In a competitive global economy, companies must improve production efficiency to compete, while also increasing revenue to enable better product development compared to competitors (Porter & Heppelmann, 2014), either by employing premium prices or reducing operating costs, or both. The Fourth Industrial Revolution has enabled a significant shift in the digital economy, with business models based on resources of greater value (Bettiol et al., 2020). Applying robotics, component manufacturing, the Internet, big data, and artificial intelligence, has gradually become the norm for enterprises when improving the efficiency of their production and business activities. Thus, cross-border business structures and processes are also significantly affected (Alcácer et al., 2016).

To a certain extent, digital transformation makes it easier for companies to operate internationally and expand their markets (Strange & Zucchella, 2017). Although internationalized processes strengthen competition between businesses, the development of advertising technology has helped businesses to market and promote their products to global customers and accumulate more data. Therefore, according to Azar and Ciabuschi (2017), companies must transform their technology to achieve good export performance. Enhancing innovation and digital transformation are prerequisites that enable exporting companies and their countries to reach higher growth levels. Industrialization can now not succeed without digitalization (Dalenogare, Benitez, Ayala & Frank, 2018), and productive enterprises can improve their efficiency by integrating the product manufacturing processes. Digital technology provides platforms for better communication and utilization of technology by exporting companies, enhancing corporate effectiveness. However, to accomplish optimal adaptation to international markets, enterprises should be innovative in their use of technology (Azar & Ciabuschi, 2017). In addition, the process of global specialization is gradually changing, with notable fragmentation since digital technology has been more widely applied to manufacturing (Laplume et al., 2016). Each production fragment takes advantage of diversity in capability around the world to be closer to purchasers and improve shipping and customer care, increasing competitiveness.

Despite being the subject of controversial opinions among scholars, the impact of digitalization has not yet been examined (Chiarvesio & Romanello, 2018). From a favorable perspective, digital technology may facilitate expansion in terms of both scale and scope by removing limitations of time and geography (Rehnberg & Ponte, 2018). Likewise, big data and accumulated data about customers are closely linked to exporting (Strange & Zucchella, 2017), giving easier access to international markets and enabling companies to navigate these markets. Companies can gather information about new trends in other countries where their products are not yet sold, and exploit product design as a means of gaining competitive advantage, with the ability to identify markets and items suitable for different locations while streamlining production processes. The potential benefits that a company accumulates after analyzing customer data include the ability to take advantage of new business opportunities, particularly when looking to expand distribution to new markets. Where businesses in emerging and fast-growing countries are underperforming, digitalization is seen to have great potential for internationalized businesses intending to expand into new markets (Strange & Zucchella, 2017).

Khan and Hou (2021a) find that economic growth is often achieved at the cost of environmental sustainability; a higher level of economic development is accompanied by reduction in quality of the environment (Khan, Hou, Le & Ali, 2021). Khan, Hou, Irfan, Zakari, and Le (2021) report empirical evidence on the link between energy consumption and economic growth in both the short run and long run. The effects of natural resources, energy consumption, and some economic and social factors on environmental quality are explored by Khan, Hou and Le (2021). New evidence on the association between energy intensity, financial development, and environmental sustainability has been found recently in Asia Pacific Economic Cooperation countries by Khan, Hou, Zakari, Irfan and Ahmad (2022), and in OECD countries by Khan, Zakari, Ahmad, Irfan and Hou (2021). In general, although previous studies have examined both the causes and influences of environmental sustainability, they have not fully exploited the attendants of environmental sustainability. Furthermore, there are channels through which the effects of environmental sustainability, or the influence of other factors on the environment, can either be identified or mitigated. However, prior studies have not considered these channels. To our best knowledge, we are the first to examine the impact of digital transformation in the public sector on enhancing environmental sustainability. In general, previous studies contend that the Internet and technology improvement stemming from digitalization are positively associated with human capital and technological progress. Modern economic growth then shifts the quantity and quality of energy from conventional non-renewables to modern renewables, thus ensuring environmental sustainability (Khan et al., 2021).

