GI provides organisations with the chance to diminish their operations' adverse effects on the environment and guarantees a competitive advantage (Awan et al., 2020). It can facilitate the development of new manufacturing processes and products that are less injurious to the ecosystem and natural environment (Khan et al., 2021). GI is "the production, application or exploitation of a good, service, process, organisational structure or management or business method that is novel to the firm and results in a reduction of environmental risk" (Ma et al., 2018). The definition of GI has various forms, e.g. green technological innovation, which encompasses product and process innovation, and green non-technological innovation comprising management innovation and organisational structure (Chang & Chen, 2013; Chen et al., 2006; Hilkenmeier et al., 2021). The former aims to assimilate various advanced and novel technologies that can improve the existing process and products to reduce energy consumption, prevent pollution and save natural resources (Fernando et al., 2019; Khan et al., 2021; Xie et al., 2022). It also alludes to process and product innovation. The latter encompasses adopting/restructuring firms' management strategies, i.e. environment, energy, quality management, green supply chain, and green marketing to minimise harmful environmental effects (Klein et al., 2021; Shu et al., 2016).

Chen et al. (2006) describe GI as "hardware or software innovation that is related to green products or processes, including the innovation in technologies that are involved in energy saving, pollution prevention, waste recycling, green product designs, or corporate environmental management." It is positioned as the main driver of long-term socio-economic progress. Several studies acknowledged the key factors that affect GI adoption, e.g. concerned stakeholders' pressure, strategic orientation, organisational learning, knowledge management, absorptive capacity, and consumers' demands (Awan et al., 2020; Dangelico, 2017; Klein et al., 2021; Shahzad et al., 2020a; Song et al., 2020). Further, organisational innovation is a driving force in enhancing industrial export, environmental performance, and, eventually, business excellence and SD (Li et al., 2020; Wu et al., 2019). In brief, GI inclines to improve competitiveness by developing innovative goods, processes, materials, and institutional frameworks.

With increased environmental deterioration and climate change envisaged by rising hazardous emissions and pollution, global sustainable economy is certainly constrained (Khan et al., 2021). Green technologies and monitoring policies are imperative to regulate and encourage GIA (Li et al., 2020). GIA requires innovative organisational strategies to switch their classical and traditional means of production to novel and sustainable operations (Anser et al., 2020). Nevertheless, transformation into sustainable operations remains difficult for organisations because multiple uncertainties and complexities are involved in the transformation procedure (Han & Chen, 2021). Different sectors have accepted and transformed their operations into green operations following SD indicators, e.g. environmental, social, and economic performance (Jahanshahi et al., 2020). GIA in businesses has also attracted experts' and researchers' attention (Forcadell et al., 2021; Han & Chen, 2021; Klein et al., 2021). Recently scholars have identified different barriers and enablers for GIA in manufacturing enterprises (Han & Chen, 2021). From prior literature, the adoption of green technology, or its acceptability, can be recapitulated as the extent of the possibility of an emerging novel technology being authorised by groups or individuals (Awan et al., 2020; Jahanshahi et al., 2020).

Many scholars modelled the critical elements of technology adoption for better decision-making, which is further developed in the UTAUT model, and verified the rationality of their attributes (Venkatesh et al., 2003). To anticipate technology adoption intention and usage of novel and innovative technologies, the UTAUT prolongs the TAM, the theory of reasoned action (TRA), diffusion of innovation theory, and a mirror of cognition theory (Zhao & Bacao, 2020). The UTAUT comprises four fundamental driving factors of intention and usage: performance and effort expectancy, facilitating conditions, and social influence (Venkatesh et al., 2003). Various studies integrated the UTAUT to explore behavioural intention to accept the latest technologies, stimulating its generalizability (Anser et al., 2020; Zhao & Bacao, 2020). The UTAUT framework is the foremost consolidated model, comprehensively describing technology adoption (Al-Saedi et al., 2020). This model was further studied by integrating others factors such as compatibleness expectancy, sustainable innovativeness, and environmentalism, in adopting green household technology (Ahn et al., 2016). Accordingly, knowledge of green products influences users' behaviour to care for the natural environment following the UTAUT model as knowledge influences all phases of the purchasing decision process (Hsu et al., 2017).

