Do Higher Education Research and Development Expenditures affect Environmental Sustainability? New Evidence from Thirty-One Chinese Provinces

Higher education R&D expenditures (HEEXP) are one of the important determinants of economic growth that facilitates science, technology, new ideas, and innovation, but its effect on environmental sustainability remains unexplored. This paper examines the nexus between HEEXP and carbon dioxide emissions (CO 2 e), followed by control variables such as electricity consumption, foreign direct investment, gross domestic product, and total population for the period 2000Q1-2019Q4. Some of the key results are as follows. First, the present ndings conrmed the long-run cointegration among variables. Second, the nding showed signicant long-term negative nexus between HEEXP and CO 2 e. Third, the ndings indicated that electricity consumption, foreign direct investment, gross domestic product, and total population are the important factors that intensify the overall situation of CO 2 e. Fourth, the results indicated that there exist a bi-directional causality between EC and CO 2 e; FDI and CO 2 e; GDP and CO 2 e; POP and CO 2 e and HEEXP, and CO 2 e. new power systems, and sequestration, clean ecological grassland development, recycling economy, biofuels, bioproducts, and integrated gasication combined systems.


Introduction
Environmental pollution is a major threat to the environment of the world. Rising economic growth and industrialization in emerging economies have fuelled the irresponsible consumption of fossil fuels. Apart from the speedy depletion of natural resources, this situation has contributed to the emanation of more waste, residues, and green-house gases (GHGs) into the environment. These toxic emissions of various types are considered as primary causes of global climate change, rising temperatures, and air pollution. Among them, carbon dioxide is one of the leading pollutants, accounting for about sixty-three percent of the total GHGs (Sharif Hossain, 2011;Wei, 2020). Wei (2020) further reported that the global mean temperature has upsurge by 0.74 centigrade during the last ten decades. Theoretically, the association between gross domestic per capita (GDP) and CO 2 e is directly linked to the consumption of different types of carbon-intensive natural resources, especially fossil fuels. Many scholars have argued that CO 2 e, fossil fuel consumption, and economic progress are intimately correlated. Researchers have stated that massive industrialization, resulting from an increase in economic activities, escalates the rate of energy consumption from various non-renewable sources, thereby causing CO 2 e (Rehman, Rauf, Ahmad, Of the few studies in the economics literature, scholars have used education as a control variable, predominantly using student numbers or percentage of students as proxies. None of the prior studies in both the disciplines have linked the HEEXP to CO 2 e. The rest of the paper is categorized as follows. Section 2 explores the literature review. Section 3, 4, 5, and 6, present the conceptual framework, model speci cations, and data sources and variables, and estimation techniques, respectively. Section 7 focuses on the interpretations of results and discussions, followed by the conclusion, policy recommendations, future directions, and limitations in Sect. 8.
2 Literature Review 2.1 The relationship between income and CO 2 e The close inverted U-shape association between environmental sustainability and economic progress has gained considerable signi cance among scholars, especially during the last three decades. Many believe that rapid economic progress and industrialization affect the environment through the excessive consumption of fossil fuels. Intellectuals have conducted extensive research to nd potential determinants of environmental pollution. Past empirical studies have established that dirty and cheap fuel sources (e.g., coal, oil, and natural gas) have been a signi cant source of increasing global temperature. After the rst industrial revolution, entrepreneurs and economies have been striving to control the CO 2 e levels to prevent the harmful impact of global warming problems. Environmental Kuznets curve (EKC) hypothesis is probably the most frequently tested framework that explains the link between aggregate income and environmental sustainability (Özcan & Öztürk, 2019). Grossman and Krueger (1991) argued that ecological pollution escalates in the initial stage of economic progress due to intense industrial consumption of cheap energy. This situation, however, improves with increased income as more e cient and clean technologies are used in the production process in the latter stages of economic development. This relationship is commonly referred to as the EKC hypothesis. Several researchers have validated the EKC hypothesis for different economies, including but not limited to, Iberia (Moutinho, Madaleno, & Bento, 2020); China (Jiang, Yang, & Ma, 2019;Mushtaq, Chen, Din, Ahmad, & Zhang, 2020;Zhou et al., 2018); India (Dar & Asif, 2017); Pakistan (Ur Rahman, Chongbo, & Ahmad, 2019); USA (Alola & Alola, 2019); Brazil (Ben Jebli & Ben Youssef, 2019); emerging economies (Wawrzyniak & Doryń, 2020); NAFTA and BRIC (Rahman, Cai, Khattak, & Hasan, 2019); Ukraine (Melnyk, Kubatko, & Kubatko, 2016); SEE economies (Obradović & Lojanica, 2017); developed and developing economies (Anser et al., 2020); and OECD (Manzoor Ahmad, Khan, Rahman, Khattak, & Khan, 2019).