Many controversies and discussions have revolved around the impact of digitization on environmental performance, with both negative and positive impacts reported (Liu et al., 2019). According to Feroz et al. (2021), digitalization affects environmental activities directly and indirectly through diverse transmission channels in many ways. The Internet underpins significant achievements in production and business activities in the economy; however, it also indirectly harms the environment (Salahuddin & Gow, 2016). Evidence for this argument is that electricity consumption has increased rapidly since the advent of digitization, leading to a scarcity of resources and the removal of green energy from the energy structure (Majeed & Tauqir, 2020). Despite these shortcomings, in general, digital technologies still contribute significantly to sustainable environmental development through applications and technologies for pollution control, waste management, and sustainable manufacturing and urbanization (Feroz et al., 2021).

To accomplish a circular economy and achieve environmental sustainability, recycling of e-waste and reuse of material still necessitates significant support from innovative, modern technology (Holger et al., 2020). Products with the ability to improve environmental sustainability can be advertised more widely using digital platforms (Feroz et al., 2021), while carbon emission reduction and other waste reduction can also be promoted thanks to the contribution of artificial intelligence (AI), the Internet of Things (IoT) and other technology-based data analytics (Melissa et al., 2019). Regarding uncertain, interactive, and dynamic environmental problems, AI is also considered to be an effective tool to help minimize complexity (Ye et al., 2020). In general, the demand for digital technologies to support environmental sustainability has been examined in different dimensions with different transformation tools. For instance, Weersink, Fraser, Pannell, Duncan and Rotz (2018) assert that food system traceability becomes more convenient and manageable based on big data and other best practices applied to production processes. Furthermore, we can expect that big data will help arrive at a turning point in control of CO2 emissions by enabling the large-scale deployment of next-generation green vehicles. In addition, Sharma et al. (2020) argue that AI and big data can be used with waste management, global warming, geographic information systems, and land-use planning problems. Esmaeilian et al. (2018) and Leng et al. (2020) favor the notion that blockchain plays a crucial role in sustainable manufacturing and business activities, including prolonging product life cycles, maximizing the efficiency of resource usage, and reducing carbon emissions. Gradually, digitalization is becoming key to greener and cleaner manufacturing processes and supply chains (Kerdlap, Low & Ramakrishna, 2019; Mao, Wang, Tang & Qian, 2019; Wang, Liang, Li & Cai, 2018). Substantial evidence demonstrates the effect of digitalization in encouraging green production by reducing the cost of renewable energy (Moyer & Hughes, 2012). In a recent trend, combining smart and sustainable cities, and promoting social welfare associated with ecosystems, has been found to result in sustainable urbanization (Bibri & Krogstie, 2017; Huang, Wu & Yan, 2015; Malik, Sam, Hussain & Abuarqoub, 2018). From the consumer perspective, digitalization stimulates and enhances the demand for non-fossil fuels and more environmentally friendly products (Pickl, 2019). Ultimately, the digital global socioeconomic system has significantly reduced transaction costs between spatially separated actors, eliminated asymmetric information, and further boosted green production and consumption through R&D spillover effects (Zafar, Ullah, Majeed & Yasmeen, 2020).

TGG represents the export value of green goods. Data on trade activities in APEC products were taken from the UN Comtrade database using the six-digit level of the 2007 version of the Harmonized System (HS 2007). Values are all expressed in current USD. To cover the years 1996 to 2019, the HS codes listed in APEC were converted from HS 2007 into HS 1996 using UN Trade Statistics.5 We take a natural logarithm of TGG (LnTGG) before incorporating it into the model.

