The Antecedents of Willingness to Adopt and Pay for the IoT in the Agricultural Industry: An Application of the UTAUT 2 Theory

Abstract

This paper aims to examine the factors influencing the willingness of Bangladeshi farmers to adopt and pay for the Internet of Things (IoT) in the agricultural sector by applying the theoretical framework of the Unified Theory of Acceptance and Use of Technology 2 (UTAUT 2). To this end, the study employed a quantitative research methodology and obtained data from 345 farmers from the northern districts of Bangladesh. Using a cross-sectional survey design and convenience sampling method, a study of premium fruit growers was undertaken to assess IoT use in agriculture, and the primary survey data were analyzed using the Structural Equation Modeling (SEM) approach via AMOS 26. The study confirmed that effort expectancy, performance expectancy, facilitating condition, hedonic motivation, government support, price value, personal innovativeness, and trust influence the willingness of Bangladeshi farmers to adopt the IoT. Additionally, predictors such as trust and willingness to adopt were observed to influence the willingness to pay for the IoT, while the construct ‘performance expectancy’ produced no effect. The study also revealed that the willingness to adopt moderates the association between performance expectancy, price value, and willingness to pay for the IoT. This research has novel implications because it investigates the behavior of rural customers with respect to innovation adoption, which in this case is the IoT in agriculture. It outlines precise reasons for the willing adoption of the IoT in agriculture, which will, in turn, assist marketers of IoT technology in the design of appropriate marketing strategies to increase acceptance in rural areas. Using the proposed model that incorporates farmers’ willingness to pay, this empirical study takes the first step in examining whether farmers in a developing economy such as Bangladesh will adopt and pay for the IoT.

1. Introduction

Agriculture resulted in the birth of sedentary human civilization and remains one of the world’s most significant sectors. However, despite its importance, a vast majority of agricultural techniques remain conventional [1]. With a predicted worldwide population of 9.6 billion by 2050, a 70% increase in global food production is required to fulfill the ever-rising demand [2]. According to the UN Environment Programme’s (UNEP) Food Waste Index Report 2021 [3], almost 17% of the world’s food output may be wasted. Demand growth, food waste, infrastructure gaps, and inefficient production practices all put enormous strain on the global food supply chain. For this reason, there is an urgent need for a wide-ranging industrial growth of the agricultural sector, as well as the deployment of novel technology to modernize conventional agrarian practices [1]. Internet of Things (IoT) is a term that has become increasingly popular in recent years in this context. The capacity to deal with dynamic workflow environments enables the IoT to address many current agricultural concerns [4]. The IoT and its accompanying technologies have huge potential to boost agricultural productivity [5] and can play a critical role in reshaping the agriculture industry [6].

The agriculture business has recorded a rise in interest in IoT applications over recent years. The most common use of the IoT in the agriculture industry is monitoring, followed by control and actuation [7,8]. The IoT can help reduce food losses and ensure food safety by monitoring the entire agriculture supply chain. Using the IoT, Big Data, Blockchain, and Machine Learning, smart agriculture may increase productivity, reduce water consumption, and reduce pesticide usage [9,10]. Although academics have undertaken a good deal of research on agricultural IoT applications, the authors of this article still discovered significant research gaps in the existing literature. Therefore, further empirical inquiry is still required in the field of IoT adoption in agriculture. The majority of the existing literature is focused on how the IoT can be integrated with new technologies and applied in different industries [2,9,10,11,12]. The IoT may be used in a wide range of applications in agriculture, but many hurdles must be overcome before it can be widely adopted [7,8]. The adoption of the IoT in the agricultural sector has become the subject of a few empirical studies [13,14,15,16] and literature reviews [17,18], but despite this, only a minimum amount of work has been put into its understanding. Therefore, it may be highlighted that the existing material focuses on technological challenges, rather than economic, social, or cognitive considerations [11].

Although the IoT has piqued the interest of academics and practitioners alike, its study has been restricted to healthcare [19,20], city buildings [21], and the prospect of the IoT in innovative applications and services [22]. To the best of the authors’ knowledge, no previous research has provided an in-depth examination of the various factors influencing the adoption of the IoT in Bangladesh’s agricultural economy. Similarly, no empirical research has specifically discussed the behavioral intention of Bangladeshi farmers towards the adoption of the IoT in agriculture. According to Lozhko [23], a major barrier to the adoption of the IoT in the B2B environment is the high upfront cost of constructing an IoT infrastructure, thereby amplifying the importance of willingness to pay. According to a Boston consulting group [24], the pricing of the IoT is critical to the market’s adoption of the technology; therefore, an understanding of customers’ willingness to pay more may help marketers devise effective pricing strategies. None of the existing studies have researched the willingness to pay for the IoT in the agricultural sector. Furthermore, the study discovered some contradictions in the empirical outcome for some constructs such as trust, performance expectancy, and Behavioral Intention (BI), and those constructs were the significant original model, thereby demanding a reassessment of the new research paradigm in the agricultural sector.

