Firm-Level Digitalization for Sustainability Performance: Evidence from Ningbo City of China

Abstract

Climate change is a significant and urgent threat, gaining traction in the scientific community around the globe and requiring immediate action across many sectors. In this context, the digital economy could provide a mutually beneficial solution by utilizing innovation and technical breakthroughs to establish a sustainable future that addresses environmental deterioration, promotes economic growth, and encourages energy conservation. Against this background, this study examined the diffusion of innovation modeling-based factors affecting small and medium-sized firms’ (SMFs) adoption of the Internet of Things (IoT) technology and its impact on SMFs’ sustainability performance related to environmental, economic, innovation, and energy conservation perspectives. The key findings revealed that (i) the relative advantage, trialability, and observability drive IoT adoption. However, compatibility and complexity hinder IoT adoption. (ii) When prioritizing the adoption factors, the relative benefit is the strongest driver, and compatibility is the most significant barrier to IoT adoption. (iii) IoT technology adopter SMFs spent less on natural resources and more on renewable energy and environmental monitoring systems than non-adopter firms, boosting their environmental sustainability. (iv) IoT technology adopter firms had greater revenue, profits, and credit access than non-adopters and lower input costs, improving their economic sustainability. (v) IoT adopter firms spent more on innovative products than non-adopter enterprises, demonstrating innovation performance. (vi) Compared to non-adopter firms, IoT technology adopter SMFs had lower utility expenses and spent more on energy-efficient technologies. (vii) To realize the full potential of the IoT for a more sustainable and inventive future, authorities may pursue a variety of policy actions involving the strengthening and implementation of IoT technology standards and regulations, securing the incentivization of financial resources to SMFs, diverting the allocation of resources to research and development avenues, prioritizing the capacity development and environmental awareness, and focusing on IoT infrastructure development.

1. Introduction

Our world is confronted with the difficult challenge of striking a balance between economic development and environmental conservation. The environment, economy, and innovation are the three essential components that are involved in this tightrope walk. As per the Energy and Natural Resources Report 2023, the conservation of energy is at the center, and it has an impact on both the health of the environment and the well-being of the economy [1]. Climate change poses a significant threat, which is a direct result of an imbalanced system. According to the Global Risks Report 2024, four structural forces are expected to materialize and manage worldwide risks, including future paths of global warming and its associated impacts on Earth systems (i.e., climate change); temporal variations in population number, growth, and composition worldwide (demographic bifurcation); the developmental trajectories of frontier technologies (technological advancement); and the process of material progression in the concentration and origins of geopolitical power (geostrategic transition) [2]. These arguments emphasize that environmental concerns are among the crucial global risks to be monitored in the coming decades. Throughout history, economic expansion has occurred at the price of the environment. The pollution of air and water by industries led to an increase in the usage of resources [3]. The tide, however, is beginning to change. The general public is becoming more aware of environmental concerns, and the economic penalties that are associated with environmental deterioration over the long term are becoming more apparent. In the process of bridging the gap between a strong economy and a healthy environment, innovation is emerging as the bridge [4]. Innovative technological developments in the fields of renewable energy, sustainable materials, and energy efficiency provide potential solutions that have the potential to divorce economic expansion from environmental degradation [5]. Energy conservation, a fundamental principle underlying these advancements, plays an essential part. By decreasing our dependency on fossil fuels, we not only reduce our impact on the environment but also open up chances for developing new environmentally responsible sectors [6].

It is important to note that the way forward to low-carbon development is full of challenges. Significant investments are required to transition to a low-carbon economy, and to navigate this transformation successfully, a detailed assessment of the economic and social implications demands careful attention. In addition, the advantages that come with innovation could not necessarily be dispersed equitably, which might exacerbate the inequality that already exists [7]. The possibility of a situation in which both parties benefit exists, notwithstanding the difficulties. We can build a future in which a robust economy and a healthy planet go hand in hand by encouraging innovation in energy conservation and environmental preservation [8]. When it comes to preventing climate change, which is the ultimate result of an uneven system, this strategy is absolutely essential [9]. Due to the fact that climate change presents a significant risk to human civilizations, as well as ecosystems, it is of the utmost importance to discover sustainable solutions that may effectively handle this worldwide problem.

The digital economy, which emerged from the proliferation of information and communication technologies (ICTs) in the mid-1990s, has experienced a significant and noteworthy evolution [10]. Originally propelled by the internet and mobile communication, it centered on e-commerce and fundamental online services. However, the narrative did not conclude at that point. This initial phase laid the foundation for a more complex and detailed digital environment [11]. The advent of social media, cloud computing, and big data analytics has fundamentally transformed the way businesses function and how consumers engage with one another [12]. The growth of e-commerce led to the emergence of dominant online companies and the disruption of conventional physical retail outlets. Emerging sectors such as ride-sharing and on-demand services have thrived due to the convenience and effectiveness of digital platforms [13]. Currently, the digital economy is an indisputable power intricately woven into the structure of contemporary economies. It serves as a potent catalyst for economic expansion, promoting creativity and efficiency across various industries. Digital technologies are revolutionizing the production of things, delivery of services, and dissemination of knowledge across several sectors, such as manufacturing, agriculture, education, and healthcare [14]. The digital revolution is transforming both our work and consumption patterns while also influencing the trajectory of economic development.

The digital economy has the potential to significantly contribute to the attainment of the United Nations’ Sustainable Development Goals (SDGs). For instance, smart grids, facilitated by digital technology, have the capability to include renewable energy sources and improve the distribution of electricity. This leads to a decrease in dependence on fossil fuels and the promotion of clean energy alternatives. Also, IoT sensors can monitor and track energy use in both buildings and companies. This allows for specific interventions to be implemented and encourages the promotion of energy efficiency [15], which is in line with SDG-7. Next, utilizing digital technologies such as 3D printing and sophisticated robots may optimize manufacturing processes, save waste, and establish more environmentally friendly production systems [16]. In addition, digital platforms can establish more efficient connections between enterprises, suppliers, and consumers, therefore promoting innovation and ethical industrial practices that contribute to SDG-9. After that, the digital economy has the potential to encourage and support more environmentally friendly consumer behaviors. E-commerce platforms can provide environmentally friendly items and transparent supply chains [17]. Further, digital technologies may enable customers to make well-informed decisions on their purchases. In addition, data analytics helps pinpoint places where waste can be reduced and resources can be optimized, hence promoting responsible manufacturing methods in line with SDG-12. In the end, the digital economy is essential in reducing the impact of climate change. IoT sensors may enhance agricultural practices by optimizing water utilization and fertilizer application, hence lowering greenhouse gas emissions. In addition, digital communication technologies may enable remote cooperation and decrease corporate travel [18], thus helping to cut carbon emissions in line with SDG-13.

The digital transformation process is changing how businesses function, resulting in a series of beneficial effects on a firm’s environmental, economic, and creative performance. In this context, digital instruments continuously monitor industrial processes, enabling the detection and minimization of waste [19]. Smart grids and sensors enhance energy efficiency by optimizing energy use inside buildings. In addition, digital platforms enable communication and cooperation, promoting the development of environmentally friendly ideas such as creating green products and establishing sustainable supply chains [20]. Furthermore, this change optimizes processes, reducing resource consumption and lowering manufacturing expenses. Data analysis enables organizations to optimize pricing strategies and efficiently focus marketing efforts, resulting in enhanced income [21]. Also, digital technologies have the potential to enhance communication and cooperation inside the organization, resulting in expedited decision-making and enhanced overall efficiency [22].

The process of digital transformation is now impacting both global and Chinese SMFs [23,24,25]. Studies on digital transformation in small and medium-sized firms (SMFs) analyze the interactions between customers and these firms. To this end, SMFs use digital technology to create novel digital goods and services, broaden their customer base, and enhance their company performance [26]. Rapid networks provide the connection between SMFs and business owners, as well as vendors and clients. This connection allows them to obtain immediate facts and promptly adapt to evolving markets and supply networks. Furthermore, SMFs frequently employ virtual specialists and artificial intelligence (AI) to fulfill consumers’ requirements [27]. The digitalization of managerial duties also contributes to the continuous improvement of operations work, resulting in increased productivity and reduced input expenditures [28]. SMFs undergo a digital transformation that enhances company outcomes and boosts personnel productivity and production. Furthermore, such transformation enables the adoption of novel methods for financial administration and transactions [29]. Digital finance facilitates the process of providing financial support to SMFs, hence enhancing their access to funding [30]. This, in turn, contributes to promoting inclusive finance [31], which can potentially elevate corporate environmental sustainability [32].

Business operations are being revolutionized by digital technologies like the Internet of Things (IoT) technology, which is causing a domino effect of beneficial outcomes to spread throughout the business world [33]. The incorporation of sensors and internet connectivity into physical assets enables businesses to acquire real-time data regarding the utilization of resources and processes. As a result, environmental benefits such as less pollution can be achieved through the optimization of production and waste management [34]. From a financial standpoint, businesses are able to recognize and eradicate inefficiencies, thus reducing the costs of materials and energy. The continuous flow of data also serves as a source of innovation, as businesses are able to analyze patterns of usage in order to develop new products and services [35]. Last but not least, IoT devices have the potential to directly increase energy conservation by automating controls for lighting, heating, and machines, which can result in considerable savings in overall usage [36]. In a nutshell, implementing the IoT generates a domino effect of environmental responsibility, economic benefit, inventive thinking, and the utilization of sustainable energy.

Prior studies investigating IoT adoption and its role in the corporate sector demonstrated significant progress. Those studies can be classified under the following research streams: determinants of IoT adoption and its impact on corporate performance. The first research stream included works examining the determinants of IoT adoption at the firm and individual levels. For instance, Kronlid et al. [37] applied a sociotechnical systems framework to analyze 24 distinguished sociotechnical factors influencing IoT adoption across the healthcare sector. The authors employed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) tool during March and April 2022 and revealed that more practical research was needed on IoT adoption in the healthcare industry. Rey et al. [38] conducted a questionnaire based on 21 close-ended questions in Campania (Italy) to study the factors influencing IoT adoption at the firm and individual levels in the transportation and logistics industrial sectors. The authors unfolded that absorptive capacity, firm size, and firms’ perception of the IoT technological advancement upsurged IoT adoption. In their study, Jaspers and Pearson [39] applied structural equation modeling (SEM) to primary data gathered from 930 respondents in New Zealand to inspect the factors affecting IoT acceptance by the residents. The authors found that trust and perceived usefulness positively influenced IoT acceptance, while privacy concerns did not play a significant role in the adoption phenomenon. Ali et al. [40] examined the driving forces of IoT adoption in supply chain management by using the decision-making trial and evaluation laboratory (DEMATEL) on data collected from 231 respondents from Canada, Bangladesh, Australia, and the USA. Their findings showed that information protection, efficient logistics structures, and business knowledge penetration proved to be the most influential driving forces of IoT adoption.