Relationship between foreign direct investment and CO 2 e
The positive link between FDI and CO 2 e is known as the pollution-haven-hypothesis (PHH). This concept explains how sources of pollution transfer between countries and regions due to asymmetries in environmental regulations and industrial locations. Prior evidence indicates that pollution-intensive units, factories, or plants facing stringent regulations and policies in rst-world economies moved and sought refuge in developing and the third-world economies where laws were either non-extant or extremely weak.

The relationship between electricity consumption and CO 2 e
Electricity is one of the primary sources of energy for all industries. Even though electricity consumption is not directly associated with CO 2 e, the vast quantities of non-renewable fossil fuels used for power generation emit high CO 2 e (Zhang, 2019). Previously, few academics have examined the relationship between electricity consumption and CO 2 e. For example, Zhang (2019) investigated the relationship between electricity consumption and carbon intensity among twenty-seven rms in China using a STIRPAT framework. The results indicated that electricity consumption played a mitigating role in CO 2 e.
Balsalobre-Lorente, Shahbaz, Roubaud, and Farhani (2018) concluded that electricity consumption increased CO 2 e in the long-run across the European nations. Bélaïd, and Youssef (2017) tested the association between energy (renewable and non-renewable) consumption and CO 2 e for Algeria. The ARDL estimates validated the renewable energy consumption-CO 2 e led hypothesis. Yorucu and Varoglu (2020) studied the nexus among industrial production, electricity consumption, economic growth, and CO 2 e in selected small island states. Based on the FMOLS and DOLS estimations, the authors found that a one percent increase in electricity consumption predicted an upsurge of 0.79 percent in CO 2 e. In the same way, others studies have also reported a positive connection between electricity consumption and CO 2 e for China (Akadiri et al., 2020;Munir & Riaz, 2020;Ou, Xiaoyu, & Zhang, 2011;Xu, Hong, Ren, Wang, & Yuan, 2015;Zhang, 2019); Spain (Zarco-Soto, Zarco-Periñán, & Sánchez-Durán, 2020); South Asian economies (Munir & Riaz, 2019); Bangladesh (Shahbaz, Salah Uddin, Ur Rehman, & Imran, 2014); ASEAN countries (Lean & Smyth, 2010); Pakistan (Rehman et al., 2019) BRICS (Cowan, Chang, Inglesi-Lotz, & Gupta, 2014;Haseeb, Xia, Saud, Ahmad, & Khurshid, 2019); and Kuwait (Salahuddin, Alam, Ozturk, & Sohag, 2018). Figure 3 illustrates the conceptual framework, depicting the mechanism through which HEEXP may affects CO 2 e. For long, the HEIs have been contributing to the advancement of knowledge, economy, cultivating students, and conducting research in many elds. Whether it was the intervention of government or a self-driven agenda, HEIs around the world have undergone enormous transformation and restructuring in areas like organizational practices, research focus, controls, funding structures, and autonomy (Wendt, Söder, & Leppälahti, 2015). Governments' funding, therefore, has been crucial for many HEIs to support basic and advanced level research, especially in elds like environmental sciences, energy and resources e ciency, sustainability, and other similar areas. Many academic institutions have set up separate departments for energy economics, sustainability, green technology, and eco-innovation, while simultaneously initiating programs and activities to achieve green education, green campus, and green economy. With the support of their respective governments, industries, and other institutions, academic institutions are actively conducting research and developing solutions for sustainable production, responsible consumption, and environmental preservation. These projects re ect two facets: i) research on green and sustainability technology, methods, processes, and products; ii) developing and promoting green campuses (GC). Congruent with the above, academic institutions and governments are equally focused on addressing various crucial issues related to energy consumption and production. A possible explanation resides in the energy resources possessed by a country. If the energy demand exceeds the supply, countries are left with no choice but to import expensive energy from other countries