the eﬃciency of the investment for renewable energy in Korea using data envelopment analysis. Renewable & Sustainable Energy Reviews, 47, 694–702. Levinson, A. (1996). Environmental regulations and manufacturers’ location choices: Evidence from the Census of Manufactures. Journal of Public Economics, 62(1), 5–29. Li, H., Fang, K., Yang, W., Wang, D., & Hong, X. (2013). Regional environmental eﬃciency evaluation in China: Analysis based on the Super-SBM model with undesirable outputs. Mathematical and Computer Modelling, 58(5–6), 1018–1031. Liu, J.S., Lu, L.Y.Y., Lu, W.-M., & Lin, B.J.Y. (2013). Data envelopment analysis 1978–2010: A citation- based literature survey. Omega, 41(1), 3–15. Lundgren, T., & Zhou, W. (2017). Firm performance and the role of environmental management. Journal of Environmental Management, 203(Pt 1), 330–341. Maggioni, D., & Santangelo, G.D. (2017). Local environmental non-proﬁt organizations and the green investment strategies of family ﬁrms. Ecological Economics, 138, 126–138. Martin, P.R., & Moser, D.V. (2016). Managers’ green investment disclosures and investors’ reaction. Journal of Accounting and Economics, 61(1), 239–254. Maxwell, J.W., & Decker, C.S. (2006). Voluntary environmental investment and responsive regulation. Environmental and Resource Economics, 33(4), 425–439. Murillo-Luna, J.L., Garcés-Ayerbe, C., & Rivera-Torres, P. (2008). Why do patterns of environmental response diﬀer? A stakeholders’ pressure approach. Strategic Management Journal, 29(11), 1225–1240. Pearce, D.W., Markandya, A., & Barbier, E. (1989). Blueprint for a green economy. London, United Kingdom: Routledge. Porter, M.E., & van der Linde, C. (1995). Toward a new conception of the environment-competitiveness relationship. Journal of Economic Perspectives, 9(4), 97–118. Schaltenbrand, B., Foerstl, K., Azadegan, A., & Lindeman, K. (2016). See what we want to see? The eﬀects of managerial experience on corporate green investments. Journal of Business Ethics. doi:10.1007/s10551-016-3191-x Sengupta, A. (2012). Investment in cleaner technology and signaling distortions in a market with green consumers. Journal of Environmental Economics and Management, 64(3), 468–480. Shen, H., & Feng, J. (2012). Media monitoring, government supervision, and corporate environ- mental disclosure. Accounting Research, 20(02), 72–78+97. in Chinese. Singal, M., Parsa, H.G., Segarraoña, M., Jang, S., Chen, R., & Singh, A.J. (2014). The link between ﬁrm ﬁnancial performance and investment in sustainability initiatives. Cornell Hospitality Quarterly, 55(1), 19–30. Song, M., & Zhou, Y. (2016). Quantitative analysis of foreign trade and environmental eﬃciency in China. Emerging Markets Finance and Trade, 52(7), 1647–1660. Song, Y., Yao, S., Yu, D., & Shen, Y. (2017). Risky multi-criteria group decision making on green capacity investment projects based on supply chain. Journal of Business Economics & Management, 18(3), 355–372. CHINA JOURNAL OF ACCOUNTING STUDIES 89 Stoever, J., & Weche, J.P. (2018). Environmental regulation and sustainable competitiveness: evaluating the role of ﬁrm-level green investments in the context of the porter hypothesis. Environmental and Resource Economics, 70(2), 429–455. Tone, K. (1997). Several algorithms to determine multipliers for use in cone-ratio envelopment approaches to eﬃciency evaluations in DEA. In H. Amman, B. Rustem & A. Whinston Computational Approaches to Economic Problems(pp. 91–109). Boston, MA: Springer. Tone, K. (2001). A slacks-based measure of eﬃciency in data envelopment analysis. European Journal of Operational Research, 130(3), 498–509. Voica, M.C., Panait, M., & Radulescu, I. (2015). Green