Moreover, utilizing the SEM approach to data compiled from 508 IoT users through IoT-based bulletin board systems and Facebook, Hsu and Lin [41] analyzed the factors affecting the continued utilization of IoT services. They uncovered that perceptional benefits had a substantial positive contribution to IoT services, while privacy concerns had a mild influence on those services. A study by Shaqrah and Almars [42] explored the adoption factors of IoT in the education industry by applying the partial least square-based structural equation modeling (PLSEQ) technique to data collected from 400 students of Taibah University (Saudi Arabia). The authors unveiled that public support, facilitating conditions, effort expectancy, and innovativeness had substantial contributions, while perceived usefulness and expected performance had the least contributions to IoT adoption in the education industry. Langer et al. [43] studied the questionnaire-based data from 212 dairy farmers from Germany to evaluate the influence factors of IoT sensors used in dairy farming. The authors found that high capital costs impeded the use of IoT sensors, while attitude and behavioral control promoted the same. In addition to this, employing PLSEQ to data compiled from 228 Chinese adopters of blockchain technology, Esfahbodi et al. [44] assessed the determining factors of blockchain technology in the e-commerce industry. They empirically argued that traceability and cost-saving features had important contributions to the perceived usefulness of blockchain technology; however, perceived ease of use had no impact on the intention formation of users in blockchain technology adoption.

The second research stream focused on the role of IoT adoption in determining firm performance in heterogeneous contexts. For example, Qin [45] made use of PLSEQ on data from 860 Chinese respondents gathered through emails and WeChat platforms to inquire about the impact of IoT and AI acceptance on frugal innovation. The author disclosed that both the IoT and AI positively drove the said innovation. Ge et al. [46] used data from China’s A-share listed companies to inspect the contribution of digitalized innovation networks on firms’ innovation performance by employing an ordinary least squares (OLS) approach. The authors established that embedding digitalized innovation in firms encouraged them to adopt social cohesion and implicit knowledge to enhance innovation performance. A different study by Attaran et al. [47] investigated the interaction between industrial IoT and physical production. They found that IoT utilization improved the economic and environmental performance of industries. Tan et al. [48] used the secondary data of 432 firms, including 168 adopters and 264 non-adopter firms, for the year 2015 to analyze the impact of IoT usage on the financial performance of the firms. Employing the OLS methodology, they disclosed that IoT adoption improved corporate financial performance. Applying a business model development approach, Nalajala et al. [49] evaluated the role of IoT in virtual manufacturing processes and real-time data collection for seamless supply chains to improve the production activities of the companies. The authors spotlighted that data analytics driven by IoT implementation facilitated productivity levels even through basic usage of the IoT. In their work, Masoomi et al. [50] surveyed 175 firms belonging to renewable energy supply chains in the Iranian context to investigate the sustainability performance of those firms in response to IoT adoption. Utilizing PLSEQ, they unmasked that IoT adoption was able to capitalize on the environmental and economic sustainability of the firms under analysis.

Furthermore, Mishra et al. [51] employed the Fuzzy Vikor approach to data compiled from a panel of 10 experts with applied side knowledge of AI, IoT, and blockchain technology to evaluate the influence of their implementation on decarbonization. They evidenced that the said technological solutions were helpful in decarbonizing the corporate sector. Conducting a systematic literature review and applying several statistical testing approaches, Musarat et al. [52] assessed the advantages, challenges, and uses of the IoT in the Malaysian construction industry. The authors asserted that the IoT exacerbated the safety, productivity, and product quality of the industry by posing security and data privacy risks. Ding et al. [53] analyzed the environmental performance of the logistics sector in China from 2011 through 2018 by applying Grey correlational analysis. They shed light on the evidence that increased IoT adoption was correlated with reduced levels of particulate matter and carbon emissions. Exploring the role of cellular IoT in zero energy systems, Abbas et al. [54] examined and found that upcoming 6G technology devices would not rely on batteries and charging sources, retrieving the minimal level of energy consumption by those devices. Guan [55] empirically examined the role of IoT in low-carbon urban development structures by recording atmospheric pollution levels. The author uncovered that atmospheric pollution was lower in low-carbon cities with the IoT in place than those without IoT applications. Finally, a study by Hu [56] evaluated the impact of IoT applications in smart grids to identify their energy conservation capabilities within a load–pressure framework. The author disclosed that IoT intervention significantly impacts the energy conservation features of thermal energy management systems.

The third research stream explored the possible risks associated with implementing the digital transformation process. For instance, Sun et al. [57] utilized data from China’s A-listed firms over the 2011–2021 period to investigate the combination of digital and real economies and their influence on the risks of collapsing stock prices. They revealed that, on the one hand, digital economy integration lowered the risk of stock price collapse. On the flip side, it reduced the efficiency of capital market pricing, which upsurged the said risk factor. Employing data compiled from Chinese cities and firms from 2011 through 2020, Zhao and Weng [58] studied the effect of the digital economy on firm-level activities and the resulting gaps among the cities. For data analysis, they utilized the OLS estimator to find that the digital economy boosted entrepreneurial activities while creating an inter-city gap in terms of those activities, supporting the “Mathew Effect”. Using the multiple indicators multiple causes (MIMIC) model on data from 30 Chinese provinces from 1995 through 2020, Lv et al. [59] analyzed the influence of digitalization on the informal economy. The authors evidenced that, in the initial phase, the digital economy reduces the informal economy, but it promotes the same in the advanced stages of development. However, Riaz et al. [60] delved into the role of IoT adoption in influencing the security risk concerns in the agricultural farming sector. Using a quantitative assessment framework, they evidenced and argued that the risk environment can be monitored in real time through adaptive security frameworks to improve the efficiency of agricultural activities.

The current body of knowledge on digital transformation, particularly in terms of the adoption of the Internet of Things (IoT) technology, is focused on examining the determinants of IoT utilization within a diverse range of theoretical frameworks [37,38,39,40,41,42]. Nevertheless, none of them opted for Rogers’ diffusion of innovation modeling (DIM) framework [61] to investigate those determinants of IoT adoption. Moreover, there is a limited amount of empirical evidence that establishes a connection between digital transformation and the sustainability performance of firms using several metrics [8,21,62,63]. In this regard, there is a lack of research on the influence of IoT implementation on the sustainability performance of small and medium-sized firms (SMFs) captured through their spending on environmentally friendly, energy-efficient, and innovative activities. Some research works have examined the effects of digital transformation on firm revenue and profitability [22,64,65]. However, none of these studies have specifically investigated the influence of IoT technology on firms’ economic sustainability behavior. In particular, no previous study has been found to capture such an impact through the metric of SMFs’ access to credits.

The objective of the current investigation is to thoroughly examine the factors influencing the adoption of IoT technology and the impact of such adoption on the sustainability performance of SMFs. This research provides distinct and novel contributions to the scientific literature. In this regard, the current investigation is the initial attempt to thoroughly examine the SMFs’ environmental, economic, and innovation performance, as well as the energy conservation linked to the use of IoT technology. Furthermore, in the theoretical frontiers, the current research considers the DIM framework-based factors influencing IoT technology adoption, which is a novel approach in the existing literature. Also, the present inquiry considered the collective incorporation of firm-level factors such as expenditures in environmental monitoring systems, energy-efficient products, and innovative products, which were not previously considered in the prior research. The present research utilizes the PLSEQ to empirically assess the parameters that influence the adoption of IoT technology. Additionally, this investigation utilizes propensity score matching (PSM) to examine the sustainability performance of SMFs in relation to their decisions to adopt IoT technology. The analysis relies on data obtained from an extensive survey conducted in four districts of Ningbo City, located in the Zhejiang Province of China.

The rationale for choosing PLSEQ and PSM can be enumerated as follows: Fundamentally, this research addresses two types of questions: First, do the DIM-based factors affect IoT adoption by SMFs? Second, how does IoT adoption by SMFs influence the firm performance? For the former type of question, we employ PLSEQ, because we have latent constructs (or variables) to examine. The latent constructs are unobserved variables that need multiple observed indicators to be measured effectively. Other alternative techniques, such as logistic regression and probit model, are suitable for binary-dependent variables and continuous or binary-independent variables; however, those techniques are not meant for latent variables. Therefore, in the current case, PLSEQ is the optimal choice to handle latent variables. Several prior studies can be referred to that used PLSEQ to determine the factors influencing the adoption of different technologies involving renewable energy technology [66], AI chatbots [67], geographic information systems [68], dairy farming IoT sensors [43], and building information modeling [69], among others. For the latter type of question, PSM was chosen because it can prevent selection bias while evaluating the difference in the firm performance between IoT adopters and non-adopter firms. To illustrate, IoT adopter and non-adopter firms may have pre-existing differences that could affect the firm’s performance. If those differences are left uncontrolled, they would lead to selection bias. PSM can control those pre-existing differences by matching the IoT adopter (treatment category) firms with similar non-adopter (controlled category) firms. Thus, by generating a set of comparable IoT non-adopter firms, PSM performs a randomized experiment in which differences in firm performance can be reliably attributed to the IoT adoption phenomenon. Existing studies applied the PSM approach in various contexts, including the impact of biogas adoption on household welfare [70], the influence of hybrid rice variety adoption on the technical efficiency of farming [71], and the effect of soil and water conservation practices on food consumption at the household level [72]. These arguments make PLSEQ and PSM the best-suited choice for the current investigation.

Ningbo was selected as the investigation laboratory for the following reasons: Typically recognized for its robust manufacturing sector, the city has adopted digital technologies to enhance the capabilities of its plants. Industrial robots, the IoT, and big data are employed to enhance production lines, establish intelligent factories, and enhance efficiency. Integrating traditional manufacturing with advanced digital solutions is a crucial element of China’s comprehensive strategy for digital transformation. Furthermore, Ningbo’s endeavors have received recognition. The Chinese government has acknowledged the city for its accomplishments. It has completed more than 10,000 digital transformation projects and gained national recognition for its 5G-powered factories. As a result, it has become a role model for other cities seeking to enhance their businesses through digital upgrades. Ultimately, although manufacturing continues to be obligatory, Ningbo is actively promoting the development of a strong services industry. By 2025, the city’s objective is to establish itself as a prominent national center for software development and a frontrunner in the field of the industrial internet. The city’s emphasis on cultivating a varied and technology-oriented economy establishes it as a leader in China’s transition to a more knowledge-driven economic model.

Regarding the scope and significance, while the findings of the current study stem from a survey carried out in a specific geographic area (i.e., Ningbo City of China), the implications of these results reveal the DIM framework-based factors influencing SMFs’ decisions to adopt the IoT. SMFs worldwide might face similar prospects and challenges while adopting IoT technology; therefore, the findings and insights obtained on DIM factors from the case of Chinese SMFs would be interesting to the international industries and global community while integrating IoT into their industrial systems. Furthermore, investigating the role of IoT adoption on the sustainability performance of SMFs, this study covers global concerns such as economy, energy, environment, and innovation, which will aid in evolving broad-spectrum discussions on overcoming the climate change mitigation challenges.