The existing empirical adoption research for the IoT in agriculture employed various research models such as the Behavioral Reasoning Theory [13], the Technology Acceptance Model [14], and the Innovation Diffusion Theory [16]. However, there is a dearth of study on the factors that encourage and discourage the use of new agricultural technologies such as the IoT within a UTAUT framework, both internationally and locally (in Bangladesh). The UTAUT model has been found to be a better fit for the healthcare sector in Bangladesh [19,20]. Consequently, a comprehensive model such as the UTAUT is necessary for experimentation to help derive a comprehensive understanding of current advances in the agricultural industry, which is our contribution to the existing body of knowledge. We consider Bangladesh for this empirical study for two reasons: First, Bangladesh is an agrarian economy, and close to 50% of its total population is employed in the agricultural sector [25]. If we could know the barriers to IoT adoption through investigation, this could further amplify the job positions as well as contribute to the current growing GDP of this country. Second, Bangladesh is lagging behind in its agricultural productivity due to the lack of mechanization in the agricultural production process [26]. This investigation can shed light on knowledge gap on how the agricultural sector can be modernized using IoT technology from the least developed country’s perspectives. However, this study bridges the theoretical deficit by investigating the elements that influence the adoption of the IoT in the agricultural business and also establishes the research topic that will be focused on farmers’ attitudes toward the IoT in agriculture.

Thus, the research seeks to provide an answer to the following question: “What are those things that make Bangladeshi farmers more likely to adopt and pay for the IoT in the agricultural industry?” In response, the present study utilizes the UTAUT 2 theory to analyze the factors impacting the willingness of farmers to adopt and pay for the IoT in agriculture and validates the provided theoretical model. Furthermore, this research will also examine the moderating roles of the facilitating conditions on the association between performance expectancy and willingness to adopt the IoT.

2. Literature Review and Underpinning Theory

2.1. IoT and Its Application in Agriculture

The IoT is a relatively new paradigm that “connects real-world objects to the Internet, allowing objects to collect, process, and communicate data without human intervention” [27]. Users can get smarter services from IoT technology applications due to their utilization of ubiquitous computing and real-time processing. The IoT is a massive network that links people, data, and applications, and enables digital services management and control [28] (

Table 1. Usage of IoT for various agricultural purposes.

Application	Functions
Analysis of yield data	Composite layers are created by condensing numerous years’ worth of yield data [36]
Variable-rate technologies	Allow the application of site-specific agricultural inputs [15]
Field mapping using GPS	Achieve accurate acreage measurement for fields and roads using maps created by farmers [15]
Weather forecasts at the field-level	Measurements made by sensors aid in the forecasting of local weather and precipitation [37].
Autosteer technology	Allows precision farming machinery to operate on autopilot, improving accuracy and production for farmers [15].
Optimization tools for machines	Agriculture inputs are reduced via the use of precise Global Positioning Systems (GPS) and sensors to record agricultural activities [38].
Services measuring productivity	Yield monitoring systems collect data from harvesting trucks and sensors on soil conditions, moisture, and crop yields [15].

). The IoT’s underlying network architecture can connect a wide range of smart devices—ranging from microsensors to huge agricultural tractors—over the Internet. Smart farms, intelligent greenhouses, scientific diseases, pest monitoring, livestock movement monitoring, controlled fertilizer usage, irrigation management, and asset tracking may all benefit from the IoT [29]. The IoT equips farmers with automated technologies and decision-making tools that seamlessly integrate information, services, and goods for higher quality, productivity, and profitability in farming [30]. A multitude of papers illustrating the uses of the IoT in agriculture now exists [29,30,31,32,33,34,35].

It is primarily through the IoT that conventional agriculture can benefit from the increased sensing and monitoring of production processes, improved understanding of specific farming conditions (e.g., weather, environmental conditions, and pest and disease management), concise and remote control of farm operations such as application of fertilizers and pesticides, and autonomous weeding [7,39] (

Table 2. Research on the adoption of IoT in the agricultural industry.

Sources	Research Method/Sample Size/Country	Analysis Tools	Theoretical Framework/ Models	Factors	Limitations
[13]	Empirical /Farmers’ interview/220/ India	PLS-SEM	Behavioral reasoning theory	Attitude, reason for, reason against, and value of openness to change	Personal innovations and risk-taking ability could be used as the moderator
[17]	Literature Review	MICMAC methods	Modified total interpretive structural modeling	Crop management, government initiative, soil quality management, and irrigation management	Results are based on a literature review and are not an empirical research
[14]	Empirical/ Farm owners and managers’ interview/395/ Thailand	SPSS/ Multiple regression	Technology acceptance model	IoT readiness, e-learning, and institutional support perceived usefulness.	Results are based on only small farms and did not consider government support as a construct
[15]	Empirical/ Questionnaire survey/492/USA	SEM-STATA	None	Perceived risk, perceived value, trust, age, farm size	Did not consider contextual factors such as price value and other constructs like trust.
[16]	Empirical/ Face-to-face farmers’ interview/400/Tanzania	Structural Equation Modeling (SEM) (AMOS)	Innovation diffusion theory	Awareness, relative advantages, ease of use, compatibility, visibility	Trust and perceived risk factors could be included to enhance the explanatory power
[18]	Empirical/company experts/35/China	Cross-Impact Matrix Multiplication Applied to Classification (MICMAC) analysis	Interpretive structural modeling	Cost savings, perceived benefit, external pressure, technical knowledge, executive support, trust, technological compatibility, complexity, scale of the enterprise, and government support.	The study is limited to its analysis method as no robustness was tested using other latest statistical methods
Current study	Empirical/ Farmer’s interview/345/ Bangladesh	SEM-AMOS	UTAUT 2	Personal innovativeness, social influence, effort expectancy, willingness to adopt, facilitating condition, performance expectancy, hedonic motivation, price value, government support, and trust.