that undermine their security and environment. With the potential role of renewable and green energy, technologies, products, and services, many countries and institutions have been investing heavily in academic research and development related to eco-innovation, green technologies, and renewable energy solutions. As a result, the number of eco-related patent applications and green research has increased manifolds in the past few decades across developed and developing nations. In terms of environmental bene ts, these patents have been used across many industries to solve problems, including energy shortages, fossil fuel dependency, and carbon footprint, and low energy e ciency.

Conceptual Framework
Beyond that, academic institutions have been developing and institutionalizing the concept of green campus (GC) and green education. Simply put, GC embodies the development of two critical aspects in an academic institution: a) energy and resource-e cient campus (ERSC); b) campus energy management system (CEMS). The concept of CEMS emphasizes the construction of green education and environmental-related technologies for ERSC. The ERSC, however, requires the integration of green ideology into capital operation, infrastructure, logistics, and other departments. The primary purpose of GC is: to achieve energy and resources e ciency by saving materials, water, energy, land; promote the use of green and clean energy sources during o cial hours; encourage sustainable development in higher education; improve R&D for faculty, staff, students, and society at large; enhance stakeholder engagement on sustainable decision making; sponsor students and faculty participation in green and sustainability-related activities; and to designing and implement green curricula. Thus, GC plays a vital in the implementation of the sustainable development goals and green policies. Above all, the exchange and cooperation activities among academic institutions for the advancement of GC ideology offer multiple bene ts, in terms of national policy formulation for GC development; attainment of Strategic Development Goals, encouraging collaborative research, enabling the diffusion of carbon and energysaving programs, innovation, and carbon-reduction technology in HEIs, initiating training programs for faculty members, and establishing real-time experiment, labs, and demonstration centers for green research, education, green campus development, and strategy implementation. Through the proper utilization of HEEXP, the GC can nd a new way to set the foundations for disseminating the soft power of eco-protection, achieving low-carbon goals, and enabling a smooth transition to a green economy and campus. That said, the development of GC necessitates the need for educational institutions to focus on the hardware and software of GC, simultaneously. The former pertains to the integration of green aspects in construction, building, infrastructure, and operations, and the latter refers to the development and promotion of green culture, humanity, green citizenship, and cultivation of talent for social entrepreneurship. This process, if properly executed, will result in the formation of core green values at all levels (economy, education, society, business), enabling sustainable progress (Tan, Chen, Shi, & Wang, 2014). In short, it is proposed that the development of GC (through HEEXP) not only helps in mitigating CO 2 e, but also play an important role in promoting sustainable consumption and production across residential and commercial sectors.
The rationale for the use of FDI, EC, GDP, and POP as control variables is brie y discussed henceforth. First, China has become one of the most attractive FDI destinations due to low labor costs and weak environmental regulations. Many multinational companies from developed nations have transferred their technologies (FDI), converting China into a pollution-haven. Second, China is among the top energy generation countries, where almost eighty percent of electricity was generated from coal. Third, it is one of the largest economies in the world, vis-a-viz the GDP growth rate. Fourth, China is one of the most populous economies in the world, where population growth has created contributed to energy consumption among residential and non-residential consumers, directly and indirectly causing CO 2 e.