Despite these four fundamental variables, Venkatesh et al. (2012) underlined the need to integrate more relevant prognosticator variables to forecast behavioural intentions for the technology adoption perspective by modifying the UTAUT to provide a new predicting model, namely UTAUT2. This latest model has progressively been implemented for investigating multiple queries such as self-service technology, adoption of mobile technology, mobile banking and commerce, online education, and online healthcare services (Huang & Kao, 2015). Hedonic motivation and cost of innovation are considered more important factors of UTAUT2, which are further integrated into the research framework of this study to emphasise efficacy and utility. Moreover, Ma et al. (2017) established that, compared to non-labelled items, the sustainable label reading behaviour of products increases the purchasing of sustainable and green products, while the increasing ecological cognisance among individuals (Chen, 2008) suggests that they are eager to pay a greater value for eco-friendly goods. Since this study's primary aim is to discover the factors that influence GIA, the UTAUT2 model can offer better insights; therefore, it is employed as the research framework, as shown in Fig. 1.

Fig. 1.

Research model.

(0.16MB).

Performance expectancy is a key variable of the UTAUT model, which influences behavioural intentions. It is "the degree to which an individual believes that using the system will help him or her to attain gains in job performance" (Venkatesh et al., 2003). It comprises four fundamental measures that gauge performance: perceived usefulness, job fit, extrinsic motivation, and comparative edge (Huang & Kao, 2015). It is the most significant contributor to identify individual intention to accept new technology and satisfaction (Zhao & Bacao, 2020). Prior studies have certified that performance expectancy positively and considerably affects adoption and continuing usage of the latest technologies, i.e., mobile banking. In the setting of this study, GPE may have a considerable impact on GBI, as various green factors such as supplier selection, procurement, industrial engineering and consumerism all have a considerable impact on green purchase intention (Anser et al., 2020). Recent studies specified that green product knowledge positively affects individual green behaviour (e.g. Hsu et al., 2017). Thus, the subsequent hypothesis is proposed:

• H1: Green performance expectancy positively affects green behavioural intention.

Effort expectancy is one of the dominant constructs of UTAUT, described as "the degree of ease of use associated with the usage of a new technology or a technology product" (Huang & Kao, 2015). It is a comparable construct with ease or complexity of use (Zhao & Bacao, 2020); the latter are identified as the extent to which innovative technology is complex or easy to use and comprehend. The complexity of innovative technology may harm its adoption (Dangelico, 2017). It is expected that the larger the ease of use of innovative technology, the lower the individual behavioural intention (Al-Saedi et al., 2020). Some research studies found that effort expectancy harms using novel technologies, i.e., internet banking and shopping online (Chopdar and Sivakumar, 2019). The latest studies identified that effort expectancy significantly affects innovative technologies' utilisation and satisfaction by employing and validating the UTAUT model (Anser et al., 2020; Shang & Wu, 2017). Further, for our study context, green product labelling enhances the individual green behaviour and intention to utilise sustainable and green products compared to non-labelled products (Ma et al., 2017). Accordingly, the subsequent hypothesis is proposed:

• H2: Green effort expectancy positively affects green behavioural intention.

Hedonic motivation, known as perceived enjoyment, refers to internal pleasure, fun, or satisfaction experienced using the latest innovative technology and articulates a key role in contributing to the UTAUT2 model (Tam et al., 2020). An individual with utilitarian motivation focuses on instrumental values, while one with hedonic motivation focuses on fun and pleasure (Wang et al., 2020). It has been demonstrated to be a more fundamental driver than other UTAUT components and a core estimator of behavioural intention (Venkatesh et al., 2012). Empirical research identified that hedonic motivation affects technology adoption both in individual and organisational contexts (Ashfaq et al., 2021; Huang & Kao, 2015). In the context of GHM, users' hedonic motivation captures a vital role in predicting green buying behaviour (Choi & Johnson, 2019). Prior studies acknowledge that individuals' thinking and green motivation incite their urge to purchase eco-friendly and green products (Ali et al., 2020). Motivation for adopting smart technologies is a pertinent factor that affects individuals' intentions to enhance their households' sustainability and sustainable consumption behaviour (Ahn et al., 2016). Furthermore, individuals' novelty-seeking behaviour and green consumerism also impact green purchase intention (Anser et al., 2020; Choi & Johnson, 2019). Thus, subsequent to the above discussion, we posit the below hypothesis:

• H3: Green hedonic motivation positively affects green behavioural intention.