).

2.2. UTAUT 2 and Hypothesis Formulation

According to the first UTAUT, only extrinsically motivating elements are considered, and a heavy focus is placed on the utilitarian value of the technology. This is represented by the construct of performance expectancy in the sense of utility, which also represents the strongest influencing factor for the intention to use in the UTAUT [40]. This extrinsic motivation component, according to Venkatesh et al. [41], is augmented by the inner motivation component and hedonic motivation, in the UTAUT2. Along with the existing construct (Personal innovativeness, social influence, effort expectancy, willingness to adopt, facilitating condition, performance expectancy, hedonic motivation, price value, except habit) of the UTAUT2, this study experiments with three additional constructs, including trust, government support, and willingness to pay (Figure 1). The constructs are subsequently explained.

2.2.1. Performance Expectancy (PE)

Following the basic concept of PE and its adaptation to the current situation, we defined this factor as the degree to which individuals feel that accessing the IoT will assist them in undertaking a certain activity [42,43,44]. In the agricultural industry, PE is defined as the degree to which a device can assist users in monitoring their daily yield data analysis, GPS field mapping, productivity measurement, and field-level weather forecasts [15]. If farmers’ perceptions of improved harvest management, easier access to agricultural services, more efficient use of resources, and higher overall productivity enhance their PE for connected crop care equipment, then their willingness to use and pay for these devices will rise accordingly. The BI to utilize healthcare technology has been strongly associated with PE in previous studies [42,45,46], and according to Gu and Liu [47], PE is positively associated with the willingness to pay for Q&A platforms. Therefore, the higher the performance expectancy of the IoT technology, the higher the chances of farmers’ willingness to pay for it. Thus, the following hypotheses are proposed:

Hypothesis 1 (H1).

PE is significantly and positively related to the willingness to adopt the IoT.

Hypothesis 2 (H2).

PE is significantly and positively related to the willingness to pay for the IoT.

2.2.2. Government Support

The IoT business, particularly in the agriculture sector, relies heavily on government backing. The government’s willingness to engage in the business may help the agricultural business flourish via the enactment of policies that benefit both investors and service providers. According to research by Goo and Heo [48], government involvement has a favorable influence on the growth of Fintech owing to the lowering of uncertainties. According to research conducted by Chong and Ooi [49] on the acceptance of RosettaNet in Malaysia, the Malaysian government actively encourages the use of the standard by offering grants and tax exemptions. Following Marakarkandy et al.’s [50] findings, government support for new technologies has a favorable correlation with their adoption. These rules and regulations assist both the agricultural business and its infrastructure, as well as the government’s transformational role towards automation via the expansion of Internet networks, reduction of VAT on accessories, and enhancement of technical capabilities, amongst other things.

Hypothesis 3 (H3).

Government support has a positive relationship with the willingness to adopt the IoT in the agricultural industry.

2.2.3. Facilitating Conditions

FC is the user’s belief that support and infrastructure are available to assist them in the usage of their desired technology [41]. Technical and infrastructure support for system utilization is usually classified under FC, as it has an impact on both user intent and actual use [40,41]. Technology or organizational assistance is required by farmers to employ crop care equipment. The government should encourage the development of the physical, network, and electrical infrastructure, as FC makes it possible for users to have a better understanding of the resources and facilities available to them, such as their talents, technical expertise, and the help and direction of IT experts [44,51]. Boontarig et al. [52] discovered that FCs have a favorable effect on behavioral intention in their investigation. FC is perceived as the most important factor in determining a consumer’s intention to utilize mHealth [19,53]. As a result, the following hypothesis is proposed:

Hypothesis 4 (H4).

FC is positively related to the willingness to adopt the IoT.

2.2.4. Social Influence

As highlighted in several studies, SI has a multifaceted role in the acceptance of new technology [54,55]. To put it another way, SI is the degree to which people believe that important individuals agree with their actions or technological choices [19,40,44]. Specifically, SI has demonstrated an excellent predictive power in the context of health-related technology use [55,56,57], wireless devices, smartphones [58], and mobile diet applications [56]. In the agricultural context, when a farmer receives a direct or indirect suggestion of utilizing a smart agricultural technology from other farmers, experts, peers, or those whom they follow, they will be encouraged to adopt that technology. The following hypothesis is thereby proposed:

Hypothesis 5 (H5).