2. Theoretical Foundation and Hypotheses Formulation

The widespread dissemination and use of digital technologies are potentially as crucial as their creation in stimulating economic development. The impact of new technology on economic development can only be fully realized when the new technology is broadly disseminated and used [73]. The spread of technology depends on a sequence of choices made by both individuals and businesses, which include weighing the unpredictable advantages of the technology against the uncertain expenses of adopting it [74]. Knowledgeable and proficient people, as well as innovative firms generally situated in affluent economies, have consistently praised the beneficial influence of digital technology and gained substantial benefits in the face of their technical expertise [75]. Nevertheless, a significant portion of the global population is still in the early stages of embracing digital technologies.

Transitioning to digital technologies is not simply about individual businesses on a national scale; it is a driver of economic development, environmental advancement, and creativity across whole economies. There is a correlation between firms that undergo digital transformation and higher efficiency, which, in turn, leads to increased economic production and productivity [76]. Additionally, digital technologies may be used to monitor and control environmental effects, hence supporting sustainable practices and the conservation of resources [77]. Another advantage of digital transformation is that it encourages a culture of creativity. This is because data-driven insights and collaborative platforms are the driving forces behind creating new ideas and technologies that benefit the whole economy [78]. To put it simply, digital transformation serves as a driving force behind a prosperous, environmentally conscious, and technologically advanced future.

Moreover, digital technologies enable organizations to analyze extensive quantities of data, resulting in discovering fresh prospects and creating ingenious solutions [79]. Cloud computing and big data analytics provide rapid and efficient experimentation and concept testing, hence expediting the innovation cycle [80]. In addition, digital technologies may facilitate cooperation among both internal and external stakeholders, enabling the integration of many viewpoints to enhance innovative problem-solving. These technologies are essential for maximizing energy efficiency. Intelligent sensors and building management systems have the capability to automatically modify lighting, heating, and cooling according to real-time requirements [81]. In addition, digital technologies may enhance the incorporation of renewable energy sources and encourage remote working methods, significantly diminishing a firm’s carbon footprint [82]. Ultimately, digital transformation offers firms a compelling chance to attain a mutually beneficial outcome. By adopting digital technologies and implementing effective methods, companies may improve their environmental performance, stimulate economic development, and expand their ability to innovate, all while saving energy and promoting a more sustainable future [83].

2.1. DIM-Based Factors and IoT Adoption

The adoption of innovative technologies has been a hotly debated concern of several well-known theories for the last several decades. In this regard, the technology acceptance model (TAM) by Davis [84] explains the mechanisms through which technological usefulness and ease of use shape an individual’s adoption decisions. TAM has been used by several studies from various perspectives, including but not limited to online buying intention [85], smart city services acceptance [86], and digital banking adoption behavior [87]. Moreover, the Unified Theory of Acceptance and Use of Technology (UTAUT) theory by Venkatesh et al. [88] covers the technological adoption factors such as performance expectancy, effort expectancy, social influence, and facilitating conditions. The prior studies have applied UTAUT in different scenarios, like public acceptance of autonomous modular transit [89], farmers’ intention to purchase alternative fuel tractors [90], and electric vehicle adoption behavior [91]. In addition, the theory of planned behavior (TPB) by Ajzen [92] deals with the psychological disposition of humans in intention formation for decision-making. TPB particularly emphasizes the constructs of subjective norms, attitudinal constructs, and perceived behavioral control. A wide range of studies have employed TPB to understand the adoption behaviors of individuals involving farmer’s uptake of low-carbon agricultural technologies [93], digital technology adoption in rainbow trout aquaculture [94], and acceptance of solar rooftop technology in emerging markets [95]. Despite reflective theoretical developments of the above-mentioned theories, they largely lacked the systematic stages of technological acceptance behavior.

As opposed to the theories mentioned earlier, the diffusion of innovation modeling (DIM) framework by Rogers [61] posits that consumers have to move via multiple phases while deciding whether to embrace or refuse an innovation. DIM aids in comprehending the process of making choices to embrace an innovative item like the IoT, which is particularly beneficial for new customers who may have numerous uncertainties and concerns prior to purchasing a novel technology [96]. These features were not embedded in the construction of previously implemented theories like TAM, UTAUT, and TPB. The DIM elucidates the process by which either the approval or denial choices regarding a novel good are influenced by the acceptance of innovations. It suggests that customers get interested in the procedure of creative thinking only after they begin to acquire knowledge regarding emerging technologies [97]. The relevant data act as a foundation for determining whether to embrace or refuse the novel item. Ref. [61] suggests that certain innate features, including personal habits, societal attributes, and interpersonal protocol, can influence how decisions are made. Within this paradigm, the acceptance of IoT is primarily facilitated via the processes that involve information acquisition, persuasiveness, decision-making, application, and approval, as illustrated in Figure 1.

The information acquisition channel involves the exposure of individuals or entities (firms, organizations, etc.) to the existence of an innovation without having an in-depth knowledge of how it works. Such a channel uses mass media to spread awareness about innovation to audiences on a broad spectrum. Once the basic knowledge is obtained, individuals propagate to the persuasiveness channel through which they can conceptualize opinions in favor or against the innovation. Using this channel, individuals carry out discussions with peer groups (colleagues, friends, relatives, family members, and leaders) to obtain a deeper understanding of the benefits and drawbacks of the innovation. After grasping a sufficient level of understanding of the innovation, individuals and other entities decide to accept or reject the innovation based on its compatibility with their requirements. In this channel, trials of innovation, personal experiences, and pilot and demonstration programs are effective in shaping acceptance or rejection decisions. As the adopters reach the application stage and start using the innovation, they might face certain technical challenges that could be resolved by providing technical support from experts, as well as innovation providers. Finally, innovation adopters revisit their adoption decision by seeking whether the innovation is suitable in their specific context. Herein, their satisfaction will confirm their continuous use of the innovation, while their dissatisfaction might lead to dis-adoption behavior.

The current research focuses on the persuasiveness component of the DIM framework, which refers to the stage where prospective users of the novel technology (specifically, IoT users) are introduced to the concept of adopting the novel item. Throughout this stage, prospective users actively seek data that will help them choose to accept such technology. Persuasiveness is derived from five key features: (a) Relative advantage, (b) Compatibility, (c) Complexity, (d) Trialability, and (e) Observability. These features play a crucial role in influencing the ultimate choice of whether to embrace or dismiss the IoT. In the following paragraphs, an expanded description of each feature is provided in relation to the widespread acceptance of the IoT.

• The relative advantage of the IoT refers to the level at which the IoT is considered superior to traditional manual options. The progression of the use of the IoT depends on the level of relative advantage it offers. To illustrate, by utilizing real-time data from connected devices, firms are able to make data-driven decisions that increase their decision-making process, optimize operational efficiency, reduce waste, and enhance the quality of their products. This results in a notable benefit in areas such as operational effectiveness and customer contentment [98]. Regarding distinction through innovation, IoT technology has the potential to generate novel opportunities for the creation of products. Firms can develop inventive, data-centric goods and services that meet changing client needs and address environmental issues [99]. Emphasizing sustainability and efficiency can serve as a crucial factor that sets us apart from competitors in the marketplace. The IoT facilitates enhanced operational visibility by providing a comprehensive perspective of operations, allowing companies to discover areas of congestion, anticipate equipment malfunctions, and optimize the allocation of resources. The increased visibility leads to a competitive edge by enhancing responsiveness, accelerating turnaround times, and lowering production expenses [100]. Given these arguments, the following relationship can be hypothesized:

H₁. 

Relative advantage is expected to have a favorable influence on the adoption of IoT technology.

• The compatibility of the IoT refers to its alignment with present norms, prior knowledge, and the preferences and requirements of users. In the scenario showing that the IoT is not aligned with societal expectations and principles, its pace of uptake will likely be slow. To illustrate, businesses that are contemplating adoption may face a substantial obstacle in the shape of a lack of compatibility across the various IoT devices and platforms [101]. Let us imagine a situation in which different manufacturers of machines utilize communication protocols or data formats that are incompatible with one another. This results in difficulties when attempting to include them into a unified IoT system, which necessitates extra investments in adapters, gateways, or even whole system refurbishments [102]. It is possible for businesses, particularly smaller ones, to be dissuaded from entering the realm of IoT technology due to the intricacy and considerable upfront expense involved [103]. In addition, the ongoing development of IoT standards might give rise to specific worries over the futureproofing of any system that is selected. This can cause firms to be cautious about committing to a technology that may become outdated in a few years [104]. Along these lines, the following association can be formulated:

H₂. 

Compatibility is expected to have a detrimental effect on the adoption of IoT technology.

• The complexity of IoT alludes to the level of individuals’ comprehension of its application, as certain technologies are designed to be easily operated by users. However, some technological products are challenging for the users to comprehend. In this respect, the diversity of devices and protocols in the IoT might lead to issues with integration due to their wide range. Integrating sensors, machinery, and software from many manufacturers may be a complicated and costly process, requiring specialist knowledge and sometimes impeding smooth data interchange [105]. The substantial volume of data created by IoT devices gives rise to significant issues about data security and privacy. Corporations must have robust cybersecurity protocols to safeguard sensitive data from vulnerabilities or illegal entry, necessitating investments in security solutions and specialized knowledge [106]. In addition, when it comes to interfacing with existing systems, the effective integration of IoT with the current ICT infrastructure might provide a hurdle. Firms may be required to enhance outdated systems or allocate resources towards new platforms in order to effectively handle and analyze the large amount of data generated by IoT devices. This will contribute to the total expenses and intricacy of the deployment process [107]. Based on these theoretical points, the following connection can be established:

H₃. 

Complexity has the potential to have a detrimental effect on the adoption of IoT technology.

• The trialability of the IoT describes the level at which customers can engage with the IoT by experimenting with it via different endeavors. Under such circumstances, consumers are inclined to embrace emerging innovations like the IoT faster if they have previously worked with trial usage before choosing to embrace it. To explain further, contrary to extensive ICT initiatives that need significant initial expenses and lengthy implementation schedules, IoT solutions may often be deployed in a modular manner [81]. Companies may initiate small-scale trial initiatives in certain areas, such as monitoring the energy use of a particular manufacturing line or a building. Firms develop faith in the potential of the technology by directly witnessing its advantages, such as less waste or enhanced efficiency, in a controlled setting [75]. By reducing the perceived risk and expenditure involved in a complete implementation, this makes the adoption of the IoT more appealing to enterprises that are hesitant to explore unfamiliar technology domains [27]. The ability to quickly test and experiment with IoT applications encourages firms to adopt a “test-and-learn” strategy. This method helps companies find the most effective uses of the IoT and customize solutions to meet their individual requirements. As a result, the adoption of IoT technology is accelerated on a larger scale [108]. Against these arguments, the following association can be hypothesized:

H₄. 

Trialability is expected to have a favorable influence on the adoption of IoT technology.