The key independent variable, DPSi,t={DPS_UCi,t,DPS_BMi,t,DPS_KEi,t}, consists of three different indicators. DPS_UC denotes user-centricity that captures the extent to which (information about) public service is provided online, how the online journey is supported, and whether public websites are mobile-friendly. DPS_UC is calculated as the weighted average of indicators reflecting the level of online availability, usability, and mobile friendliness. DPS_BM is business mobility that captures the extent to which public services that are aimed at foreign businesses are available online, usable, and implemented with electronic identification (eID) and electronic document (eDocument) capabilities. This indicator is calculated as a weighted average of business mobility online, usability, eID cross-borders, and eDocuments cross-borders. DPS_KE is a key enabler that captures the extent to which technical pre-conditions for eGovernment service provision are used. The key enablers used for measuring the quality of the services to businesses and citizens include eID and eDocument, authentic sources, and digital posts. We take the data for DPS implementation from the eGovernment Benchmarking report and studies for digitalization conducted by Capgemini. The dataset is available from 2012 to 2019.

Regarding other control variables, we use theories of international trade and empirical studies in the environment literature, such as Fu, Chen, Jang and Chang (2020), as well as in the trade literature, such as Agosin, Alvarez and Bravo-Ortega (2012), Cadot, Carrère and Strauss-Kahn (2011), Espoir (2020), Gnangnon (2019) and Parteka and Tamberi (2013), to select explanatory variables. The set of explanatory variables includes income level (INC) measured by real gross domestic product per capita at a constant 2010 price, as in Cadot et al. (2011), Parteka and Tamberi (2013), and many other studies. The other main macroeconomic indicators are savings (SAVE) as a share of GDP, inflation (INFLA) measured as the annual percentage change in GDP deflator,6 as in Ben Hammouda, Karingi, Njuguna and Sadni Jallab (2006). Also incorporated into the baseline model are population level (POPU), which is taken as the natural logarithm of the total population, and industrialization level (INDUS), which is the value added to GDP. These variables are available from the World Development Indicators (WDI). We follow Cabral and Veiga (2010) to consider the influence of political and institutional variables, including level of democratization (DEMO) from the democratization index taken from the World Bank Governance Indicator (WBGI).

To start a procedure of empirical estimation regression, we first merge data from different sources and clean the merged data to ensure there are no gaps in the panel data. The detailed descriptions of included variables; including definitions, measurements, data sources, and descriptive statistics, are shown in Table 1. The final sample consists of 208 observations covering 25 European countries, from 2012 to 2019. Detailed information on these countries is provided in Table A.1 in the Appendix. The correlation matrix between all variables is displayed in Table 2, which presents the correlation coefficients between the variables. The results show a positive correlation between lnTGG with DPS_UC and DPS_BM, but its association with DPS_KE is negative. Since our sample is characterized by a sample N and short T period, cross-sectional dependence (CD) tests are applied (Pesaran, 2021). The results of the CD test are displayed in Table 3, and show that most of the variables in the model have CD except for the DEMO variable. Hence, we apply the PCSE model to examine the association between TGG and DPS. According to Beck and Katz (1995), the PCSE method helps deal with issues arising from the simultaneous correlations between objects in a full N x N cross-sectional matrix. To address the possibility of multicollinearity with heteroskedasticity and endogeneity, we take the one-year lag of all the explanatory variables. As contended by Canh and Thanh (2020) and Schneider and Enste (2000), the simultaneity resulting from the possible feedback effect of TGG on other economic variables can be resolved in this way. For a robustness check, we also employ an FGLS model to minimize the variance of change, as recommended by Canh, Schinckus, Thanh and Hui Ling (2020) and Liao and Cao (2013). The two-step GMM is also applied to further deal with the endogeneity presented in Eq. (1), to ensure the accuracy of our findings, as Gala, Camargo, Magacho and Rocha (2018) and Sweet and Eterovic (2019). In addition, this study measures the short-term and long-term relationship between DPS and TGG. To serve this purpose, the ARDL method is employed. According to Pesaran and Smith (1995), this method helps to distinguish between long-run and short-run effects for exogenous and endogenous variables (Table 4). Furthermore, potential problems related to the existence of time- and country-specific fixed effects are also addressed through the dynamic fixed-effects estimator (DFE) employed in the ARDL model (Canh & Thanh, 2020). For further discussion, we examine the moderating effects of institutional quality in the relationship between DPS and TGG. Interaction between variables reflecting institutional quality and digitalization variables is incorporated in Eq. (1).