SI has a positive influence on the willingness to adopt the IoT.

2.2.5. Hedonic Motivation

Hedonic Motivation (HM) has been defined as the degree to which the usage of new technology results in satisfaction or pleasure, and it has been identified as playing a critical role in the adoption of new technology and its use [41]. Joy, fun, entertainment, playfulness, and other intangible benefits are all included in HM (usefulness, efficiency, performance, etc.). To encourage clients to accept new technology, it is important to show them how the use of the new technology will provide them joy and happiness [59,60]. If a person thinks that the usage of mobile applications for healthcare services is amusing, interesting, and pleasant, he or she is more likely to use the application [19,61]. Consequently, a customer’s desire for new technologies such as the IoT employing GPS-based mapping or yield control systems could infuse fun or enjoyment in its usage, which in turn increases the user’s behavioral intention towards the IoT. As a result of this, we may make the following assertion:

Hypothesis 6 (H6).

HM is positively related to the willingness to adopt the IoT.

2.2.6. Effort Expectancy

People’s willingness to utilize a new technology is influenced by its ease of usage [42]. EE refers to “the ease with which a system may be used” [40]. Farmers who utilize IoT devices are more likely to perceive technology as good and useful when EE is present. As a result, the amount of energy required to run the system increases [62]. Innovation in technologies that are quick and easy to implement with minimal effort is often viewed as a sign of progress by users [63,64]. Researchers [45,51,65] discovered that EE was an excellent indicator of behavioral intention. Thus, because farmers, in many cases, are less educated and less technology conscious, they are likely to adopt the IoT if they perceive or discover that it is not a complex system to operate, or requires less effort to learn. Thus, we formulated the following hypothesis:

Hypothesis 7 (H7).

Effort expectancy is significantly related to the willingness to adopt the IoT.

2.2.7. Trust

Trust is the reliance or confidence in a particular service. The IoT will be more widely adopted if people are confident in its ability to deliver on its promises, and in the company that provides it. The IoT application’s capacity to function during an emergency may be hampered if users harbor mistrust in the service provider. Although most studies have indicated a favorable association between trust and BI, some have revealed otherwise [63,66,67]. According to Alalwan et al. [63] and Chao [68], trust is a key factor in the adoption of mobile technology. Similarly, the researchers [65,66] discovered trust to have a significant impact on students’ BIs toward the utilization of mobile learning; however, Kabra et al. [66] revealed the non-existence of an association between trust and BI. According to Jiang et al. [69], a positive association exists between technological trust and the willingness to purchase and adopt autonomous cars. Augusto et al. [70] investigated the elements influencing the willingness of customers to subscribe to premium steaming services and found trust to be a significant factor in this regard. As a result, the following hypotheses are suggested:

Hypothesis 8 (H8).

Trust has a positive influence on the willingness to adopt the IoT.

Hypothesis 9 (H9).

Trust is significantly and positively linked to the willingness to pay for the IoT.

2.2.8. Price Value

Price value is another component that has been included in the UTAUT 2 [41]. Price Value (PV) is defined as a consumer’s cognitive tradeoff between the perceived system value and the cost of acquiring or utilizing a new technology [41]. With the usage of new technology, end-users are constantly comparing the cost incurred with the resulting savings that they might derive from the new technology [59,71,72]. With the use of information technology such as the IoT, the agricultural production process may benefit from the delivery of crop care information services at a more affordable rate as compared to the traditional method of physical inspection for pest control or irrigation. By assuming that a product’s benefits surpass its expenses, its pricing is considered to be positive [41]. However, there is evidence to suggest that price value is an important factor in consumer behavior in the adoption of the IoT. As a result, researchers proposed the following hypothesis:

Hypothesis 10 (H10).

PV is positively related to the willingness to adopt the IoT.

2.2.9. Personal Innovativeness

Personal innovativeness was defined by Lu et al. [73] as the willingness of an individual to try out new technologies. Adoption of new technologies is mostly driven by a person’s willingness to tolerate the existence of the technology. According to this study, user innovativeness is characterized as the readiness to experiment with IoT services and eagerness to explore new technologies. Earlier studies have revealed a correlation between user inventiveness and technological acceptance [74,75,76,77,78], and similarly, Leckie et al. [79] demonstrated the impact of service innovation in the generation of total service value appraisal, customer engagement, and loyalty. O’Cass and Carlson [80] discovered that perceived website innovation predicts a customer’s trust. The perceived innovativeness of a website is related to the notions of originality and utility—according to the authors—and these underlying characteristics encourage people to have faith in a website. As a result, farmers perceiving the IoT as a cutting-edge technology tend to be more willing and able to use it. Based on the foregoing argument, the following hypotheses are advanced:

Hypothesis 11 (H11).

Perceived innovativeness has a positive relationship with trust.