• The observability in the context of the IoT pertains to the assessment and outlook on the IoT based on the feedback and experiences shared by the general people who have used the IoT. Collaborative conversation may also catalyze the uptake of novel technologies like the IoT. One major obstacle in using IoT technology is the apprehension of handling an intricate network of devices and the immense volume of data they produce. The observability component of the IoT is of utmost importance in this context, since it enables firms to gain real-time insights about the condition and functioning of their interconnected equipment [8]. By using centralized monitoring dashboards, anomaly detection, and data visualization tools, observability enables companies to proactively discover and resolve problems, optimize maintenance procedures, and ensure the efficient functioning of their IoT devices [15]. Increased visibility in the technology instils confidence and decreases the perceived difficulty of maintaining an IoT network, eventually promoting wider acceptance and unleashing the whole capabilities of the IoT for greater performance [109]. In the light of these arguments, the following hypothesized link can be formulated:

H₅. 

Observability is expected to have a favorable influence on the adoption of IoT technology.

2.2. IoT Adoption and Firm-Level Sustainability Performance

The adoption of IoT technology offers organizations a ground-breaking chance to enhance their performance in terms of the environmental, economic, and innovation aspects, as well as energy conservation. This can be accomplished through several theoretical pathways, affecting businesses at multiple tiers (see Figure 2). Here is a snap of several vital components:

2.2.1. Environmental Sustainability Performance

By using IoT sensors, energy usage can be monitored in several areas of business activities, including production lines, lights, and HVAC systems, with a focus on optimizing resource utilization [110]. The utilization of real-time data enables precise modifications, hence minimizing energy inefficiencies and decreasing the total environmental impact. IoT sensors utilized in manufacturing or waste management procedures can promptly identify pollutants and leaks, thus facilitating real-time pollution monitoring and mitigation [111]. This allows for immediate action and implementation of methods to mitigate the problem, resulting in more environmentally friendly production and decreased pollution [112]. From a circular economy standpoint, the IoT enables the monitoring and control of resources at every stage of their lifespan. By adopting circular economy strategies like product reuse and recycling, businesses can effectively reduce their environmental footprint and perhaps generate additional sources of income [113]. Given these theoretical arguments, the following hypothesized relationship is drawn:

H₆. 

IoT adoption is expected to have a favorable influence on SMFs’ environmental sustainability performance.

2.2.2. Economic and Innovation Performance

The implementation of IoT technology results in cost savings on energy bills, raw materials, and maintenance by optimizing resource utilization and waste elimination [114]. The funds saved can be redirected towards research and development, stimulating innovation in environmentally sustainable products and procedures [115]. Regarding enhanced productivity, the utilization of real-time data obtained from IoT sensors can be employed for prognostic maintenance, thereby averting equipment failures and interruptions in production. This results in enhanced efficiency, elevated productivity, and, ultimately, augmented profitability [116]. Concerning the new business models, the data gathered via IoT technology presents prospects for inventive business models [117]. Companies have the option to provide services where customers pay based on their usage of the product or develop solutions that use data to improve the efficiency of resource usage and provide other sources of income [118]. Based on the arguments discussed above, the following hypotheses can be developed:

H₇. 

IoT adoption is expected to have an advantageous influence on SMFs’ economic sustainability performance.

H₈. 

IoT adoption is expected to have a beneficial influence on SMFs’ innovation performance.

2.2.3. Energy Conservation

IoT technology facilitates the incorporation of renewable energy sources into smart grids, resulting in a more equitable and effective energy distribution system. This can aid in diminishing the dependence on fossil fuels and advocating for sustainable energy policies [119]. IoT technology enables the implementation of demand response systems by providing detailed information on energy consumption trends, therefore allowing for effective control of the demand response [120]. Businesses have the ability to modify their energy consumption in response to the current grid conditions, which leads to a decrease in high-demand periods and a reduction in energy expenses [83]. Regarding behavioral change, IoT-powered dashboards can visibly display a firm’s energy use data, increasing staff knowledge and encouraging energy-saving behaviors [25]. The adoption of energy efficiency practices can result in enduring environmental and economic advantages. These arguments allow the formulation of the following hypothesis:

H₉. 

IoT adoption is expected to have a favorable influence on SMFs’ energy conservation capability.

3. Methodology

This section provides a detailed account of the investigation’s layout, including information on the research location, the process of selecting respondents, and handling the questionnaire conduction. Additionally, this section provides a concise overview of the estimating approach employed.

3.1. Study Location and Research Design

This study included conducting a survey questionnaire from May to June 2024 in four districts (Zhenhai, Beilun, Haishu, and Yinzhou) of Ningbo City in Zhejiang Province, China. Please refer to Figure 3 for the research location. Ningbo is a significant sub-provincial city located in the northeastern region of Zhejiang Province. This city, situated in the southern commercial hub of the Yangtze River Delta, has always had a significant position in international commerce due to its historic and financial significance. The Port of Ningbo-Zhoushan is a highly advanced and efficient deep water port that combines inland, estuary, and seaport operations. It has the world’s highest cargo throughput and the world’s third-largest container volume, making it a versatile and comprehensive facility. As to the data from the Ningbo Bureau of Statistics, Ningbo’s regional gross domestic product (GDP) reached RMB 1.57 trillion (about USD 214.59 billion) in 2022, indicating a growth of 3.5% compared to 2021, whereas the per capita GDP was RMB 163,911 (about USD 22,403). Ningbo is rated 12th among China’s 300 cities in terms of business environment. The city has a robust and concentrated industry cluster specializing in innovative manufacturing. The manufacturing sector in the city saw a 3.3% rise from the previous year in its value addition, reaching about RMB 668.117 billion (around USD 91.31 billion) in 2022.

Since the implementation of the 13th Five-Year Plan, Ningbo has placed significant emphasis on the advancement of the service industry as a crucial means to accelerate the transformation of the economy. In line with its objectives, the city’s 14th Five-Year Plan aims to solidify its position as a national applications center and a regional frontrunner in the field of industrial internet connectivity by 2025. The plan targets generating over RMB 300 billion (nearly USD 41 billion) in revenues from software applications and data solutions. As of March 2023, the Ministry of Industry and Information Technology (MIIT) has acknowledged nine 5G-powered industries in the city as national exemplars, indicating substantial advancements in this field. The city has effectively implemented digital transition in more than 10,000 projects, with 81 projects being acknowledged as municipal demonstrative models for the usage of 5G and industrial internet technology. In the inaugural quarter of 2022, the city had a significant increase in its technologically advanced businesses, resulting in a value addition of RMB 77.09 billion (about USD 11.86 billion), which is the largest among the cities in the province. Significantly, this number represents an annual growth rate of 1.7% and accounts for 58.4% of the total obtained by large-scale industrial firms in the city over the same time frame. The city’s high-tech industries received an infusion of RMB 16.37 billion in investments, representing a growth of 11.8% compared to the previous year.

Regarding the researched areas, Zhenhai District, which is well known for its pharmaceutical and maritime businesses, has a sizeable number of SMFs operating in these fields. In Beilun District, which is host to the Ningbo-Zhoushan Port, a significant trade center, many SMFs are active in importation and exportation and associated logistical operations. Haishu District is the commercial and economic center of Ningbo, and as a result, it is home to a wide variety of SMFs operating in diversified service sectors, including technical services, advisory services, and banking. Finally, some SMFs in the service sectors, information technology, and small-scale production are the specialty of the Yinzhou District, which is a center for commerce, transit, and academia.

Surveys were administered to the owners of SMFs, who were either adopters or non-adopters of IoT technology. The owners of the firm were engaged in person to complete the questionnaires. Throughout this procedure, the principles and contents of the questionnaire were thoroughly explained to the firm owners in order to elicit well-informed replies. The completion of the questionnaire by each respondent required an average time of around 1.5 h. Since confidentiality was demanded by some of the individuals, their identities have been kept confidential to ensure data safety and consistency.

A partially structured survey questionnaire, developed after reviewing the relevant literature, was distributed for the pre-testing information it contained. The contents of the questionnaire were revised and amended based on the perspectives of the firms’ owners. Once all the survey materials were finalized, a total of 587 personally administered survey questionnaires were floated in English, as well as Chinese versions. Out of these, 491 questionnaires were gathered with valid and properly completed responses and were used for further empirical examination. The rate of responses (83.65%) exceeds the minimal standard of 20% and is thus suitable for further research [121]. Ref. [122] presented multiple grades to assess the suitability of the collected data size. The ratings were assigned numerical values as follows: very poor (50), poor (100), good (300), very good (500), and outstanding (1000). Considering the grades as mentioned above, the present sample size of 491 is close to the very good (500) grade, which is sufficient for data analysis.

Table 1. Sampling attributes.

Survey Elements	Facts
Survey administration timeframe	May 2024 to June 2024
Research site	* Zhenhai, Beilun, Haishu, and Yinzhou
Size of the sample	587
The number of valid responses	491
The number of adopters	259
The number of non-adopters	232
Rate of responses	83.65%

Note: * Selected four districts of Ningbo City of Zhejiang Province, China.

records the characteristics of the investigated sample. The survey questionnaire was organized to incorporate three distinct sorts of sections. The first segment focused on collecting the demographic characteristics of the respondents and the SMFs. The second segment focused on the parameters derived from the DIM framework that determine the adoption of IoT technology by enterprises. A five-point Likert scale was used to calculate the constituents for this segment. Respondents were instructed to indicate their level of agreement for each topic by assigning a number between “1” (representing “strongly disagree”) and “5” (representing “strongly agree”). The third segment consisted of inquiries about the sustainability performance of firms in order to examine the influence of IoT adoption on these performance metrics. The information included in this component consisted of quantitative (numeric) data.

3.2. Study Sample’s Demographic Attributes

The majority (56.59%) of the respondents who were the owners of the firms belonged to the middle-aged category (42.97%), after which came the young category (31.98%) and the old category (25.05%). The research sample consisted of 97 females, while the remaining 394 respondents were male. Regarding the educational background of firm owners, the majority (35.44%) had completed senior high school education, which typically involves 12 years of schooling. This was followed by individuals who held a bachelor’s degree (24.23%). By comparison, the respondents who had less than six years of primary education made up the smallest percentage, accounting for 2.65%. In terms of firm size, the proportion of responses from small-sized firms (56.82%) exceeded that of medium-sized firms (43.18%). Out of all the surveyed firms, the majority (51.73%) were classified as high-earning firms, earning more than 1,000,000 RMB per year. The second-largest group (29.53%) consisted of medium-earning firms, making between 500,001 and 1,000,000 RMB per year. The computer technology industry has the highest proportion of firms (26.68%), followed by the textile and garments industry (24.03%), among others.

Table 2. Demographic attributes of firms and respondents.