Data Sources And Variables
The data for HEEXP, FDI, EC, GDP, POP, and CO 2 e were collected from the National Bureau of Statistics (2019) for the period 2000 to 2019. Consistent with the previous studies (Sbia, Shahbaz, & Hamdi, 2014;Shahbaz, Hoang, Mahalik, & Roubaud, 2017), the accuracy and frequency of the data were enhanced through the quadratic match-sum method. All variables were converted into logarithmic forms for added reliability and consistent results. Table 1 shows the data sources and descriptions. Testing the cross-sectional dependence (CSD) among the series was the rst step in the panel data analysis. This test was conducted to identify and deal with the problems of unit-root and CSD in the data series. As the CSD is associated with factors, including, economic union, nancial shocks, demand shocks, supply shocks, pandemic diseases, globalization, and trade wars, it must be dealt with accuracy and precision. If ignored, it could be led to bias cointegration and stationarity results (Z. Khan, Ali, Jinyu, Shahbaz, & Siqun, 2020). The Pesaran (2015) cross-sectional dependence test (PCSDT) and the M.H Pesaran and Yamagata (2008) slope homogeneity test (SHT) were applied for addressing the CSD and homogeneity problems, respectively. In the next step, the order of integration was examined for all variables using the second-generation Pesaran and M.H (2003) (PMCADF) and Pesaran (2007) (PCIPS) unit-root tests. Conventional or rst-generation panel unit-root tests are based on the hypothesis of crosssectional independence (CSI). The second-generation unit-root tests, however, allow for the assumption of CSD in the data series. With the results of second-generation tests providing strong evidence on the existence of CSD across the provinces in China, these tests were appropriate for estimating the order of integration. For robustness check, the Clemente, Montañés, and Reyes (1998) unit root test (CMRURT), with multiple structural breaks was employed the aggregate data on EC i , FDI t , GDP t , POP it , HEEXP t , and CO 2 e it .

Cointegration testing
For cointegration testing, this study adopted the Westerlund (2007)  The paper adopted three cointegration tests for checking robustness- Kao (1999) residual-based cointegration test (KRCPT), Pedroni (2004) cointegration test (PCT), and the Gregory and Hansen (1996) cointegration test (GHCT) (with structural breaks and regime shifts). 6.3 Long-run coe cients estimation Several economic techniques have been introduced in the past decades for addressing the CSD and parameter heterogeneity problems. Some of the widely accepted methods include the M. Hashem Pesaran and Smith (1995)  , and an intercept to deal with xed components. Then, this estimator averages the computed individual-speci c slope (without or with wrights). For dynamic cases, this estimator proves to be reliable for large N and T, if the coe cients exhibit heterogeneity in groups. This estimator, however, fails to offer information about common factors (CFs), which may exist in the panel data. The CFs are referred to as time-speci c effects, which are common in provinces, countries, or regions. By incorporating the averages of the cross-sections of the independence and the dependent variables as surplus regressors when applying OLS to speci c units, the CCEMG method allows for TVU and CSD with heterogenous effect in panel members. Identi ed by the averages of CS, the unobserved CF can be any xed digit. With superior small sample characteristics and short-run estimation properties, the CCEMG technique is relatively robust to non-cointegrated and nonstationary CF, structural breaks, and some serial correlations. As an alternate method, the AMG initially computes an augmented pooled model (with year dummies) through the rst difference OLS. The calculated year dummies are then compiled to construct a new variable, representing the common dynamic process. This new variable is used as an extra regressor for single group-speci c regressor model, along with an intercept for capturing the time-variant xed impacts. Similar to the CCEMG technique, the AMG method helps in dealing with multi-factors error-terms and non-stationary variables, particularly considering CSD. The AMG estimator is superior to the CCEMG, in terms of creating a set of unobservable CF as a common dynamic process. Dissimilar to a scenario in which the unobservable factors are considered as a nuisance, the alternate treatment may offer helpful interpretations, depending on the context (Heshmati, 2019).