Hypothesis 12 (H12).

PI has a positive influence on the willingness to adopt the IoT.

2.2.10. Willingness to Adopt

The model’s main idea is hinged around people’s willingness to try new things, which is referred to as Behavioral Intention. Many researchers focus on the intention to utilize technology rather than its actual utilization. A close examination of people’s (and managers’) willingness to pay for the IoT reveals their ability to price it when the technology is launched. The willingness of consumers to pay more for a product or service is influenced by their attitudes toward, and intentions to adopt, newer technology such as green information and communication technologies, renewable heating systems, and electric automobiles [81,82,83]. Therefore, the following hypothesis is postulated:

Hypothesis 13 (H13).

Willingness to adopt has a positive relationship with willingness to pay.

2.2.11. Willingness to Pay

Willingness to pay (WTP) is defined as ‘the highest price that a consumer is willing to pay for a product or service’ [84]. Managers must realize that because the utmost number of consumers are willing to pay, they must be ready to create an effective pricing strategy that maximizes profits while also satisfying consumers’ needs and expectations [85]. Individuals’ buying intents and WTP are influenced by product, service, and technology values. While the ‘desire to pay more’ measures the influence of monetary units on values, purchase intention measures consumers’ intention to purchase a service or technology [84]. Although consumers could expect an amazing experience from the use of IoT services, it is crucial to grasp the value–price tradeoff in the market. The value–price tradeoff can be better grasped when included in the ‘desire to pay more’ variables.

2.2.12. The Moderation Effect of Facilitating Condition

As earlier stated, facilitating condition refers to the necessary facility such as knowledge, technological awareness, and technical and infrastructural support. Various scholars have applied a variety of support facilities as a moderation tool. For instance, Li and Zhao [86] studied the connected classroom climate as a moderating effect in UTAUT constructs such as performance expectancy and behavioral intention and discovered a significant relationship. In their research, Abubakar and Ahmad [87] hypothesized that technological awareness moderates the link between performance expectancy and behavioral intention, while others directly applied the FC as a moderator. For example, Humida et al. [88] assessed the moderation effect of facilitating conditions between the perceived usefulness (same as performance expectancy) and behavioral intention towards the e-learning system in Bangladesh, albeit its infrastructural deficiency among the Bangladeshi students. They concluded that no significant moderation of facilitating conditions was observed, owing to the respondents being students who were knowledgeable about technology. This research aims to retest the same proposition in the context of farmers who are presumed to have deficit knowledge and access to technology. Thus, the higher the facilitating condition, the greater the possibility of predicting the willingness to adopt the IoT through the performance expectancy of technology.

Hypothesis 14 (H14).

Facilitating condition positively and significantly moderates the association between performance expectancy and willingness to adopt the IoT.

3. Research Design

This is a quantitative study that adopts a cross-sectional survey design. Data were obtained from respondents in different northern districts of Bangladesh using a convenience sampling approach based on self-administered questionnaires.

3.1. Population, Sample, and Data Collection Procedure

Farmers residing in Bangladesh’s rural areas make up the population, and the sample in this study comprises premium fruit growers from three northern districts of the country. The research’s scope and objectives were established with subject matter experts (5 people who undertook the research in the same area) before sampling was carried out. Subsequently, 50 people from five villages partook in a pilot survey to assess the reliability and validity of the constructs. For the final survey, questionnaires were administered to Bengali farmers who responded from Union Parishad rural farms, as well as from rural marketplaces and rural homes. The study was conducted among the farmers who were involved in the cultivation of premium fruits (strawberry, orange, guava, jujube, dragon fruits, etc.).

The thumb rule was used to verify that primary data were collected and that adequate sample size was obtained [89]. The maximum number of items in the latent variable is 32, indicating that the sample size must be 10 times more, i.e., 320. To collect data, a standardized questionnaire was administered in the native language of the respondents. Only farmers who utilized technology for farming purposes, and not just IoT technology, were the survey’s primary target. Before the questionnaires were filled by the farmers, they were shown videos and demonstrations of IoT technology-based devices in their language. The convenience sample approach was implemented in this study, which was undertaken in 25 villages in the Bangladeshi districts of Rajshahi, Natore, and Chapai Nababganj, where farmers already adopted the use of various farming technologies. These districts were selected based on the possession of premium fruit producers who supplied a majority of the premium fruits to the Bangladeshi markets.

The survey was conducted among farmers who were either owners or were managers of farms in the Union Parishad, as well as at rural hospitals and rural markets, and among rural families that were visited. The farmers selected for this study were a good fit because they were well-versed in farming technologies and had substantial acreage and income to back up their opinions. The total number of farmer-respondents was 395, of which 345 were deemed suitable for analysis, resulting in an overall response rate of 87.34%.

Table 3. Respondents profile.