Social influence means that social networks incline individuals' decisions since they frequently evaluate the ideas and opinions of others when deciding whether or not to espouse innovative technologies (Anser et al., 2020). It is described as "the degree to which an individual perceives that important others believe he or she should use the new system" (Venkatesh et al., 2003). It denotes encompassing the individual decision-making process to accept innovative technology affected by others' opinions (Ashfaq et al., 2021; Dangelico, 2017). Social influence is also considered a subjective norm in TAM and social norms in TRA (Zhao & Bacao, 2020). Recent research identified that social influence significantly influences the adoption of innovative technologies and behavioural intention at all points in time (Wang et al., 2020). In this research setting, prior studies identified that social influence related to environmental conservation is the most influential element in predicting and adopting green technology (Ahn et al., 2016). Moreover, it helps shape individual behaviour towards green intentions and sustainable purchase decisions, e.g. purchasing unique biodegradable packaging and carrying bags (Choi & Johnson, 2019). The greater the social influence of green technology adoption, the greater the individual's persistence in using it. As a result, the following hypothesis is advanced:

• H4: Green social influence positively affects green behavioural intention.

Facilitating conditions are the final and central element of the UTAUT model. Facilitating conditions are "the factors in an environment that hinder or make an activity easier to perform for an individual" (Venkatesh et al., 2003). Individual-level and group-level are the two forms of facilitating conditions. The former is about the individual insight into environmental support; the latter is about organisational support available for groups (Ahn et al., 2016). Without a comprehensive set of facilitating conditions, it is challenging to adopt and use the latest technology. However, it is rational in a green context since the facilitating conditions, e.g. training and guidance about innovative and green technology (software and hardware), persuade usage and GBI (Tariq et al., 2016). The prospective barriers to use can be eliminated or reduced significantly (Venkatesh et al., 2012). Prior literature identified that organisations' employees would accept and adopt new technology when they received support and facilitating assistance (Tam et al., 2020). Nysveen and Pedersen (2016) identified that an individual with accession to a conducive series of facilitating conditions is expected to adapt and accept new technology. Wong (2013) suggested that adoption of green technologies enables organisations to reduce adverse ecological impact, facilitating SD outcomes. So, the following hypothesis is proposed:

• H5: Green facilitating conditions positively affect green behavioural intention.

Innovation cost is another of the most critical variables in the UTAUT2 model, as product cost significantly influences technology adoption (Tam et al., 2020). The price value is conventionally specified as arbitration between cost and benefit analysis. When the advantages of adopting new technology are superior to the financial costs, the innovation cost shows positive results and positively influences adoption intention (Venkatesh et al., 2012). Besides, GI is not free; however, it is lucrative for organisations in the long run (Zailani et al., 2015). Prior research acknowledged that environmental compliance is an extra financial burden and increases production costs instead of considering it an essential strategy to avert harmful ecological effects (Liu et al., 2021). However, the number of environmentally conscious consumers is rising; they prefer to use eco-friendly products (Chang & Chen, 2013). They desire innovative and green products and are determined to pay a greater price for green items (Chen, 2008). Further, Wei et al. (2018) stated that less environmental motivated consumers are likely to pay less for green products. However, highly environmental motivated consumers are likely to pay high. Green processes and product innovation diminish adverse ecological impacts and enhance production efficiency and sustainable financial performance through cost and waste minimisation (Zailani et al., 2015). Prior studies show contradictory arguments regarding this relationship, so re-investigation of this relationship is indispensable. Thus, the subsequent hypothesis is proposed:

• H6: Green innovation cost positively affects green behavioural intention.