Panel causality testing
For panel data, Dumitrescu and Hurlin (2012) proposed a test to examine causal relationships between variables. This test outperforms the traditional causality tests by allowing for the hypothesis of causality existence in at least one cross-section, against the non-existence of homogenous Granger-causality relationship. In this way, the Dumitrescu and Hurlin (2012) panel-causality test (DHPCT) accounts for the CSD between the sample province or countries. Moreover, the DHPCT is insensitive to the variance among the cross-sections and the time difference in the panel. It generates e cient results, even if the size of the cross-sections and time series are smaller or larger than others (Ceyhun, 2019). Table 2 depicts the results of PCSDT. As seen below, the null hypothesis of no CSD for the EC it , FDI it , GDP it , POP it , HEEXP it , and CO 2 e it was rejected at 10, 5, and 1 percent signi cance levels. This implied that all the provinces in China were interdependent in a way that an economic shock in one region may affect other regions, too. As reported in Table 3, the SHT highlighted heterogeneity problems in the model.   Table 4 displays the results of the PCADF and PCIPS unit-root tests. These tests were used to check the integration order of all the study variables. The results con rmed that all the study variables were nonstationary at level but became stationary at the rst difference, even though these tests were unable to deal with structural breaks in the data. Given that most global economies have experienced many structural changes, it is considered imperative to trace structural breaks in the data series for China. There was a high probability that the PCADF and PCIPS could be given bias results, if structural changes were underestimated. This problem was addressed through the CMRURT that allowed for multiple structural breaks in the data.   (Wong & Zheng, 2004). In 2004, China faced one of the worst historic in ationary pressures, partly triggered by real-estate speculations. With an increase in the costs of raw material and energy and over-investments in some industries, China raised interest rates and applied administrative control to abate the pace of investment in some sectors and industries (Morrison, 2010). In 2005, Lenovo Group acquired the personal computer division of IBM for a hefty sum of USD1.75 billion. Indeed, this acquisition is considered as an economic breakthrough. Apart from gaining access to foreign, facilities, operations, and R&D, China strengthened its presence in the US (Morrison, 2010). From 2008-2009, the global nancial crisis pushed China to revisit its economic policies to sustain economic growth. While the economic growth rate was disrupted in 2009 relative to the past years, this slowdown in growth was reasonably modest, especially when compared with the total shrinkage in the world output (Lardy, 2012). Although the incoming FDI experienced a sharp decline, the inbound foreign investments reached an all-time high in 2010, increasing by around two-third, i.e., USD185 billion. There was almost twenty percent contraction in outbound FDI in 2009, but the outbound FDI increased by thirty-seven percent and touched an all-time high of USD60 billion (Lardy, 2012). Moreover, the inclusion and internationalization of RMB in the Special Drawing Rights currency basket by the IMF in 2010 was another important milestone, which enabled China to expand its nancial presence in the global nancial markets (Cassis & Wójcik, 2018). With all the study variables exhibiting the same integration order, the study applied the cointegration analysis, including the WECPT, KRCPT, PCT, and the GHCT. Note. CO 2 e = Carbon dioxide emissions; EC = Electricity consumption; FDI = Foreign direct investment; GDP = Gross domestic product; POP = Population, HEEXP = Higher education R&D expenditures. **, *** indicates 5% and 1% level of signi cance, respectively. Table 6 depicts the outcomes of the cointegration analysis without structural breaks. The rst two columns (G t , G a ) indicate the group means statistics for the total cointegration, whereas the remaining two columns (Pa, Pt) show panel statistics. The WECPT outputs con rmed a sustainable long-term association among all the study variables. In Table 7, the results of the cointegration analysis with structural break and regime shifts were found to be consistent with the WECPT, KRCPT, and the PCT.  (2007)  *, **, *** indicates 10%, 5% and 1% level of signi cance, respectively.  alone, the faculty and staff from HEIs constituted 11.3 percent of the overall research and development population. Using almost 8.5 percent of the total national R&D spending, these researchers have shown impressive results. These individuals conducted 62.2 percent of the all research projects and activities, received 28.8 percent of the total patents, applied for 21.6 percent of the total patents, and produced 64.4 percent of the entire scienti c publications. Following the 'new normal' of fostering the nation with education, science, innovation, and developing a green economy, the Chinese government has placed a signi cant focus on green and sustainable technology research. Currently, Chinese scholars are the leading the global research related to green production, sustainability, green technology, environment, and green energy. More so, the government has been allocating a considerable amount of funds for sustainability-oriented R&D projects. From 2000-2009, these funds have increased from just RMB7.67 billion to RMB46.7 billion, constituting almost eight percent of the total national spending on R&D. A total of RMB14.5 billion were allocated to basic research, accounting for nearly fty-three percent of the total national research budget (Hu, Liang, & Tang, 2017).