Demographic Attribute	Classifications	Respondents/Firms	Proportion (%)
Age of firm owner (years)	Young (24–40)	157	31.98
	Middle-aged (41–55)	211	42.97
	Old (>55)	123	25.05
Gender of firm owner	Male	394	80.24
	Female	97	19.76
Qualification of firm owner (schooling years)	Below primary (<6 years)	13	2.65
	Primary education (6 years)	21	4.28
	Junior secondary education (9 years)	48	9.78
	Senior high school education (12 years)	174	35.44
	Bachelor’s degree (16 years)	119	24.23
	Master or PhD (18 or above)	116	23.62
Firm size * (number of employees/annual revenue)	Small-sized (<300/0.5 to 5 million RMB)	279	56.82
	Medium-sized (300–2000/5 to 200 million RMB)	212	43.18
Firm revenue (RMB per annum)
Low-earning firms	100,000 to 500,000	92	18.74
Medium-earning firms	500,001 to 1,000,000	145	29.53
High-earning firms	Above 1,000,000	254	51.73
Firm type	Textile and garments	118	24.03
	Information technology	131	26.68
	Electronics	88	17.92
	Foods and beverages	69	14.05
	E-commerce and traders	85	17.31

Notes: RMB: Renminbi (China’s local currency unit). * This research utilized the standards outlined in the “Regulations on the Standards for Classification of Small and Medium-sized Enterprises” introduced in 2011 to classify small and medium-sized firms (SMFs) in China. These regulations establish the criteria for classifying SMFs, which are determined by three primary indicators: number of employees, annual revenue, and overall assets. However, we only consider the first two metrics for our particular requirements. It should be noted that the total of the fractions might not be exactly 100 due to rounding.

displays the demographic characteristics of the sample.

3.3. Explanations of Study Variables

Table 3. Classifying the variables of the study.

Variable(s)	Variables’ Classification
Dependent variable	The Internet of Things (IoT) adoption (binary in nature); IoT adopter = 1, IoT non-adopter = 0
Outcomes	Environmental sustainability performance (expenditures on natural resource consumption, expenditures on renewable energy technology products, expenditures on environmental monitoring systems), Economic sustainability performance (firms’ input costs, firms’ revenues, firms’ profits), Innovation performance (expenditures on innovative products), Energy conservation (expenditures on energy-efficient technology products, utility bills)
Independent variables	Demographic attributes (Age of firm owner, gender of firm owner, qualification of firm owner, firm size, and firm type), DIM framework factors (relative advantage, compatibility, complexity, trialability, observability)

classifies the variables of this investigation into three categories: dependent, outcome, and independent variables. In the case of the current study, IoT adoption is the dependent variable, represented by binary values where the adopter is coded as 1 and the non-adopter is coded as 2. The outcome factors encompassed environmental sustainability performance, economic sustainability performance, innovation performance, and energy conservation. Expenditures on natural resource consumption, renewable energy products, and environmental monitoring systems gauge environmental sustainability performance. Economic sustainability performance is gauged by firms’ input costs, revenues, and profits. Expenditures on innovative products measure innovation performance, while expenditures on energy-efficient products and utility bills gauge energy conservation. The questions defining the metrics of the above-mentioned sustainability performance indicators of SMFs can be referred to in

Table A1. Structured questionnaire.

1. Demographic Information of Respondents and Firms’ Attributes Respondents Are Requested to Indicate the Relevant Option for Each of the Following Statements:
1.1. Age of firm owner (years)	24–40 (young); 41–55 (middle-aged); Above 55 (old)
1.2. Gender of firm owner	1. Male; 2. Female
1.3. Qualification of firm owner (schooling years)	Below primary (<6 years); 2. Primary education (6 years), 3. Junior secondary education (9 years), 4. Senior high school education (12 years), 5. Bachelor’s degree (16 years), 6. Master or PhD (18 years or above)
1.4. Firm size (measured by number of employees and annual revenue of the firm)
1.4.1. Number of employees of the firm	1. less than 300 (small-sized); 2. 300–2000 (medium-sized)
1.4.2. Firm annual revenue	1. 0.5–5 million RMB (small-sized); 2. 5–200 million RMB (medium-sized)
1.5. Firm category by annual earning (RMB)	100,000–500,000 (low-earning firms); 2. 500,001–1,000,000 (medium-earning firms); 3. Above 1,000,000 (high-earning firms)
1.6. Firm type	Textile and garments; Information technology; Electronics; Foods and beverages; E-commerce and traders
2. Measurement items of exogenous constructs based on Diffusion of Innovation Modeling (DIM) framework factors Respondents are requested to indicate their degree of agreement or disagreement with the provided explanations of measurement items.
	1 = Strongly Disagree	2 = Disagree	3 = Neutral	4 = Agree	5 = Strongly Agree
Relative advantage (RLTV)
RLTV1: I believe that I have the financial capability to invest in IoT.
RLTV2: I believe adopting IoT will enhance the competitiveness of my business.
RLTV3: I believe adopting IoT will make business transactions much easier than manual methods.
RLTV4: I believe adopting IoT can partially substitute the labor force of my business.
RLTV5: I believe the process of marketizing IoT is beneficial.
RLTV6: I believe adopting IoT will enhance the efficiency of doing business than ever before.
Compatibility (CMPT)
CMPT1: I believe adopting IoT incurs excessive costs.
CMPT2: I believe the current structure of my business is difficult to modify through the implementation of IoT.
CMPT3: I believe that the number of employees who are capable of operating the new IoT configuration in my business activities is limited.
CMPT4: It will become difficult to upgrade IoT infrastructure if the future devices are incompatible with existing ones.
Complexity (CPLX)
CPLX1: I believe that IoT products are difficult to manage initially.
CPLX2: I believe that IoT products’ applications would necessitate frequent updates.
CPLX3: I believe adopting IoT might risk my business to security breaches.
CPLX4: I believe adopting IoT will involve data overload as the IoT devices generate a lot of data.
Trialability (TRLB)
TRLB1: I believe it might be appealing for IoT producers to provide a reimbursement policy on purchasing IoT technology.
TRLB2: I believe that prior expertise with technology would facilitate the acceptance of IoT technology products.
TRB3: I believe launching pilot programs of IoT applications in businesses could facilitate the acceptance of IoT technology products.
Observability (OBSR)
OBSR1: IoT is considered a valuable technology by my peers.
OBSR2: The IoT product users recommend adopting this technology due to its ground-breaking characteristics.
OBSR3: I believe that IoT technology wins broad social acceptance.
Endogenous construct: Adoption of IoT (IoTA)
IoTA1: I plan to adopt or have previously adopted IoT technology.
IoTA2: I believe that the use of IoT technology is quite valuable.
IoTA3: I am willing and able to autonomously decide on adopting IoT technology for my business.
3. Questions regarding the sustainability performance of small and medium-sized firms (SMFs) Respondents are requested to respond to each of the following questions.				Response
3.1. Environmental sustainability performance
3.1.1. How much does your firm spend annually on natural resources (water, coal, oil, natural gas, raw materials) (in RMB)?
3.1.2. How much does your firm invest annually in renewable energy technologies (in RMB)?
3.1.3. How much does your firm spend annually on environmental monitoring systems (in RMB)?
3.2. Economic sustainability performance
3.2.1. What is your firm’s annual input cost (in RMB)?
3.2.2. How much has your firm received in credits or financing (in RMB)?
3.2.3. How much annual revenue does your firm earn (in RMB)?
3.2.4. What is your firm’s annual profit (in RMB)?
3.3. Innovation performance
3.3.1. How much did your firm spend on sustainable technology innovations last year (in RMB)?
3.4. Energy conservation
3.4.1. How much does your firm invest annually in energy-efficient technologies (in RMB)?
3.4.2. How much does your firm spend in terms of utility bills (in RMB)?
4. IoT adopter versus non-adopter SMFs Respondents are requested to respond to the following questions in Yes/No.				Yes (adopter)	No (non-adopter)
4.1. For SMFs belonging to the “Textile and garments” industry
4.1.1. Does your firm use IoT sensors for predictive maintenance of equipment?
4.1.2. Is your firm using automated quality control systems with IoT integration to track defects during production?
4.1.3. Has your firm implemented IoT-enabled machinery for real-time monitoring of production processes?
4.1.4. Does your firm use Radio Frequency Identification (RFID)-based tracking or other IoT technologies to monitor inventory in real-time?
4.1.5. Does your firm use IoT data analytics to optimize production schedules or inventory management?
4.2. For SMFs belonging to the “Information technology” industry
4.2.1. Does your firm develop IoT solutions or products for clients?
4.2.2. Is your firm’s IoT data integrated with cloud-based or other computing systems?
4.2.3. Does your firm invest in research and development for IoT technologies?
4.2.4. Does your firm use IoT for remote monitoring and maintenance of client IT infrastructure?
4.2.5. Does your firm use IoT data analytics to offer predictive analysis for clients?
4.3. For SMFs belonging to the “Electronics” industry
4.3.1. Does your firm manufacture IoT-enabled consumer electronics (e.g., smart appliances, wearables)?
4.3.2. Does your firm use IoT-enabled sensors for real-time monitoring and control of manufacturing processes?
4.3.3. Has your firm implemented IoT for predictive maintenance of production equipment?
4.3.4. Is IoT integrated into your firm’s logistics and distribution operations for real-time tracking of shipments?
4.3.5. Does your firm collect and analyze data from IoT-enabled products for insights into customer usage or performance?
4.4. For SMFs belonging to the “Foods and beverages” industry
4.4.1. Does your firm use IoT-enabled sensors for real-time monitoring of production processes (e.g., temperature, humidity)?
4.4.2. Has your firm implemented IoT solutions to ensure product quality (e.g., monitoring for spoilage or contamination)?
4.4.3. Does your firm use IoT for real-time tracking of raw materials and finished products in the supply chain?
4.4.4. Does your firm use IoT-enabled systems to track food and beverage shipments and ensure they are stored at optimal conditions (e.g., cold chain management)?
4.4.5. Does your firm use IoT solutions to track expiry dates and automate inventory rotation to reduce food waste?
4.5. For SMFs belonging to “E-commerce and traders” industry
4.5.1. Does your firm use IoT to monitor real-time inventory levels in warehouses?
4.5.2. Does your firm use IoT-enabled systems for sorting, packing, and shipping orders?
4.5.3. Does your firm use IoT-based solutions to track shipments in real-time, from suppliers to your warehouse and from your warehouse to customers?
4.5.4. Does your firm use IoT systems to monitor environmental conditions (e.g., temperature, humidity) in warehouses or storage facilities?
4.5.5. Are IoT systems integrated with your business’s Enterprise Resource Planning (ERP) or Customer Relationship Management (CRM) systems to improve decision-making and customer service?

(Appendix A). The independent variables encompass demographic factors, as well as DIM framework factors, including relative advantage, compatibility, complexity, trialability, and observability. The DIM factors are adapted and modified from the study of Ahmad et al. [123], who investigated the factors influencing biogas technology adoption by constructing DIM factors.

3.4. Statistical and Econometric Techniques

On the one hand, this study utilizes partial least square-based structural equation modeling (PLSEQ) to evaluate the factors that influence the uptake of the IoT. The PLSEQ is employed due to its quantitative advantages, as it is well suited for multiple types of factor analyses. This approach aids in estimating the connections among latent (hidden) factors by utilizing observable items obtained from surveys. The PLSEQ demonstrates a strong performance by maintaining uniformity for data that do not have a normal distribution and mitigating the impact of small sample biases. Therefore, this study chooses this approach to investigate the queries under analysis. In accordance with the standards defined by [124], this study employs a dual-stage methodology that incorporates measurement, as well as structural modeling layouts. In order to achieve this, the SPSS plugin Amos (version 23.0) has been utilized in this study.