Psychologists and social scientists acknowledge that behavioural intentions always strongly affect actual behaviour (Straub, 2009; Zafar et al., 2020). Nevertheless, the prediction of actual behaviour is still challenging. Behavioural intention denotes "the degree to which a person has formulated conscious plans to perform or not perform some specified future behaviour(s)" (Huang & Kao, 2015). The prior researcher, Venkatesh et al. (2012) identified that behavioural intention regarding technology adoption plays a magnificent role in actual technology adoption. Several researchers employ intention behaviour as a surrogate of actual adoption behaviour (Karampournioti and Wiedmann, 2022; Zafar et al., 2020). GI is now growing a competitive strategy due to increasing environmental regulations and optimal sustainability outcomes. Further, GIA is a long-run effort that necessitates an organisation to create substantial developments in processes and products, which inevitably invoke environmental risks (Jahanshahi et al., 2020; K. Lee et al., 2021). Larger organisations are ready to assimilate innovative technologies, capabilities, and external and internal environments; they are more likely to put potential risks beneath control (Albino et al., 2009). Thus, consistent with the underlying theory and research model, we expect that GBI will substantially influence GIA. Hence, we propose the following hypothesis:

• H7: Green behavioural intention positively affects green innovation adoption.

This research seeks to determine whether GBI acts as a mediator among diverse decision-making factors of UTAUT and green innovation adoption. Behavioural intention focuses on the desire to actual usage or adoption; such desires can be dominant and irresistible; still, it does not basically ascertain the actions. Several studies provided the theoretical background and critical role of behaviour intentions for actual technology adoption (Ashfaq et al., 2021; Ifedayo et al., 2021; Venkatesh et al., 2012). Ifedayo et al. (2021) identified behaviour intentions as a prognosticator of podcast technology acceptance in Nigeria directly and indirectly as well. Further, J. Lee et al. (2021) also acknowledged that eco-friendly behavioural intentions significantly influence the adoption decisions regarding electric vehicles. The previous scholars have extensively conferred how green thinking and motivation relate to green behaviour and adoption intention (Ali et al., 2020; Choi & Johnson, 2019). Moreover, Casey and Wilson-Evered (2012) also emphasise the key mediating role of behavioural intention among performance expectancy, effort expectancy, and trust in new technology. Therefore, based on the extant literature, we propose that GBI plays a mediating role among integrated constructs and GIA. Thus, the following hypotheses are proposed.

Green behavioural intention mediates the relation among green performance expectancy (H8a), green effort expectancy (H8b), green hedonic motivation (H8c), green social influence (H8d), green facilitating conditions (H8e), and green innovation cost (H8f) to green innovation adoption.

Generally, the number of employees at any particular geographical location is known as organisational size. Several researchers have identified that organisational characteristics have a higher propensity for the behavioural intention to adopt innovative technologies (Aibar-Guzmán et al., 2022). Following previous studies, this research also considers organisational size as moderating variable (Ma et al., 2018; Shu et al., 2016). Lin and Ho (2008) emphasised that organisational resources, including quality of resources and organisational size, further influence the adoption of new green technology. Further, Lin et al. (2020) highlighted that organisational resources significantly affect green technology adoption. More resourceful and larger organisations have higher chances of adopting and integrating technological changes into their operations, as this is a lengthy process and needs massive investment. Organisations can implement an advanced environmental strategy by adopting green technologies; when the organisation has higher resources and greater size, the adoption capacity of innovative technologies is higher. So, the following hypothesis is proposed:

• H9: Organisational size significantly moderates the aforementioned relations towards green behavioural intention and green innovation adoption in confounding ways.

A questionnaire survey comprised of two portions was adopted to gather data. The first portion is associated with the demographic evidence of respondents and organisations (see Table 1). The second consists of different measures related to targeted variables. The employed instrument is adopted from prior studies with multiple validated and reliable items. All measurements were concluded following the endorsements of a panel included of three professors and professionals to ensure face validity. GPE, GEE, GSI, GFC, and GBI were evaluated by four, four, three, five, and three items, respectively (Venkatesh et al., 2003); GHM, GIC, and GIA were assessed using four, four, and six items, respectively (Venkatesh et al., 2012). All items were answered on a seven-point Likert scale "1=strongly disagree" to "7=strongly agree". Before conducting the formal survey, pilot testing was undertaken to ensure content validity, but few modifications were necessary to certify data validity and reliability.