investments – between necessity, ﬁscal constraints and proﬁt. Procedia Economics and Finance, 22,72–79. Wang, C., Wu, J., & Zhang, B. (2018a). Environmental regulation, emissions and productivity: Evidence from Chinese COD-emitting manufacturers. Journal of Environmental Economics and Management, 92,54–73. Wang, R., Wijen, F., & Heugens, P.P.M.A.R. (2018b). Government’s green grip: Multifaceted state inﬂuence on corporate environmental actions in China. Strategic Management Journal, 39(2), 403–428. Yu, Y., Huang, J., & Luo, N. (2018). Can more environmental information disclosure lead to higher eco-eﬃciency? Evidence from China. Sustainability, 10(2), 528. Zhang, B., Chen, X., & Guo, H. (2018). Does central supervision enhance local environmental enforcement? Quasi-experimental evidence from China. Journal of Public Economics, 164, 70–90. Zhao, X., Zhao, Y., Zeng, S., & Zhang, S. (2015). Corporate behavior and competitiveness: Impact of environmental regulation on Chinese ﬁrms. Journal of Cleaner Production, 86, 311–322. Zheng, Y. (2007). De Facto Federalism in China: Reforms and dynamics of central-local relations. Singapore: World Scientiﬁc Publishing. 90 Y. CHEN AND J. FENG Appendices Appendix A Table A1. Variable deﬁnitions. Variable Description Data Source Inputs – Corporate green investments by ﬁrm i during ﬁscal year t: GI = the total amounts of corporate green China Security Market and Accounting Research prevention investments and green treatment (CSMAR) database investments. (We deﬁne corporate green investments as internal investments in equipment, technologies, materials, energy and services that can prevent, control and reduce environmental pollution, produce environmental beneﬁts, and reduce environmental costs, with the goals of improving corporate environmental performance, developing green management and reducing environmental risks.) GPI = the amounts of items that are categorized as China Security Market and Accounting Research green prevention investments from the (CSMAR) database increased construction in progress, the increase in R&D expenditure and purchased ﬁxed assets. (Items are related to investments that aﬀect the production process and act to prevent pollution.) GTI = the amounts of items that are categorized as China Security Market and Accounting Research green treatment investments from the (CSMAR) database increased construction in progress, the increase in R&D expenditure, purchased ﬁxed assets and overhead expenses. (Items are related to investments that deal with already emitted pollution and do not aﬀect the actual production processes.) Undesirable outputs – Corporate pollutant emissions during ﬁscal year t of ﬁrm i: COD = annual COD emissions. The emission data is hand collected form ﬁrm N = annual NH N emissions. NH annual reports, corporate social responsibility 3 3 Wastewater = annual Wastewater discharges. reports, sustainability reports, and SO = annual SO emissions. environmental reports 2 2 NO = annual NO emissions. X X Soot = annual Soot emissions. Solid Waste = annual Solid Waste emissions. Corporate green investment eﬃciency for ﬁscal year t of ﬁrm i: GIE = the measure of ﬁrm-level green investment Results from SBM-DEA models eﬃciency calculated by SBM-DEA models. A higher value of GIE indicates more eﬃcient green investments of the sample ﬁrm in terms of reducing undesirable outputs (pollutant emissions). Input = the ratio of the absolute value of input’s slack Ineﬃciency variable to actual input. Output = the ratio of the absolute value of output’s slack Ineﬃciency variable to actual output. GPIslack = the slack variable of green prevention investments