In addition, this study utilized the propensity score matching (PSM) technique to examine the impact of IoT adoption on firms’ sustainability performance. When doing statistical evaluation, it is important to take into account the key distinguishing characteristics between the various categories. This is necessary to address possible errors in estimating, such as self-selection and explicit biases. The current investigation employs a matching strategy to address the biases that arise from the variations in observable characteristics between the control and treatment categories. In order to facilitate comparisons, individuals who utilize the IoT and those who do not are evaluated based on various discernible characteristics. At first, the matching implies that the categories being analyzed contain disparities in traits that cannot be directly observed. The matching procedure considers all discernible characteristics of both adopters and non-adopters of the IoT. Nevertheless, incorporating a growing variety of factors diminishes the likelihood of finding a suitable match [125]. The escalation of the categories’ matching characteristics significantly amplifies the seriousness of the issue by expanding the number of matched dimensions, a phenomenon referred to as the dimensionality curse [126]. Utilizing the PSM method can resolve this problem [127]. The PSM method compares the characteristics of every treatment individual (consumer) to those of someone in the untreated group (control) who has similar characteristics. This is done to produce a match score rather than simply juxtaposing the predefined parameters. The PSM establishes a controlled category that exhibits visible characteristics that are comparable to the persons receiving treatment. As a result, it effectively reduces the risk of encountering a “dimensionality curse” by generating a distinct matched score.

Concerning the current investigation, there is a possibility of bias related to self-selection due to potential differences in the characteristics of adopters and non-adopters. To address this bias, the present study employs PSM, which mimics the process of the randomization of samples by matching adopters and non-adopters according to their covariate attributes [127]. The similarity of the basic characteristics between the underlying two categories allows for the assessment of treatment effects. Yet, the lack of arbitrary samples may hinder this evaluation, given the presence of different essential traits. The implementation of the IoT serves as a means of treatment. The key characteristics of the treatment category were determined by considering factors such as the gender of the firm owner, firm age, firm revenue, and firm size. Using those traits, the respondents’ propensity scores were computed using STATA 17 software. Following is the formulation of the PSM technique:

$$Treatment=\left(P_i/1-P_i\right)={\phi }_0+{\phi }_{1}G+{\phi }_{2}A+{\phi }_{3}Q+{\phi }_{4}FR+{\phi }_{5}FS+{\phi }_{6}FT+{ϵ}_i$$

(1)

where $P$ stands for probability; thereby, $P_i$ represents the probability that a firm adopts the IoT, while $1-P_i$ indicates the probability that a firm does not adopt the IoT. Hence, the expression $\left(P_i/1-P_i\right)$ refers to the odds of adopting the IoT. Therefore, this expression models the binary outcome of IoT adoption. The constant term ${\phi }_0$ represents the drift parameter of the model, whereas ${\phi }_1$ through ${\phi }_6$ represent the parametric slopes associated with the respective demographic characteristics. The variable $G$ represents the gender of the firm owner, with a value of 1 indicating male and 0 indicating female. $A$ depicts the age of a firm owner in years. $FR$ denotes the firm revenue in Chinese renminbi (RMB). The firm size is abbreviated as $FS$ and measured by the number of employees, as well as annual revenue. $FT$ stands for firm type, which is defined as per business categories of the firms. In addition, ${ϵ}_i$ represents the probabilistic element of the model, which encompasses additional factors that influence the propensity score but are not part of the model. This study utilized a z-score to expand the spectrum of the scales in accordance with the notion of normalization, as stated below:

$$z_i=x_i-\overline{x_i}/{\sigma }_i$$

(2)

where $z_i$ represents the z-score, $x_i$ reflects the score of the primitive scale, $\overline{x_i}$ denotes the average of the sample, and ${\sigma }_i$ represents the standard deviation of the sample. The utilization of the z-score enables the equalizing impact of the given independent variables while computing the propensity scores. The PSM utilizes algorithms that consider numerous criteria to match individuals based on their associated treatment. The current investigation utilizes optimal pair matching (OPTM), nearest neighbor matching (NNB), and kernel matching (KRM) algorithmic procedures to ensure reliable and strong results. To determine the average treatment effect on the treated (ATET), a significant level of shared support is necessary [128].

The procedure of generating data showed an upward skewness, indicating the presence of non-normally distributed data. On the other hand, parametric approaches rely on the premise of normal distribution, making them unsuitable for use, particularly when dealing with smaller sample sizes, such as in the current scenario. Against this background, the utilization of nonparametric bootstrapping approaches was crucial in the present study to calculate the ATET while testing the hypotheses. These approaches take into account the empirical distribution of the dependent variable based on its distributional statistical operation rather than considering a predefined distribution of the data. In this regard, the statistical distribution is formed by considering the subsequent sampling of the initial data set to create confidence bands and test the associated hypotheses. Confidence bands and testing of hypotheses can be constructed without the requirement of adhering to the normally distributed data, stable variance, and specific preset distributional topologies [129]. These characteristics contribute to the method’s popularity when dealing with data samples that lack an established distribution. Experimental research has shown the statistical advantage of this strategy compared to other nonparametric procedures, particularly for skewed data [130]. Therefore, the resampling process entails these subsequent steps:

$$UD\to y=\left(y_1,y_2,y_3,\dots ,y_n\right)\to DS\left(y\right)$$

(3)

$UD$ represents the manifestation of the probability distribution of an unknown sample, resulting in an arbitrary sample denoted as $y$. The term $DS\left(y\right)$ refers to the specific statistic that is desirable. By employing the “with swap” function on the set $\left(y_1,y_2,y_3,\dots .,y_n\right)$, a bootstrap sample is generated by randomly producing $n$ samples. This process yields the expression as follows:

$$\widehat{SD}\to y^{\mathrm{*}}=\left({y_1}^{\mathrm{*}},{y_2}^{\mathrm{*}},{y_3}^{\mathrm{*}},\dots ,{y_n}^{\mathrm{*}}\right)\to DS\left(y^{\mathrm{*}}\right)$$

(4)

The symbol $\widehat{SD}$ represents a statistical distribution obtained through reprocessing to create a bootstrap sample, denoted as $y^{*}$. The term $DS\left(y^{*}\right)$ refers to the desirable statistical result of $\widehat{SD}$. The selection of an optimal number of replications, which refers to the number of samples used to bootstrap created throughout the bootstrap procedure, is of utmost importance. Prior research has contended that the repetitions would be satisfactory as long as the inclusion of extra repetitions does not substantially modify the $DS\left(y^{*}\right)$ [131]. Based on specific criteria, it was determined that $5\times {10}^3$ cycles were appropriate for the current investigation. Prior investigations have determined the resampling technique’s optimal precision and power statistic based on the total number of repetitions [132].

4. Results and Discussion

4.1. Results of Structural Equation Modeling

To begin, the Fornell-Larcker approach is utilized, as proposed by Fornell and Larcker [70], to determine if discriminant validity (DSV) exists. The DSV is based on the assertion that the average variance extraction-based squared root (SQRAVE) for every construct of the latent nature (CLT) has to be greater than the reflective association of that construct in order to show DSV [133]. The results of the DSV assessment are recorded in

Table 4. Outcomes of discriminant validity.

Factors	$\mathit{R}\mathit{L}\mathit{T}\mathit{V}$	$\mathit{C}\mathit{M}\mathit{P}\mathit{T}$	$\mathit{C}\mathit{P}\mathit{L}\mathit{X}$	$\mathit{T}\mathit{R}\mathit{L}\mathit{B}$	$\mathit{O}\mathit{B}\mathit{S}\mathit{R}$	IoTA
RLTV	[0.876]
CMPT	0.517	[0.838]
CPLX	0.139	0.136	[0.810]
TRLB	0.612	0.464	0.408	[0.872]
OBSR	0.263	−0.385	0.237	0.336	[0.854]
IoTA	0.398	0.419	0.581	0.192	−0.623	[0.835]

Note: [ ] enclose SQRAVE values.

. The cross-association between every CLT and the other CLTs is not higher than their respective SQRAVE, demonstrating DSV’s presence in the data under analysis.

After confirming DSV, it shows the convergence validity (CNV) and reliability of internal consistency (RIC) of the itemized components of CLTs (refer to

Table 5. Outcomes of convergence validity and reliability of internal consistency.

	CNV		RIC
CLTs and Respective Itemized Components	External Loads	AVE	CMR	CR-Alpha
Relative advantage (RLTV)
RLTV1: I believe that I have the financial capability to invest in IoT.	0.819	0.795	0.857	0.711
RLTV2: I believe adopting IoT will enhance the competitiveness of my business.	0.836
RLTV3: I believe adopting IoT will make business transactions much easier than manual methods.	0.801
RLTV4: I believe adopting IoT can partially substitute the labour force of my business.	0.825
RLTV5: I believe the process of marketizing IoT is beneficial.	0.733
RLTV6: I believe adopting IoT will enhance the efficiency of doing business than ever before.	0.751
Compatibility (CMPT)
CMPT1: I believe adopting IoT incurs excessive costs.	0.832	0.778	0.892	0.742
CMPT2: I believe the current structure of my business is difficult to modify through the implementation of IoT.	0.806
CMPT3: I believe that the number of employees who are capable of operating the new IoT configuration in my business activities is limited.	0.781
CMPT4: It will become difficult to upgrade IoT infrastructure if the future devices are incompatible with existing ones.	0.814
Complexity (CPLX)
CPLX1: I believe that IoT products are difficult to manage initially.	0.815	0.740	0.884	0.729
CPLX2: I believe that IoT products’ applications would necessitate frequent updates.	0.768
CPLX3: I believe adopting IoT might risk my business to security breaches.	0.729
CPLX4: I believe adopting IoT will involve data overload as the IoT devices generate a lot of data.	0.761
Trialability (TRLB)
TRLB1: I believe it might be appealing for IoT producers to provide a reimbursement policy on purchasing IoT technology.	0.824	0.732	0.857	0.713
TRLB2: I believe that prior expertise with technology would facilitate the acceptance of IoT technology products.	0.767
TRB3: I believe launching pilot programs of IoT applications in businesses could facilitate the acceptance of IoT technology products.	0.731
Observability (OBSR)
OBSR1: IoT is considered a valuable technology by my peers.	0.816	0.783	0.875	0.741
OBSR2: The IoT product users recommend adopting this technology due to its groundbreaking characteristics.	0.839
OBSR3: I believe that IoT technology wins broad social acceptance.	0.728
Adoption of IoT (IoTA)
IoTA1: I plan to adopt or have previously adopted IoT technology.	0.836	0.719	0.851	0.701
IoTA2: I believe that the use of IoT technology is quite valuable.	0.798
IoTA3: I am willing and able to autonomously decide on adopting IoT technology for my business.	0.843

Notes: AVE: average variance extracted, CMR: composite reliability, and CR-Alpha: Cronbach alpha.