Analysis of this investigation was based on quantitative data collected via a questionnaire from various manufacturing industries in one of the emerging markets, i.e., Pakistan, between November 2021 and March 2022. The current government establishes stringent environmental regulations to reduce dependence on coal energy and shift manufacturing to renewable energy sources. However, renewable energy resources are also limited. So, it needs to take some corrective measures to promote green initiatives in the current challenging environment. Similarly, it faces various SD issues requiring vigorous green product development and process innovation (Awan et al., 2020). Therefore, Pakistan has been identified as an appropriate context for evaluating our research hypotheses. The questionnaires were distributed online using Google docs and WhatsApp and offline through personal visits including a cover letter illustrating the aim of this research and assuring respondents' data confidentiality. Due to the epidemic, we conveniently contacted upper, middle, and front-level staff members to obtain higher responses from different manufacturing industries, including textiles and clothing, petroleum and chemicals, electronics and IT, food and beverages, metal manufacturing, and leather products.

To enhance the response rate, reminders and follow-ups were sent to concerned respondents. These corporations were listed in the "Pakistan Stock Exchange (PSX)" and registered with the "Securities and Exchange Commission of Pakistan (SECP)." 980 questionnaires were dispersed to 399 manufacturing units in Pakistan; we received 516 functional responses – a response rate of 52%. These respondents signify the organisation as a whole. Usually, in survey studies, scholars have a low response rate due to respondents' busy schedules and non-access to the internet (Hair et al., 2017). Due to the pandemic, many employees were working from home and had easy access to the internet, so we had a higher response rate than usual. Also, a large sample leads to more precise estimation and results (Asiamah et al., 2017). The majority of respondents held supervisory positions, i.e., 45%, responsible for executing organisational strategies and implementing policies; 60% were male, and the majority were aged between 18 and 35. (see Table 1). The current research adopted a 10X rule for sample size as guided by Hair et al. (2017), which is "10 times the largest number of structural paths directed at a particular latent construct in a structural model". The sample size was derived through G*Power software proposed by Prajapati et al. (2010) to ensure the sample's adequacy for the research model. A set of power analyses revealed that our sample is suitable for further investigation.

Common method bias (CMB) variance is specified as "variance that is attributable to the measurement method rather than to the constructs the measures represent" (Cohen, 1988). It is argued to be a main concern in the questionnaire survey. Initially, CMB was estimated using Harman's single factor, where the first factor has a cut-off value of less than 50% (i.e., 31.15%) (Harman, 1976). Besides, a more rigorous method for testing CMB vis full collinearity evaluation was also implemented (Kock, 2015). The resulting variance inflation factor (VIF) values were less than 3.3 (Kock, 2015). These findings imply that CMB is unlikely to cause severe concern.

The construct reliability method ("Cronbach's alpha (CA), rho_A, and composite reliability (CR)") and validity ("discriminant and convergent validity") was used to estimate the measurement model by following Hair et al. (2017). Referring to the results in Table 2, the CA score ranges from 0.741 to 0.841, whereas the figures of rho_A are in the range of 0.742 and 0.842, and the statistics of CR squeeze a range from 0.853 to 0.889. All statistics are greater than the threshold of 0.70; subsequently, the construct reliability is established (Cohen, 1988; Hair et al., 2017). The loading of factors and "Average Variance Extracted (AVE)" were assessed to determine the convergent validity (CV). These statistics were also larger than the threshold of 0.50, as Hair et al. (2017) advised. The resulting statistics authorised the CV of variables.

Furthermore, the discriminant validity (DV) is affirmed using a traditional but vastly familiar approach (Fornell & Larcker, 1981) and a recent and latest approach heterotrait-monotrait (HTMT) ratio (Henseler et al., 2015). In the first approach, the square root of AVE should be larger than the correlation among targeted components. The second HTMT approach acclaims a cut-off value of 0.85 (Sarstedt et al., 2017). The findings in Tables 3 and 4 approve both criteria of DV.