Results And Discussion
Next, China initiated the 211 Project and 985 Project to uplift the standard of its HEIs. These projects were aimed at developing globally competitive rst-class universities, programs, and scienti c disciplines to promote sustainable and green socio-economic development in China. Hu, Liang, and Tang (2017) argued that the fteen years of the 211 Projects have been extremely fruitful, in terms of setting the foundations for green innovation in education, research and service, and transitioning to a green economy. China spends around two percent of its total GDP on research, an amount that is increasing at the rate of twenty percent per year (Chung, 2015). Under the government's guidance, Chinese HEIs have dedicated time, resources, and money for research on green energy, economy, technology, education, and innovation to realize a green revolution (Liu, Strangway, & Feng, 2012). These factors have played an instrumental role in indirectly mitigating CO 2 e by raising awareness, development of green technology, and green urbanization, and green education.
In the same vein, China has been investing heavily in the green university/campus project. Many topranking and globally-recognized universities have joined hands with the government to realize the Sustainable Development Goals. For Instance, Tsinghua University has been championing the idea of green campus (GC), green technology, and green education. Peking University initiated the green university project in April 2009. As an initial step, the planning department was rebranded as the Campus Planning and Sustainable Development O ce. Beijing University has set four key objectives for achieving the GC and educational goals: 1) spatial design augmentations of the university; 2) improved and continued excellence of scienti c research and teaching; 3) propagation and restoration of culture and environmental heritage; and establishment of zero-carbon campus (Morgan, Gu, & Li, 2017). Lee and E rd (2014) further explained the idea of green universities by identifying some key attributes. Firstly, these universities place acute emphasis on environmental education and integrate environmental aspects in the teaching, research, and curriculum. Secondly, the student and faculty master the knowledge, skills, and expertise on topics related to environmental protection, sustainable development, and environmental awareness. Thirdly, the members of the green universities actively engage in the society-focused programs for environmental publicity, evaluation, and education. Fourthly, the environment becomes an important part of the campus culture, and it is integrated into all campus policies to develop a cleans and green campus environment. Gou (2019) added that green campus operations are linked to all areas, including, labs, classrooms, transportation, dormitories, and other facilities. Thus, the idea of green campus and green education entails several economic bene ts, especially for a massively-populated country like China. The GC can help to save energy, water, and other precious resources in China, particularly if the consumption of energy and water among HEIs is higher than the residential consumers.