Demographics	Classification	Frequency	Percentages
Gender (out of 345)	Male	290	84
	Female	55	16
Age (Years)	<25	60	17
	25–50	192	56
	>50	93	27
Types of Premium Fruits Yield	Strawberry	25	7
	Orange	20	6
	Guava	97	28
	Jujube (Kul)	140	41
	Dragon	25	7
	Others	38	11
Farm Size (acres)	<5	85	25
	5–10	184	53
	>10	76	22

offers a breakdown of the farmers’ demographic characteristics. For elicitation, pilot survey, and final data collection, the survey questions were translated into Bengali.

3.2. Instruments of the Study

Each variable was measured using a five-point Likert scale (from strongly agree to strongly disagree), and scales were developed for the questionnaire constructs based on the existing literature (Appendix A). The HM, PV, PE, and EE were measured using a scale validated by Alam et al. [19]. Additionally, ‘willingness to adopt’ and facilitating conditions were measured using a scale developed by Sheel and Nath [90], while ‘willingness to pay’ was measured using a scale developed by Pihlström and Brush [91]. It is worth mentioning that this study considered ‘willingness to adopt’ as synonymous with the ‘behavioral intention’. Government support and trust were measured using a single item adapted from Chong et al. [49] and Chao [68], respectively. The construct of social influence and personal innovativeness has been developed in accordance with Alkawsi et al. [92].

3.3. Data Analysis

To better understand the characteristics of respondents, a demographic analysis was carried out, with the descriptive statistics of the sample being gathered afterward. The presented hypotheses were tested using Structural Equation Modeling (SEM), which was deemed appropriate for this study owing to its ability to confirm the suitability of UTAUT-2 in the analysis of IoT adoption patterns. As an added benefit, SEM corrects measurement inaccuracies in measuring objects [93]. All of the interdependence links were examined in a single SEM study [93], and as a result, the SEM tended to generate findings that are devoid of errors. Covariance-based SEM was used to investigate the association between variables in this research because it was designated as a confirmatory study [94]. The data analysis was conducted via IBM-SPSS 26 and AMOS 26.

4. Results of the Study

4.1. Demographic Profile

A high proportion (84%) of the responders was male. Additionally, 41% of the responders were Jujube farmers, while 28% were guava fruit producers. About 53% of respondents had 5–10 acres of land, while 25% possessed less than 5 acres of land for farming purposes. The majority of people who partook in the study were aged between 25 and 50, succeeded by those above 50 (25.3%). This implies that the research respondents were mature enough to comment on the pertinent issues in this study and that medium to largescale farmers who profited from selling premium fruits was maximum (Table 3).

4.2. Measurement Model

This study utilized Anderson and Gerbing’s [95] two-step method for data analysis and deployed a measurement model and an SEM for each step. At the onset, an Exploratory Factor Analysis (EFA) was conducted to verify the independence and uniqueness of the latent variables, with the KMO measure (0.973) and Bartlett’s test (p = 0.00) highlighting the appropriateness of our data for factor analysis.

4.2.1. Common Method Bias (CMB) Test

In self-single-source and cross-sectional designs of social science research, the problem of Common Method Bias (CMB) is common. Harman’s single-factor scale was investigated using 11 factors (PE, TT, EE, PI, SI, FC, PV, HM, GS, WTA, and WTP) and 32 scale items, as recommended by Harman [96] and Podsakoff et al. [97]. However, no one factor emerged as the first construct to explain the 31.915% of variation, which is less than the 50% cut-off figure advocated by Podsakoff et al. [97]. As a result, no evidence of the CMB problem was detected in the dataset.

4.2.2. Data Normality and Multicollinearity

The results for normalcy were satisfactory, owing to the absence of a significant deviation from normality. The Kurtosis and skewness values were also below 10 and 3, respectively [98] (

Table 4. Factor loadings and normality and multicollinearity statistics.

Construct	Items	Std. Beta	Mean	SD	Skew	Kurt	CA	VIF
								Tolerance
WTP	WTP1	0.985	4.060	0.792	−1.552	1.632	0.875	0.865	1.155
	WTP2	0.790
HM	HM1	0.818	2.058	0.826	0.887	0.344	0.737	0.937	1.067
	HM2	0.706
	HM3	0.630
PV	PV1	0.630	3.546	0.840	−0.376	−0.439	0.750	0.840	1.190
	PV2	0.783
	PV3	0.724
SI	SI1	0.834	3.555	0.891	−0.462	−0.602	0.802	0.660	1.515
	SI2	0.609
	SI3	0.844
GS	GS1	0.719	3.336	0.951	−0.376	−0.844	0.778	0.943	1.061
	GS2	0.871
	GS3	0.619
EE	EE1	0.744	3.755	0.832	−0.777	0.122	0.786	0.819	1.221
	EE2	0.645
	EE3	0.836
PI	PI1	0.677	3.752	0.800	−0.679	0.299	0.834	0.687	1.456
	PI 2	0.936
	PI3	0.814
TT	TT1	0.829	4.357	0.682	−1.676	3.893	0.855	0.818	1.222
	TT2	0.868
	TT3	0.755
WTA	WTA1	0.745	3.536	0.919	−0.666	−0.380	0.823	--	--
	WTA2	0.792
	WTA3	0.816
PE	PE1	0.811	2.645	1.058	0.443	−0.876	0.809	0.888	1.126
	PE2	0.675
	PE3	0.812
FC	FC1	0.741	2.706	0.828	0.184	−0.450	0.762	0.877	1.140
	FC2	0.686
	FC3	0.729

Note: WTP = Willingness to Pay, HM = Hedonic Motivation, PV = Price Value, SI = Social Influence, GS = Government Support, EE = Effort Expectancy, PI = Personal Innovativeness, TT = Trust, WTA = Willingness to Adopt, PE = Performance Expectancy, FC= Facilitating Condition.