Following the validation of the outer model, the structural model was evaluated in order to test the hypotheses. To determine the relevance of the hypotheses, a bootstrapping approach was used (5000 resample). The findings of the model disclosed a significant and positive effect of GPE (H1: βeta value=0.182; p<0.001), GEE (H2: βeta value=0.138; p<0.002), GHM (H3: βeta value=0.154; p<0.001), GSI (H4: βeta value=0.123; p<0.033), GFC (H5: βeta value=0.212; p<0.001), and GIC (H6: βeta value=0.157; p<0.001) on GBI which support the hypotheses H1 to H6 respectively. Furthermore, hypothesis H7 revealed a significant and positive influence of GBI on GIA (H7: βeta value=0.247; p<0.000). The result of control variables revealed that these were insignificant. The overall outcomes of the hypotheses are provided in Table 5.

The mediating impact of the GBI was evaluated by the series of steps (Nitzl et al., 2016). At first, this study inspected the indirect effect of the GPE, GEE, GHM, GSI, GFC, and GIC to GIA through GBI; and found a significant effect of these variables with βeta values 0.043, 0.033, 0.038, 0.031, 0.50, 0.037, respectively. In the next step, the direct effect of GPE, GEE, GHM, GSI, GFC, and GIC was measured without removing the mediator (GBI). A significant positive outcome of these variables with βeta values 0.182, 0.138, 0.154, 0.123, 0.212, and 0.157 were found respectively. The results are specified in Table 5, which leads to partial mediation. Besides, this study noticed the sign of indirect and direct effects and found positive and in the same direction; therefore, it might be determined that the GBI has complementary partial mediation (Hair et al., 2017). Hence, H8a to H8f is fully supported.

The moderation effects of organisational size were estimated through the multi-group analysis (MGA) technique. MGA assists in estimating the significant difference among various groups in data for an identical model; predominantly when a categorical moderator is involved (Hair et al., 2017). As the organisational size is a categorical moderator, to assess its moderating effect, data were divided into three groups according to the number of employees (Less than 150-small, n=161), (151 to 250-medium, n=182), and (More than 250-Large, n=173). Results of MGA in Table 6 revealed a significant difference in GIA levels observed among these three groups. In the case of smaller organisations, the effect of GEE, GSI, and GIC on GBI was insignificant. For medium-size organisations, the impact of GHM and GSI on GBI was insignificant, whereas for larger organisations, the effect of GSI on GBI was insignificant only, still it is significant at 10% level of significance. These results suggested that the propensity for GIA among these groups has discrepancies (smaller to larger). Smaller organisations have limited resources and portfolios, so they have a low GI adoption level, unlike medium and large-size organisations. Hence, H9 is fully supported.

The model fit was established by a largely adequate method, i.e. "standardised root mean square residual" (SRMR), where the SRMR value should be less than 0.08 (Hair et al., 2017). The outcomes revealed the value of SRMR is 0.065, suggesting our model is quite well. Secondly, we also calculated GOF using the formula (GOF=√(AVE × R2)) (Wetzels et al., 2009). In our model, the GOF is 0.429, demonstrating the model fulfils the large criteria. Besides R2 (coefficient of determinants), F2 (effect size) and Q2 (predictive relevance) were also analysed. The resultant values were in good range and provided in Table 7.

Following prior social scientists (Chavoshi & Hamidi, 2019; Zafar et al., 2021), this study also employed ANN to identify each variable's relative importance and reinforce SEM results. Though the ANN has many types, the present study has employed one of the most common and renowned networks, i.e., the "multilayer perceptron" (MLP) (Zafar et al., 2021), to train the neural networks. ANNs usually include one input, more or one hidden layer, and one output layer, with no single rule for selecting the best values. The value of hidden layers is proportional to the problem's intricacy (Sheela & Deepa, 2013). The importance of predictors was assessed in two steps. First, we provide seven significant covariates as predicting variables (input layer), whereas GIA was applied as an output layer in the neural network. A sigmoid function was utilised to represent the activation function of neurons in both the hidden and output layers. Following prior researchers such as Zafar et al. (2021); Liébana-cabanillas et al. (2017), the ANN model was verified by employing the number of hidden nodes from 1 to 10. To minimise over-fitting, we employed ten-fold cross-validation, with 70% of the data employed to train the network model and 30% to test it.