Apart from enabling the generation of new ideas and patents for green production, innovation, technology, and economy, the macro impact of the GC resides in improved e ciency and social fairness in the usage of natural resources. For ecological advantages, all HEIs need to revisit their effects on energy e ciency by transforming their facilities to preserve the environment. Beyond that, the social bene ts of the GC include the conversion of students and teachers into conscious and eco-friendly consumers. Thus, the GC has the potential to reduce deprivation and poverty among regions or provinces, enhance fairness, and to expand the sustainable growth concept in the Chinese society. All these measures, if implemented correctly, can decrease CO 2 -related energy consumption and increase the use of clean technologies across China. Table 9 exhibits the parallel uctuations in HEEXP and CO 2 e.  Second, the long-run coe cients indicated a signi cant positive linkage between FDI and CO 2 e, offering empirical evidence for the acceptance of the PHH in China. A one percent increase in FDI caused a rise in CO 2 e by 0.42 (MG), 0.12 (AMG), and 0.34 (CCEMG) percent. This result suggested that some cities, provinces, and municipalities in China, with less stringent regulations, have become pollution havens in an attempt to attract FDI and pollution-intensive industries. This result validated the previous studies conducted for China (Ur Rahman et al., 2019); OECD (Manzoor Ahmad et al., 2020); newly industrialized nations (Destek & Okumus, 2019); Cote d'Ivoire (Assamoi et al., 2020); ASEAN (Guzel & Okumus, 2020); MINT countries (Balsalobre-Lorente et al., 2019); Pakistan (Nadeem, Ali, Khan, & Guo, 2020;Naz et al., 2019); MIKTA economies (Bakirtas & Cetin, 2017); BRICS (Z. U. ; Arab countries (Abdo, Li, Zhang, Lu, & Rasheed, 2020); Asian countries (M. A. Khan & Ozturk, 2020); and European countries (Mert et al., 2019). However, this results contradicts the previous studies conducted for coastal Mediterranean countries (Nathaniel, Aguegboh, Iheonu, Sharma, & Shah, 2020); OECD (Manzoor Ahmad, ; Turkey (Mert & Caglar, 2020); China (Ayamba, Haibo, Ibn Musah, Ruth, & Osei-Agyemang, 2019;Hao, Wu, Wu, & Ren, 2020); and Kyoto Annex countries (Mert & Bölük, 2016).
Third, the estimations revealed a positive association between GDP and CO 2 e-a one percent increase in GDP led to a rise in CO 2 e by 0.44 (MG), 0.75 (AMG), and 0.64 (CCEMG) percent. This result suggested that GDP growth-driven by low energy e ciency and coal consumption-had enhanced CO 2 e in China.
Fifth, the results revealed a positive electricity use-CO 2 e nexus, implying that the irresponsible consumption of electricity (by educational, residential, and industrial consumers) had signi cantly enhanced CO 2 e in China. This nding points towards the heavy reliance on carbon-intensive energy sources (e.g., coal, and oil) for domestic and industrial consumers by the power generation sector. That said, the new energy policies and installed-capacity forecast suggest that the over-dependency on fossilfuels will reduce signi cantly in the future, thereby decreasing CO 2 e. The commercial sector (e.g., tech companies) is also setting the foundations for responsible energy consumption by switching from conventual to renewable energy sources. As some tech companies have started using solar and wind for power generation, other sectors will also follow this campaign to reduce their carbon footprint. This result validates the previous studies conducted for China (Akadiri et al., 2020;Munir & Riaz, 2020;Xu et al., 2015;Zhang, 2019); Spain (Zarco-Soto et al., 2020); South Asian economies (Munir & Riaz, 2019); Bangladesh ; ASEAN countries (Lean & Smyth, 2010); Pakistan (Rehman et al., 2019) and BRICS (Haseeb et al., 2019).
Finally, Table 10 exhibits the results of the DHPCT. The causality estimates revealed a bi-directional causality between EC and CO 2 e; FDI and CO 2 e; GDP and CO 2 e; POP and CO 2 e and HEEXP, and CO 2 e.
These results suggested that government policies that target EC, FDI, GDP, POP, and HEEXP have, directly and indirectly, led to an increase or decrease in CO 2 e.