This paper examines the relationship between DPS and TGG. We report the estimation result of the baseline model in Table 5, obtained from the PCSE estimate, FGLS estimate, and two-step GMM estimate. In general, the results of all three estimation methods are not substantially different. All models show that the coefficients of the variables representing levels of DPS are positive and statistically significant. This result implies that the implementation of DPS plays a critical role in enhancing the export value of green goods. Among the three proxies of DPS, user-centricity has the most significant influence, and the results are consistent, as demonstrated by all three estimation methods. The estimation results emphasize the importance of DPS in promoting the TGG between countries. Our findings are in line with previous studies in the literature. Numerous papers highlight the influence of technology and digitalization on trade performance (Alcácer et al., 2016; Azar & Ciabuschi, 2017; Bettiol et al., 2020; Laplume et al., 2016) and trade diversification (Chiarvesio & Romanello, 2018; Rehnberg & Ponte, 2018; Strange & Zucchella, 2017). Other studies also reveal that digital technologies may help alleviate pressure on the natural environment and biodiversity, which are representative of ecosystem vitality (Liu et al., 2019). As shown by Feroz et al. (2021), digital transformation is vital for environmental sustainability, since digitalization leads to improved pollution control, waste management, sustainable production, and urban sustainability. Digitalization is a crucial driver of sustainable development (TWI2050 - The World in 2050, 2019). The influence of digital transformation on the environment and sustainability in the European region is also demonstrated by Liu et al. (2019).

Based on the results in Table 5, we analyze the influence of control variables on the export values of green goods. The results reveal that real output growth (INC), inflation (INF), and level of democratization (DEMO) all have a negative impact, and most are statistically significant except for INC, which is statistically insignificant on TGG. In contrast, savings (SAVE), population (POPU), and level of industrialization (INDUS) are significantly positive. Notably, the coefficient of INDUS in the PCSE and FGLS estimates is significant compared to other variables. This suggests that INDUS plays a vital role in enhancing the export value of green goods. The positive effects of POPU and INDUS are consistent with Fu et al. (2020). In summary, the results generally support the argument that increases in savings, population, and level of industrialization contribute positively to the increase in TGG; while increases in real GDP per capita, inflation, or level of democratization decrease TGG. Similar findings regarding the effects of economic growth and international trade can be found in Lyu et al. (2021). The effects of industrial value-added, capital formation, urbanization, population growth, and biocapacity, can also be found in Yang and Khan (2021).

As discussed above, this paper also examines the impact of DPS on TGG in the short term and long term. The results are shown in Table 6, with some evidence of a significant relationship in the short term. Meanwhile, the coefficients of the DPS variables are positive and significant at a 1% significance level in the long term. It is worth noting that the coefficient of business mobility, in the long run, is significant compared to the other two variables, indicating that implementing activities to improve the business and operating environment plays a critical role in promoting TGG in the long term. By following a similar empirical estimation procedure, Ha (2022) also highlights the role of digital business and digital public services in securing various dimensions of the energy system in the long run.

For further analysis, we examine the role of institutional quality in the relationship between DPS and TGG. In this section, we take six indicators of level of institutional quality from the International Country Risk Guide: voice and accountability (VA); political stability and absence of violence/terrorism (PV); government effectiveness (GE); regulatory quality (RQ); rule of law (RL); and control of corruption (CC). The results are shown in Table 7. Panel A, Panel B, and Panel C are, respectively, the results on user-centricity, business mobility, and key enablers. In general, the results show that the positive impact of digitalization on the value of TGG becomes more significant in an economy with a well-developed institutional system. These findings are statistically significant with positive interactions between user-centricity, business mobility, and key enablers and variables representing level of institutional quality. All interactions are statistically significant at the 1% level. In other words, the positive impact of DPS on TGG is more evident in a better institutional environment. In a similar spirit, Zakari and Khan (2021a) argue that institutional quality is the underlying factor for the positive influence of energy consumption on economic growth. Institutional quality is also essential in moderating the marginal effects of economic sanctions on environmental performance (Le et al., 2021) or the impact of economic complexity on energy security (Le et al., 2022).