The neural network prediction accuracy was estimated using the Root Mean Square Error (RMSE). The findings revealed that the average RMSE for GIA was 0.1337 for training data and 0.1323 for testing data. The disparity in produced values is minor, indicating that the model used provides high accuracy (Leong et al., 2018). Outcomes are given in Table 8. A sensitivity analysis was executed to gauge the importance and normalised importance of integrated covariates in the ANN model. The importance of incorporated constructs was computed by averaging their generated values in ten networks for predicting the output. Further, the normalised importance represents the ratio of each input variable to the highest, indicating that GFC was the most important predictor for GIA with a 0.223 importance value, followed by GPE, GHM, GSI, GEE, GIC, and GBI, i.e., 0.215, 0.208, 0.177, 0.162, 0.161, and 0.128 respectively (see Table 9). The graphical representation of the average and relative importance of each construct were shown in Fig. 2. Some minor differences were observed in the ranking of variables, but GFC and GPE ranking is comparable in both analyses. The non-linear and non-compensatory design of ANN models and their higher level of prediction accuracy may explain these differences.

Fig. 2.

Importance of each construct.

(0.12MB).

This study incorporates the UTAUT to advance the conceptual framework for estimating the influence of specified decision-making factors on GIA – a previously relatively unexplored area. The empirical findings confirmed that GPE and GEE positively affect GBI accepting H1 and H2. These results support preceding studies of Ahn et al. (2016); Anser et al. (2020) by emphasising the insinuation of adopting sustainable and innovative technologies by UTAUT. The positive effects of these variables suggested that innovative green technology is easy to implement and enhances long-term performance in the current challenging business environment. Further, GHM and GSI positively impacted GBI, leading to our H3 and H4. Our results coincided with Ali et al. (2020); Wang et al. (2020). Prior researchers recommend that green thinking and social influence shape individuals' pleasure-seeking behaviour to purchase green products, ultimately conserving the environment. Ashfaq et al. (2021) also claim that social influence and hedonic motivation significantly influence intention to use the latest technology.

GFC most significantly affects GBI accepting H5, showing the outcome is congruent with Tariq et al. (2016). The results of their research accentuated that guidance and edification about innovative and green technology induce usage and GBI. Further, as internal stakeholders of the process, employees would only accept innovative technologies when they attained a particular level of technical support and assistance (Shahzad et al., 2020a). The GIC also positively affected GBI, accepting H6. Our results contradict Tam et al. (2020). The probable cause for this deviation is perhaps that consumers are now more environmentally conscious, prefer to use eco-friendly products, and are willing to pay higher values (Liu et al., 2021). Adopting innovative technology does not personify additional costs for consumers; on the contrary, it can offer financial and non-financial benefits.

Furthermore, the acceptance of H7 shows a substantial positive effect of GBI on GIA as predicted by UTAUT and is broadly coherent with the results of Venkatesh et al. (2012). Several studies suggested that behavioural intention can be used as a surrogate for actual technological adoption (Karampournioti & Wiedmann, 2022; Zafar et al., 2020). Thus, this study also predominantly evaluated the mediating effect of GBI as it instigates GIA. Our two-step mediation results show that GBI complementary partially mediates the integrated relationship towards GIA by accepting H8a to H8f. These results have also coincided with Ashfaq et al. (2021) and Ifedayo et al. (2021) in the broader context for technology adoption.