Conclusion And Policy Implications
The main objective of this study was to explore potential long-run connections between the HEEXP and CO 2 e for thirty-one provinces in China from 2000(Q1) to 2019(Q4). The panel data were analysed using the multiple econometric techniques. First, the results of the WECPCT, KRCPT, and PCT indicated that a long-term cointegration existed between all the study variables. Second, the MG, AMG, and CCEMG supported that the HEEXP had disrupted CO 2 e, while EC, FDI, POP, and GDP had a positive interaction with CO 2 e in the long run. Third, the DHPCT re ected that a two-way causal relationship existed between CO 2 e and all other study variables-FDI, EC, GDP, POP, and HEEXP.
The following important implication were drawn from current ndings. Firstly, the current ndings assert the need for the policymakers to design speci c policies for green education, green campus, green economy. Chinese government should extend nancial support to encourage its academic institutions for developing green patents and conducting research on projects related to energy e ciency, sustainable production, green consumption, and preservation of land, soil, and environment. With the nascent awareness of environmental standards and norms, an extensive capacity building is across all academic institutions to align these institutions with global standards, eco-innovation, and sustainability practices.
Second, the current results also require the need for the adjustment of research themes with the national energy and sustainable development plans. For this purpose, the HEEXP policy should be designed in a manner that the rewards, incentives, bonuses, and funding for academic institutions are based on the quality and quantity of eco-related patents and research. These institutions should be directed to develop matrices aligned with national themes and sustainability targets, including but not limited to clean and e cient transport technologies, solar thermal technology, solar cells, wind power, new nuclear power systems, carbon capture and sequestration, clean coal, ecological conservation, grassland development, recycling economy, biofuels, bioproducts, and integrated gasi cation combined systems.
Third, the acceptance of the PHH in this study has strengthened the previous argument that FDI in developing countries have enhanced dirty technologies. Thus, policymakers are expected to tighten the environmental regulations, ensure that foreign enterprises transfer clean technologies, and improve green investment. Fourth, the positive connection between CO 2 e and electricity use calls for not only revisiting the existing energy mix, but also asserts the need for devising energy e ciency strategies to curb CO 2 e.
Policymakers should, therefore, continue to clean and expand the energy mix with more renewables for electricity generation to meet future demand. While encouraging and supporting the commercial sector to deploy solar and wind for power generation, the government should formulate energy e ciency policies for resources management, regardless of its types, i.e., non-renewable or renewable energy. If ine ciently managed, these resources face the risk of depletion. Thus, the future policies for a green economy should incorporate e cient resources management, solar and wind energy development, technology improvements, carbon-taxing, and green urbanization. Of particular signi cance, all these policies should be designed, integrated, and coordinated with multiple stakeholders (i.e., community, government, academia, and administration) for effective execution and results.
Fifth, the current ndings concerning the adverse effect of the population on the environment assert the need for developing a responsible and eco-driven aging sector. This argument stems from the fact that a signi cant majority of the existing population in China is predicted to experience aging, leaving a wide gap in the workforce in the future. While this phenomenon may decrease the level of CO 2 e, it necessitates the need policies that guarantee better healthcare, social justice, social security, and other related facilities across all provinces. If this issue is underestimated, the socially deprived and unsatis ed populace may contribute to CO 2 e, thereby disrupting the green transformation. Thus, policymakers should devise policies to encourage investments in the aging sector to address the potential future disruption in economic growth. That said, this new sector should be built on the foundations of energysaving, responsible consumption, social equality, income equality, old-age security, and equal access to quality healthcare for all provinces.
This study has some limitations that open new doors for future research. First, this study had only focused on China. The same model can be used for other developing and developed economies. Second, this study applied linear econometric techniques (MG, AMG, and CCEMG) to explore the relationship between HEEXP and CO 2 e. Perhaps, some non-linear models (e.g., NARDL) can be used to explore the same relationship and variables in a uni ed framework. Third, the current has adopted the EKC framework for examining different relationships. Researchers are encouraged to tests the current ndings using the STIRPAT framework for new insights.