). To examine the existence of multicollinearity among the independent variables, two common methods were utilized, as specified by Kleinbaum et al. [99]—the VIF (Variance Inflation Factor) and Tolerance test. As presented in Table 3, the multicollinearity statistics results indicated that the tolerance value for all the constructs exceeds 0.1, while all VIF values were lower than 10. Thus, the analysis results indicated that multicollinearity problems do not exist for independent variables.

4.2.3. Reliability and Validity

The measurement model was assessed to determine the reliability and validity of the concept, while the Cronbach Alpha (CA) score was used to assess data dependability (see

Table 5. Reliability and validity measures.

Variables	CR	AVE	MSV	MaxR(H)	WTP	HM	PV	SI	GS	EE	PI	Trust	WTA	PE	FC
WTP	0.886	0.797	0.156	0.972	0.893
HM	0.764	0.521	0.113	0.807	−0.027	0.722
PV	0.757	0.511	0.158	0.802	0.200	−0.079	0.715
SI	0.811	0.593	0.276	0.843	0.239	−0.018	0.201	0.77
GS	0.785	0.553	0.044	0.829	0.020	−0.138	0.174	0.154	0.744
EE	0.788	0.557	0.260	0.811	0.145	−0.059	0.106	0.510	0.045	0.746
PI	0.854	0.666	0.276	0.908	0.293	−0.065	0.244	0.525	0.069	0.356	0.816
TT	0.859	0.670	0.113	0.868	0.235	−0.337	0.227	0.164	0.209	0.111	0.108	0.819
WTA	0.828	0.616	0.156	0.831	0.396	0.028	0.303	0.350	0.185	0.328	0.322	0.172	0.785
PE	0.812	0.591	0.053	0.825	0.069	0.104	0.107	0.194	−0.073	0.186	0.142	−0.223	0.230	0.769
FC	0.762	0.517	0.158	0.812	−0.039	0.06	−0.397	−0.162	−0.068	−0.067	−0.073	−0.168	−0.028	−0.136	0.719

Note: In the Table, bold elements, the square root of AVE, WTP = Willingness to Pay, HM = Hedonic Motivation, PV = Price Value, SI = Social Influence, GS = Government Support, EE = Effort Expectancy, PI = Personal Innovativeness, TT = Trust, WTA = Willingness to Adopt, PE = Performance Expectancy, FC = Facilitating Condition.

). For each construct, reliability coefficients representing the willingness of usage exceeded 0.6 (between 0.609 and 0.985), adequately highlighting the reliability suggested by Nunnally and Bernstein [100]. The Average Variance Extracted (AVE) and Composite Reliability (CR) were deployed to investigate the construct validity. AVE values exceeding 0.5 highlight the convergent validity for all constructs in Table 4 [101,102], owing to the greater value of the AVE’s square root compared to other components off-diagonal. Therefore, the study results imply discriminant validity [99].

A CR score greater than or equal to 0.7 indicates a good model and is generally deemed appropriate for use in preliminary research [103]. According to the previously indicated criteria, the study’s constructions are declared statistically acceptable. Owing to its superiority over Fornell–Larcker in numerous conditions, this research additionally assessed the HTMT value for robustness [104], with HTMT values exceeding 0.85/0.90, confirming discriminant validity [104]. The cut-off value was attained in this investigation (

Table 6. HTMT and R².

WTP	HM	PV	SI	GS	EE	PI	Trust	WTA	PE	FC	R2
WTP											0.18
HM	0.019
PV	0.189	0.074
SI	0.244	0.013	0.235
GS	0.030	0.123	0.190	0.165
EE	0.166	0.043	0.106	0.484	0.023
PI	0.338	0.049	0.226	0.599	0.082	0.403
Trust	0.212	0.316	0.249	0.172	0.185	0.091	0.09				0.01
WTA	0.406	0.055	0.318	0.366	0.165	0.313	0.357	0.173			0.23
PE	0.077	0.100	0.124	0.226	0.017	0.201	0.175	0.212	0.217
FC	0.012	0.051	0.396	0.213	0.102	0.091	0.074	0.164	0.025	0.156

Note: WTP = Willingness to Pay, HM = Hedonic Motivation, PV = Price Value, SI = Social Influence, GS = Government Support, EE = Effort Expectancy, PI = Personal Innovativeness, TT = Trust, WTA = Willingness to Adopt, PE = Performance Expectancy, FC = Facilitating Condition.