Lastly, this study also conducted MGA to evaluate the moderating role of organisational size among integrated relations towards GBI and GIA. The findings are distinctive and captivating as organisational size moderated structural relationships differently by accepting H9. Not every organisation can accept technological changes in production operations; it is a long process and requires immense investment/financial resources. Larger organisations can take advantage of economies of scale to adopt GI by increasing production levels. In a developing country like Pakistan, organisations do not have a specialised product line; they have diversified product lines. If they happen to own a specialised product line, their GIA levels might be higher. These results provide adequate evidence that GIA is a long-run effort that obliges an organisation to create considerable development in processes and products, inevitably invoking environmental risks. Finally, the ANN's overall findings support the relevance of integrated components. Sensitivity analysis revealed that GFC and GPE have relatively the highest importance towards GIA as predicted by SEM. Thus, we should consider the significance of these variables for achieving GIA outcomes.

This research serves mainstream literature in a variety of areas. First, a technological adoption model based on UTAUT is validated, providing a new correlate to address the scarcity in the prior literature in the field of GI. We believe this is the first research exploring GIA through diverse decision-making factors with green attributes, i.e., GPE, GEE, GHM, GSI, GFC, GIC, GBI in a novel way through SEM and ANN in developing nations, i.e. Pakistan. Second, this study divulged the direct impact of GPE, GEE, GHM, GSI, GFC, and GIC on GBI and, further on, GIA – a novel phenomenon not operationalised in green and sustainable innovation literature previously. Besides, this study also underpinned the key mediating role of GBI and developed its complementary partial mediation, as behavioural intention reinforces actual technology adoption. The proliferation of ICT and digital manufacturing alters manufacturing processes and operations, significantly impacting GIA (Awan et al., 2020). Thus, our results demonstrate that the study of UTAUT for GIA is imperative in the current era of technology-based innovation.

Third, this research measured the moderating role of organisational size that facilitates the adoption levels of GI. The significant moderating results established that larger organisations promptly realised the importance of GI and effectively embraced SD agendas. This research evocatively contributes to the existing literature and offers a vital and comprehensive mechanism to promote green technology innovation. Finally, combining the two methodologies (i.e., PLS-SEM and ANN) yields new intuitions and emphasises the relevance of all independent variables contributing to GIA independently. The findings indicate that ANN appears to be a more capable predictive model, shown by the low RMSEs of all ANN models for testing and training datasets (Leong et al., 2018; Liébana-cabanillas et al., 2017).

This study has several practical contributions which facilitate managers and policymakers. First, the findings emphasised the relevance of diverse decision-making factors based on the UTAUT model to enhance GIA, which educates practitioners and enables organisations to achieve SD goals by promoting GI. Our work acknowledged that GIA is a helpful tool to persuade manufacturing organisations to consider and integrate innovative and cleaner production technologies into their operations to reduce the environmental burden while deciding their strategic initiatives (Awan et al., 2020). GIA stimulates organisations to offer a sustainable production and consumption model to concerned stakeholders.

Second, to reap the benefits of the SD plan 2030, developing economies such as Pakistan must undertake GI to compete with developed economies. By taking the example of China, they have achieved swift economic growth while undergoing severe resource exhaustion and ecological pollution (Zhu et al., 2010). There is growing pressure to invest in green technologies in these countries, and organisations are already burdened by emergency measures to stop environmental impact; one solution is adopting green technologies. For countries like Pakistan who are in developing mode, there is a possibility to learn from the practices of developed countries regarding environmental conservation. The government should promote and work effectively on a green business climate, i.e., "Punjab Green Development Program," to assist organisations in reducing their dependence on fossil fuels and maximising the use of renewable energy (World Bank, 2018). That will increase ecological awareness among industries and enhance economic growth. Besides, PPPs will also be helpful in providing the solution of advanced and green technologies at a lower cost.

Third, organisations should provide favourable working conditions and encourage employees to acquire more advanced knowledge for specialised business operations (including supply chain integration, innovation, and technology transfer) through education and training. Encountering software and hardware difficulties while using these innovative technologies can hinder the adoption of GI. Solving these difficulties is essential and needs top management and governmental support immediately. More investment should be allocated for the skill-building of employees regarding GIA, to help improve operational performance and profitability. Regular assistance can be offered through various means; technical consultants may offer ongoing product/service consultations to all stakeholders, and call centre services can provide prompt solutions to any problems.