calculated by SBM-DEA models, which is equal to the diﬀerence between actual GPI and best target GPI. GTIslack = the slack variable of green treatment investments calculated by SBM-DEA models, which is equal to the diﬀerence between actual GTI and best target GTI. GIslack = the total amounts of GPIslack and GTIslack. (Continued) CHINA JOURNAL OF ACCOUNTING STUDIES 91 Table A1. (Continued). Variable Description Data Source Environmental enforcement of local governments for ﬁscal year t: Enforcement = the measure of local environmental We hand-collect the data from PITI reports issued enforcement which is based on the 2016–2017 by Institute of Public and Environmental and 2017–2018 Pollution Information Aﬀairs (IPE) and Natural Resources Defense Transparency Index (PITI). We match the city Council (NRDC). where the polluting ﬁrm is registered with PITI to get the enforcement score of local environmental regulation for each sample ﬁrm- year observation. A higher value of Enforcement indicates ﬁrms that are located in this city are facing stricter environmental enforcement of local governments. Control variables: Size = the natural logarithm of the ﬁrm’s total assets at the end of ﬁscal year t. SOE = 1 if the ﬁrm is state-owned during ﬁscal year t and 0 otherwise. Age = the number of years from the establishment of China Security Market and Accounting Research the ﬁrm to the sample year t. (CSMAR) database GIscale = the total green investments of the ﬁrm during ﬁscal year t-1 scaled by total assets at the end of ﬁscal year t-1. Leverage = the total debt scaled by total assets, all measured at the end of ﬁscal year t. OCF = the net cash ﬂow generated by the ﬁrm’s operating activities during ﬁscal year t scaled by total assets at the end of ﬁscal year t. Opportunity = the ratio of market value of equity to total assets at the end of ﬁscal year t. ROA = the ﬁrm’s return on assets during ﬁscal year t. Growth = the ﬁrm’s growth rate of operating income during ﬁscal year t. LSR =the ﬁrm’s largest shareholder rate for ﬁscal year t. Duality = 1 if the ﬁrm’s chair of the board is also the ﬁrm’s manager and 0 otherwise. DAGE = the average age of the ﬁrm’s directors. PGDP = the regional GDP per capita of the province China Statistical Yearbook compiled by National where sample ﬁrm is registered for ﬁscal year t. Bureau of Statistics of China Structure = the provincial gross industrial output scaled by the regional GDP of the province where sample ﬁrm is registered for ﬁscal year t. ENpressure = the annual average PM2.5 concentration of the GREENPEACE – the international environmental province where sample ﬁrm is registered for protection organization ﬁscal year t-1. A higher value of ENpressure indicates ﬁrms that are located in this province are facing higher pressure of environmental protection. ENsupervision = 1 if the ﬁrm is registered in the province where The website of Ministry of Ecology and the central government conducted Environment of the People’s Republic of China environmental protection supervisions during (MEEC) ﬁscal year t and 0 otherwise. 92 Y. CHEN AND J. FENG Appendix B CNY 100 million 9575.5 9539.0 9219.8 9037.2 8806.3 8253.5 7612.2 7114.0 5258.4 4937.0 3668.8 2779.5 2565.2 2057.5 1750.1 1456.5 1166.7 1062.0 Year 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Figure B1. Histogram of China’s environmental investments from 2000 to 2017. %GDP 1.86 2.00 1.57 1.54 1.53 1.52 1.49 1.45 1.39 1.38 1.50 1.29 1.29 1.28 1.28 1.24 1.21 1.15 1.06 1.02 1.00 0.50 0.00 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Year Figure B2. The proportion of China’s environmental investments in GDP from 2000 to 2017.