). The CNV is evaluated by examining the external loads and the average variance extracted (AVE) of the itemized components within the CLTs. Ref. [134] stated that external loads and AVEs of more than 0.7 serve as strong evidence of CNV. This holds correct in the case of each itemized component of the CLTs in this study, confirming the measurement model’s CNV. Simultaneously, the RIC is assessed using the Cronbach alpha (CR-Alpha), as well as composite reliability (CMR). A model that has elevated RIC exhibits measures that are highly connected. For the present research, the calculated values of CR-Alpha and CMR surpass the suggested criterion of 0.70 [135], confirming the modeled data’s RIC.

The Kaiser–Meyer–Olkin (KMO) for sampling adequacy and Bartlett examination of sphericity has been applied in the subsequent phase to make an assessment of the measurement model’s connection with the analysis data (refer to

Table 6. Outcomes of KMO and Bartlett testing.

Adequacy of Sample by KMO Testing		0.950
Bartlett sphericity	χ2 approximate	4261.037
	DoF	139
	Significant @	0.000

Notes: DoF: degree of freedom and KMO: Kaiser–Meyer–Olkin.

). The KMO rating can vary between 0 and 1. If the rating is nearer to 1, it suggests a sounder association between factors, and a rating approaching 0 would suggest a weaker association. Bartlett testing assesses the independence of every factor. The outcomes recommend that the measurement model is suitable, as indicated by the KMO statistic of 0.938. Additionally, the Bartlett testing stays statistically significant at the one percent significance level, suggesting that the data under consideration is appropriate for analysis.

Once the measurement model has been established successfully, the postulated structural paths are assessed using the structural model. In order to achieve this objective, a bootstrap technique is applied considering 5000 sample repetitions, following the previous empirical works [136]. In order to assess the proposed structural paths, the probability ratings are employed at several significance levels, specifically 1%, 5%, and 10%.

Table 7. Outcomes of the structural model.

Hypothesized Path			βs	Conclusion	f2	R2	Q2
H1: RLTV	→	IoTA	0.837 ***	Acceptance	0.510	0.674	0.393
H2: CMPT	→	IoTA	−0.478 ***	Acceptance	0.291
H3: CPLX	→	IoTA	−0.381 **	Acceptance	0.232
H4: TRLB	→	IoTA	0.613 ***	Acceptance	0.374
H5: OBSR	→	IoTA	0.692 ***	Acceptance	0.422

Notes: *** and ** indicate the significance levels of 1% and 5%, respectively.

presents the assessment of structural paths using the PLSEQ methodology. The proposed routes meet the criteria for assessment, and all hypothesized paths except H₃ gained acceptance outcomes due to respective estimated βs displaying statistical significance at a level of 1%. Simultaneously, the estimated β of H₅ showed statistical significance at a level of 5%. The estimated model indicates that H₁, H₄, and H₅ favorably impact the adoption of IoT technology by SMFs. Specifically, relative advantage (β = 0.613), trialability (β = 0.542), and observability (β = 0.397) all contribute to this favorable effect. Figure 4 provides a visual representation of these interactions. When comparing them, it is evident that H₂ and H₃ have a detrimental impact on the adoption of IoT technology by SMFs due to the compatibility (β = −0.536) and complexity (β = −0.491) factors. An additional criterion known as f² is utilized to measure the magnitude of the estimated βs’ impact. According to the [137] standard, the f² classifies the effect sizes into three categories: small (≥0.02), medium (≥0.15), and substantial (≥0.35). Based on the above, all the factors are found to have a substantial impact on the adoption of IoT technology by SMFs. Furthermore, the demographic characteristics of the SMFs and respondents, such as firm revenue, firm size, age, and qualification of the firm owner, and firm type have been found to have an affirmatively significant impact on the adoption of IoT technology by SMFs.

To assess the model’s prediction power, the R² (i.e., coefficient of determination) has been analyzed [134]. The R² rating of 0.674 indicates that 67.4% of the variability in the data can be described by this study’s structural model. This suggests that the model under consideration has a significant ability to predict, as it surpasses the suggested minimum value of 25% set by previous empirical studies [134]. In addition, the prediction usefulness of the model has been assessed using Stone–Geisser’s Q² standard [138,139]. The Q² statistic for the current model (0.393) meets the recommended criteria for a non-zero rating, providing compelling proof of its prediction usefulness. In addition, f² reflects the effect size of an exogenous CLT (independent variable) on an endogenous CLT (dependent variable) within the structural model. It helps in understanding the relative importance of independent variables in terms of their contributions to the dependent variable. For a robustness check, the endogeneity and multicollinearity results are provided in

Table A2. Outcomes of the multicollinearity and endogeneity testing procedure.

Hypothesized Path			βs	Conclusion	VIF (Threshold < 10)
H1: RLTV	→	IoTA	0.825 ***	Equivalent	3.267
H2: CMPT	→	IoTA	−0.469 **	Equivalent	1.583
H3: CPLX	→	IoTA	−0.372 **	Equivalent	6.378
H4: TRLB	→	IoTA	0.598 **	Equivalent	4.117
H5: OBSR	→	IoTA	0.680 ***	Equivalent	2.956

Notes: *** and ** indicate the significance levels of 1% and 5%, respectively. VIF stands for variance inflation factor.

(see Appendix A).

The factors influencing the adoption of IoT technology by SMFs are prioritized based on their individual effect sizes (refer to Figure 5). Given the calculated f² scores, the relative advantage factor has the highest priority (with f² = 0.510), while the complexity factor is placed at the lowest (with f² = 0.232). The order of priority is as follows: RLTV > OBSR > TRLB > CMPT > CPLX. SMFs prioritize the relative advantage of novel items like IoT technology, whereas complexity is of the least significance to them. Moreover, the primary factor that strongly influences the adoption of the IoT by SMFs is the relative advantage, while the main barrier is compatibility.

Consistent with our findings, Rey et al. [38] found that the perceived usefulness of IoT adoption positively shaped the acceptance behavior of firms and individuals belonging to the transportation and logistics industries. In the same vein, Jaspers and Pearson [39] also emphasized the positive contribution of perceived usefulness and trust in technology in determining IoT adoption by New Zealanders. Such an outcome is somewhat relevant to our results, given that the trialability of the IoT may also build a certain level of trust in this technology to shape the adoption behavior of individuals and firms positively. Aligned with our findings, Ali et al. [40] empirically manifested that technological advantages, such as efficient logistic structures and information safety, could enhance the IoT adoption behavior in supply chain management. On the one hand, Hsu and Lin [41] found that the perceived benefits positively contributed to the adoption of IoT services, which is aligned with our findings. On the other hand, they found that perceived compatibility promoted the IoT services, which contradicts our outcome, because the compatibility feature is disclosed to be a barrier to IoT adoption in the current study. Aligned with our finding of the positive contribution of the observability to IoT adoption, Shaqrah and Almars [42] authenticated that public support was influential in promoting IoT adoption in the education industry in the Saudi Arabian context. In addition, Langer et al. [43] empirically evidenced that high technology cost was a barrier to adopting IoT sensors in the dairy farming industry, which presents an analogy to our finding, as compatibility is found to impede the IoT adoption behavior of SMFs in the current study. Entrepreneurs view the high capital costs of the IoT as a visible barrier to its adoption from the SMFs’ perspective. Finally, an insignificant impact of the perceived ease of use on blockchain technology adoption from the Chinese perspective found by Esfahbodi et al. [44] is not consistent with our finding of complexity, which is found to be a barrier to IoT adoption by SMFs.

4.2. Results of Propensity Score Matching

In this investigation, the OPM has been used as a fundamental modeling algorithm because of its widespread use in social sciences, among others. The significant outcomes were revealed using the predicted ATET, which may be found in

Table 8. IoT adoption and sustainability performance of SMFs: outcomes obtained by propensity score matching (PSM).

Matching Standard	Outcome Variable(s)	Treated	Controls	ATET
Optimal pair matching (OPM)	Environmental sustainability (ENS) performance	(H6: IoTA → ENS performance)
	Expenditures on natural resource consumption	193	174	−2.517 **
	Expenditures on renewable energy technology products	193	174	5.628 ***
	Expenditures on environmental monitoring systems	193	174	1.269 ***
	Economic sustainability (ECS) performance	(H7: IoTA → ECS performance)
	Firms’ input costs	193	174	−3.182 ***
	Firms’ access to credits	193	174	4.873 ***
	Firms’ revenues	193	174	9.836 **
	Firms’ profits	193	174	7.519 **
	Innovation (INO) performance	(H8: IoTA → INO performance)
	Expenditures on innovative products	193	174	2.878 ***
	Energy conservation (ECO)	(H9: IoTA → ECO)
	Expenditures on energy-efficient technology products	193	174	7.931 ***
	Utility bills	193	174	−1.994 ***

Note: ** and *** indicate 5% and 1% levels of significance, respectively.

and Figure 6. To ensure the reliability of this study’s outcomes, an additional algorithm, known as NNM, is also applied (refer to

Table A4. IoT adoption and sustainability performance of SMFs: robustness outcomes by NNM.

Matching Standard	Outcome Variable(s)	Treated	Controls	ATET
Nearest neighbor matching (NNM)	Environmental sustainability (ENS) performance	(H6: IoTA → ENS performance)
	Expenditures on natural resource consumption	201	182	−2.491 ***
	Expenditures on renewable energy technology products	201	182	5.570 ***
	Expenditures on environmental monitoring systems	201	182	1.229 **
	Economic sustainability (ECS) performance	(H7: IoTA → ECS performance)
	Firms’ input costs	201	182	−3.117 ***
	Firms’ access to credits	201	182	4.706 ***
	Firms’ revenues	201	182	9.753 ***
	Firms’ profits	201	182	7.425 **
	Innovation (INO) performance	(H8: IoTA → INO performance)
	Expenditures on innovative products	201	182	2.794 **
	Energy conservation (ECO)	(H9: IoTA → ECO)
	Expenditures on energy-efficient technology products	201	182	7.815 ***
	Utility bills	201	182	−1.839 ***

Note: ** and *** indicate 5% and 1% levels of significance, respectively.

in Appendix A to see its outcomes). Nevertheless, OPM is given preference when evaluating the size of ATET estimations.

Regarding the energy conservation aspect of SMFs, each algorithm demonstrated that the acceptance of IoT produced a significant and favorable impact (@ 1% level of significance) on SMFs’ expenditures on energy-efficient technology products. The ATET was affirmative (7.931) and statistically significant. This suggests that firms that embrace energy-efficient technology products are expected to spend approximately 7.9% more on these products compared to firms that do not adopt them. The disparity indicates that companies that used IoT technology allocated more resources towards energy conservation initiatives compared to those that did not deploy the IoT. The ATET on the SMFs’ utility bills was found to be negative (−1.994) and highly significant at the 1% level. This indicates that adopter firms anticipate paying around 2% less utility bills compared to non-adopter firms. IoT technology adopter SMFs should expect reduced utility bills as a result of the advanced technology and energy-saving features. Supporting our findings, Abbas et al. [54] brought the evidence to light that future 6G-supported cellular IoT devices would need zero energy consumption levels, because those devices would not exploit manual charging or batteries. In addition, Hu [56] reasoned that the application of IoT in the thermal energy management system helps conserve energy by realizing the feedback through real-time monitoring, making the system more responsive and flexible.

Concerning the environmental sustainability performance of SMFs, the ATET on the expenditures on natural resource consumption had a negative value (−2.517) and displayed a statistical significance at the 5% level. This suggests that firms that adopt IoT technology will probably incur 2.5% lower expenditures on natural resource consumption compared to firms that do not adopt IoT technology. The ATET of expenditures on renewable energy technology products has been determined to be affirmative (5.628) and highly significant at 1%. The findings suggest that businesses that embraced IoT technology had approximately 5.6% higher expenditure on environmental conservation initiatives compared to businesses that did not use this technology. Furthermore, the ATET value for expenditures on environmental monitoring systems was found to be affirmative (1.269) and statistically noteworthy (with a significance threshold of 1%). This study revealed that SMFs who use IoT technology allocate around 1.3% more of their resources towards environmental sustainability initiatives compared to companies that do not adopt IoT. This suggests that the IoT has a positive impact on the environmental sustainability performance of SMFs. This finding can be justified by the fact that IoT adopter SMFs employ this technology for real-time monitoring and control over resource consumption and waste reduction. IoT technology also allows firms to automate energy-saving mechanisms at the strategic level of their production activities. Moreover, IoT adopter firms view investments in renewable energy and environmental monitoring systems as long-run opportunities to cut operational costs and to have less dependence on scarce natural resources. However, IoT non-adopter firms pay less attention to these indicators, which leads to reduced resource efficiency and limited expenses for sustainable technological solutions. In conformance with our findings, Masoomi et al. [50] concluded that IoT adoption by firms’ renewable energy supply chains amplified their environmental sustainability by contributing to eco-friendly processes. Likewise, Mishra et al. [51] attested that AI, IoT, and blockchain technology solutions assist corporate decarbonization. In line with our findings, Ding et al. [53] established that a boosted level of IoT adoption by China’s logistics sector was significantly linked to the reduction of carbon and particulate matter emissions. Corroborating our findings, Guan [55] substantiated that low-carbon cities applying IoT showed reduced atmospheric pollution levels than those without its application, confirming the positive role of the IoT in environmental sustainability.

With regards to the economic sustainability performance of SMFs, the ATET on firms’ input costs was found to have a negative value (−3.182) and be statistically important at 1%. This empirical output means that firms that adopt IoT technology have input costs that are 3.2% lower than firms that do not adopt the IoT. As a result, they experience economic benefits from using IoT technology. Moreover, the ATET value for firms’ access to credits was determined to be positive (4.873) and sufficiently significant (at a significance level of 1%). These data indicate that SMFs that utilize IoT technology have a 4.9% higher access to credits compared to those that do not implement the IoT. This result is supported by Zhang et al. [30], who advocated that digitalization effectively alleviated the financing constraints of Chinese small and medium-sized enterprises through digital financial services, like credits, monetary funds, and insurance, among others. The coefficient of firms’ revenues has been shown to be positive (9.836) and significant at the 5% level of significance. This discovery indicates that companies that used IoT technology experienced a revenue increase of roughly 9.8% in comparison to companies that did not utilize this technology. The analysis revealed that the ATET on firms’ profits was determined to be positively significant (7.519) at the 5% confidence level. This result suggests that adopter firms expect to generate approximately 7.5% higher profits in comparison to non-adopter firms. This outcome implies that firms adopting IoT technology show greater economic sustainability performance. This finding can be explained through the features of predictive maintenance, enhanced productivity, and mitigated downtime of IoT adopter SMFs, enabling them to exhibit efficiency and optimized operational capability, which significantly lower their input costs. Further, IoT-enabled firms may launch innovative product solutions, generating additional revenue streams. IoT technology may also offer SMFs personalized experience of services focusing on client satisfaction to guide the growth of sales and revenues. In addition to this, IoT-based firms manifest data-driven and reflective progress, which presents them in a lucrative fashion to creditors, thus enhancing their access to credits and funds. Contrarily, IoT non-adopter firms lag behind the adopter firms in those indicators due to traditional production and marketing strategies of producing and selling their products. This finding is closely related to those of Attaran et al. [47], who unfolded the effects of industrial IoT adoption on production efficiency and product quality enhancement in the manufacturing sector. In parallel to our findings, Tan et al. [48] confirmed that IoT adopter firms had enhanced levels of productivity, financial performance, and market value. Similarly, the outcome of Nalajala et al. [49] substantiated our empirical evidence by showing that data analytics processes based on IoT technology were helpful in promoting firm production levels by cutting costs and improving efficiency. In agreement with our findings, Musarat et al. [52] validated that IoT implementation in the building sector of Malaysia instigated elevated levels of industrial productivity, sustainability, and product quality.

Finally, when the innovation performance of the SMFs was taken into consideration, the ATET of expenditures on innovative products was found to be positive (2.878), with a statistical significance of 5%. This indicates that the use of IoT technology resulted in an increase of around 2.9% in firms’ consumption of innovative products. It led to the conclusion that IoT technology enhances SMFs’ use of innovative products, contributing to their innovation performance. This finding can be justified in that IoT technology enables SMFs to have real-time access to research and development (R&D) data, which allows them to explore process and product innovation more efficiently than non-adopter firms. Additionally, resources saved due to IoT technology are allocated to drive innovation ventures to provide the adopter firms a competitive advantage over IoT non-adopter firms. Also, IoT adopter SMFs aim to provide client-centric and innovative solutions, which leads them to invest in innovative technologies. In addition, IoT technology can evolve the culture of technology-dependent growth, motivating IoT adopter firms to continue to grow through innovative ventures. On the contrary, using traditional systems, non-adopter SMFs do not make aggressive investments in innovative products. This finding supported that of Qin [45], who empirically unveiled that IoT and AI capability adoption were supportive of enhancing frugal innovation in the Chinese context. This scenario is particularly relevant for consumers in developing countries and regions with constrained incomes, where frugal innovation driven by IoT and AI technologies can ensure affordability, scalability, resource efficiency, and sustainability. Aligned with our result, Ge et al. [46] found and documented that digital innovation networks enhanced the innovation performance of Chinese A-share listed firms by improving their implicit knowledge acquisition.

5. Conclusions and Policy Recommendation

Unlike the prior studies, this investigation studied the diffusion of innovation modeling (DIM)-based factors influencing the adoption of the Internet of Things (IoT) technology by small and medium-sized firms (SMFs) and gauged the influence of such adoption on those firms’ sustainability performance. The main conclusion points may be summarized as follows: First, the factors of relative advantage, trialability, and observability positively drove IoT technology adoption. On the other hand, compatibility and complexity significantly discouraged the IoT technology adoption behavior of SMFs. Second, when evaluating the prioritization of DIM-based adoption factors, it is evident that relative advantage is the most powerful motivator, while compatibility poses the greatest obstacle to the adoption of IoT technology. Third, the IoT technology adopter SMFs spent less on natural resource consumption while they spent more on renewable energy technology and environmental monitoring systems compared to the non-adopter firms, demonstrating the improved environmental sustainability performance of the adopter firms. Fourth, when compared to the non-adopters, the IoT technology adopter firms had excessive revenues, profits, and access to credits while incurring less input costs, manifesting the enhanced economic sustainability performance of the adopter firms. Fifth, the IoT adopter firms incurred more expenditures on innovative products than those of non-adopter firms, which provided evidence of an escalated level of innovation performance of the adopter firms. Finally, demonstrating the relatively reduced utility bills and increased spending on energy-efficient technology products, IoT technology adopter SMFs secured greater energy conservation capability in contrast to the non-adopter firms.

To realize the full potential of IoT for a more sustainable and inventive future, authorities may pursue a variety of policy actions. Firstly, the strengthening and implementation of IoT technology standards and regulations should be ensured by (i) creating clear and rigorous cybersecurity standards for IoT devices to assure data security and privacy while also promoting confidence and encouraging widespread use; (ii) promoting interoperability standards across IoT platforms and devices would allow for frictionless connection and data sharing, increasing the value obtained from IoT ecosystems; and (iii) setting minimal energy efficiency criteria for IoT devices, incentivizing manufacturers to create low-power, energy-saving products. Secondly, the incentivization of financial resources should be secured via (i) providing tax benefits or subsidies to enterprises that invest in IoT technologies designed to improve environmental performance and energy efficiency, which would increase adoption and make technology more accessible, and (ii) accelerating the issue of green bonds designed expressly to fund IoT initiatives that prioritize sustainability and energy efficiency, allowing to encourage private investments in eco-friendly IoT applications. Thirdly, the allocation of resources should be diverted to research and development avenues by (i) promoting public financing for research and development of next-generation IoT technologies, with an emphasis on environmental and energy uses to promote innovation and speed up the development of new solutions, and (ii) encouraging cooperation among government agencies, educational organizations, and private firms to speed up the fabrication and deployment of environmentally friendly IoT solutions. Fourthly, capacity development and awareness should be prioritized via (i) creating training programs for organizations and workers on the advantages and uses of the IoT for ecological and financial gain, providing them with the skills and knowledge required to use technology properly, and (ii) launching public awareness campaigns to inform clients about the environmental advantages of IoT-enabled devices and services, stimulating responsible consumption and demand for sustainable alternatives. Finally, IoT infrastructure development should be focused by (i) investing in increasing broadband infrastructure to provide consistent and economical internet access in all areas to provide a solid platform for the countrywide deployment and use of IoT technology and (ii) supporting the construction of smart grid infrastructure, which may seamlessly incorporate energy from sustainable sources and improve the transmission of energy using IoT technology. Following these policy recommendations, governments may foster an environment in which firms can embrace IoT and contribute to a more sustainable, inventive, and energy-efficient future. This joint approach will pave the road for a digital economy that promotes both economic development and environmental responsibility.

Although this research has made significant advancements, there are still some drawbacks that need to be addressed in future investigations. Initially, the present investigation just used the persuasive channel of the DIM framework. However, to gain a more profound understanding of this framework, it is endorsed that future research takes into account its knowledge, decision, implementation, and confirmation channels. Furthermore, this study’s sample was restricted to the specific districts of Ningbo City in Zhejiang Province, China. However, future research might be enhanced by surveying the provincial or even national level in order to gain more comprehensive data and ensure greater generalizability. Also, while this investigation focused on the sustainability performance of SMFs, future works should also include micro and large firms to provide rich findings. In the end, this study considered mixed SMFs for data analysis, but for heterogeneous findings, future studies are encouraged to conduct multi-industrial and multi-sectoral analyses for a comparative landscape of empirical investigation.