Journal

China Journal of Accounting Studies – Taylor & Francis

Published: Jan 2, 2019

Keywords: Corporate green investment efficiency; green investments; environmental performance; local environmental enforcement; SBM-DEA

References

Green tangible investment strategies and export performance: A firm-level investigation
Proactive environmental strategy in a supply chain context: The mediating role of investments
The effect of proactive adaptation on green investment
Environmental investment and firm performance: A network approach
Testing the Porter hypothesis: The effects of environmental investments on efficiency in Swedish industry
Corporate social responsibility and the pollution haven hypothesis: Evidence from multinationals’ investment decision in China
Ecoefficiency: Defining a role for environmental cost management
The pyramid of corporate social responsibility: Toward the moral management of organizational stakeholders
Environmental efficiency analysis of transportation system in China: A non-radial DEA approach
Measuring the efficiency of decision making units
Do actions speak louder than words? An empirical investigation of corporate environmental reputation
What are the determinants of investment in environmental R&D?
Stakeholders and environmental management practices: An institutional framework
A model of green investments approach
A survey and analysis of the first 40 years of scholarly literature in DEA: 1978–2016
Green investment: Trends and determinants
Value drivers of corporate eco-efficiency: Management accounting information for the efficient use of environmental resources
Pollution abatement costs and foreign direct investment inflows to u.s. states: a nonparametric reassessment
The determinants of an environmentally responsive firm: an empirical approach
Eco-efficiency analysis of paper mills along the Huai River: An extended DEA approach
Corporate social responsibility research in accounting
Defining and measuring green investments: implications for institutional investors’ asset allocations
Interaction between output efficiency and environmental efficiency: Evidence from the textile industry in Jiangsu Province, China
Green investment scheme: experience and results in the czech republic
Pollution abatement costs and foreign direct investment inflows to U.S. states
Deregulation and differentiation: Incumbent investment in green technologies
Measuring the efficiency of the investment for renewable energy in Korea using data envelopment analysis
Environmental regulations and manufacturers’ location choices: Evidence from the Census of Manufactures
Regional environmental efficiency evaluation in China: Analysis based on the Super-SBM model with undesirable outputs
Data envelopment analysis 1978–2010: A citation-based literature survey
Firm performance and the role of environmental management
Local environmental non-profit organizations and the green investment strategies of family firms
Managers’ green investment disclosures and investors’ reaction
Voluntary environmental investment and responsive regulation
Why do patterns of environmental response differ? A stakeholders’ pressure approach
Toward a new conception of the environment-competitiveness relationship
See what we want to see? The effects of managerial experience on corporate green investments
Investment in cleaner technology and signaling distortions in a market with green consumers
Media monitoring, government supervision, and corporate environmental disclosure
The link between firm financial performance and investment in sustainability initiatives
Quantitative analysis of foreign trade and environmental efficiency in China
Risky multi-criteria group decision making on green capacity investment projects based on supply chain
Environmental regulation and sustainable competitiveness: evaluating the role of firm-level green investments in the context of the porter hypothesis
A slacks-based measure of efficiency in data envelopment analysis
Green investments – between necessity, fiscal constraints and profit
Environmental regulation, emissions and productivity: Evidence from Chinese COD-emitting manufacturers
Government’s green grip: Multifaceted state influence on corporate environmental actions in China
Can more environmental information disclosure lead to higher eco-efficiency? Evidence from China
Does central supervision enhance local environmental enforcement? Quasi-experimental evidence from China
Corporate behavior and competitiveness: Impact of environmental regulation on Chinese firms

Sign up for your 14-Day Free Trial Now!

Read and print from thousands of top scholarly journals.

System error. Please try again!

or

By signing up, you agree to DeepDyve’s Terms of Service and Privacy Policy.

Already have an account? Log in

Cite

Copy and paste the desired citation format or use the link below to download a file formatted for EndNote

APA
Chen, Y., & Feng, J. (2019). Do corporate green investments improve environmental performance? Evidence from the perspective of efficiency. China Journal of Accounting Studies, 7(1), 62-92.
MLA
Chen, Yutao, and Jian Feng. "Do corporate green investments improve environmental performance? Evidence from the perspective of efficiency." China Journal of Accounting Studies 7.1 (2019): 62-92.

Reference Managers
Export to EndNote
Export in RIS Format

Our policy towards the use of cookies

All DeepDyve websites use cookies to improve your online experience. They were placed on your computer when you launched this website. You can change your cookie settings through your browser.

cookie policy

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Do corporate green investments improve environmental performance? Evidence from the perspective of efficiency

Do corporate green investments improve environmental performance? Evidence from the perspective of efficiency

Do corporate green investments improve environmental performance? Evidence from the perspective of efficiency

Abstract

Journal

Recommended Articles

References

Our policy towards the use of cookies