), and the discriminant validity was also evaluated using the MSV and MaxR (H) models for robustness purposes. To confirm discriminant validity, the MaxR(H) value should not exceed 80 discriminant validity, according to Hancock and Mueller [105], and the square root of a construct’s AVE must exceed the correlations between the components [101]. From the data presented in Table 5, the AVEs, CRs, and MaxR(H)s all exceed their respective cut-off values, indicating the genuineness of the construct. The MSV values for all constructions were discovered to be lower than the AVE values when compared with the AVE. Thus, the appropriate reliability and validity of constructs were confirmed in order to proceed to the next stages of the analysis.

Figure 2 and

Table 7. Structural model and hypothesis testing result.

Hypothesis	STD Beta	STD Error	t-Values	p-Values	Significance (p < 0.05)
H1: PE → WTA	0.190	0.039	3.736 ***	0.000	Supported
H2: PE → WTP	0.016	0.038	0.353	0.724	Not Supported
H3: GS → WTA	0.124	0.048	2.481 **	0.013	Supported
H4: FC → WTA	0.126	0.055	2.428 **	0.015	Supported
H5: SI → WTA	0.159	0.050	3.206 ***	0.001	Supported
H6: HM → WTA	0.105	0.048	2.039 **	0.041	Supported
H7: EE → WTA	0.173	0.048	3.379 ***	0.000	Supported
H8: TT → WTA	0.139	0.066	2.822 ***	0.005	Supported
H9: TT → WTP	0.175	0.065	3.843 ***	0.000	Supported
H10: PV → WTA	0.224	0.059	4.220 ***	0.000	Supported
H11: PI → TT	0.121	0.046	2.469 **	0.014	Supported
H12: PI → WTA	0.145	0.060	3.070 ***	0.002	Supported
H13: WTA → WTP	0.354	0.053	7.101 ***	0.000	Supported
H14: PE × FC → WTA					Supported

Note: *** Significant at the 0.01 level (2-tailed). ** Significant at the 0.05 level (2-tailed). S = Significant, NS = Not Significant. WTP = Willingness to Pay, HM = Hedonic Motivation, PV = Price Value, SI = Social Influence, GS = Government Support, EE = Effort Expectancy, PI = Personal Innovativeness, TT = Trust, WTA = Willingness to Adopt, PE = Performance Expectancy, FC = Facilitating Condition.

depict the structural model of our study. Sequel to the CFA test of the measurement model, the structural model’s validation assessed the proposed model’s goodness of fit indices. The SEM values (X²/df = 2.221) revealed that the data were well fitted, and to be sure, RMSEA cut-off values lower than 0.08 were met, because the actual value was 0.068 [106]. Several fit indices (CFI, GFI, IFI, and TLI) satisfied the criteria of 0.9 or higher [107].

4.3. Structural Model and Hypotheses Results

To examine the structural model and the hypotheses, the significance of the path coefficient, t-value, p-value, and R² (the variance explained) were employed in this study. The critical value for the two tests was 1.96 at a 5% level of significance, and the results of this study revealed that the R² values for ‘willingness to adopt’ and ‘willingness to pay’ were 0.23 and 0.18, respectively, both of which are considered to be moderate [108].

Because the measurement model passed the CFA test, the structural model evaluation examined the goodness of fit indicators of the proposed model, and it was observed from the results of the SEM that the conceptual framework had a good data fit (X²/df = 2.581, GFI = 0.924, CFI = 0.933, IFI = 0.928, NFI = 0.918). According to the Root Mean Square Error Approximation (RMSEA), cut-off values lower than 0.08 support the results [106], and as for the other fit indices, they all attained or exceeded the requirement of values close to 0.9 and above [107]. Accordingly, the results of this investigation reveal a strong fit to the model, as evidenced by good-fit markers.

The outcome demonstrated that performance expectancy (β = 0.190; t = 3.736; p < 0.01), government supports (β = 0.124; t = 2.481; p < 0.05), facilitating condition (β = 0.126; t = 2.428; p < 0.05), social influence (β = 0.159; t = 3.206; p < 0.01), hedonic motivation (β = 0.105; t = 2.039; p < 0.05), effort expectancy (β = 0.173; t = 3.379; p < 0.01), trust (β = 0.175; t = 2.822; p < 0.01), price value (β = 0.224; t = 4.220; p < 0.01), and personal innovativeness (β = 0.121; t = 2.469; p < 0.01) all played influencing roles in the willingness to adopt the IoT in the agricultural industry. Additionally, the results in Table 3 depict that trust (β = 0.175; t = 3.843; p < 0.01) and willingness to accept (β = 0.354; t = 7.101; p < 0.01) influence willingness to pay, while performance expectancy (β = 0.016; t = 0.353; p > 0.05) revealed no impact on willingness to pay. Thus, the results support H1 and H₃–H₁₃, and reject H2.

The study assessed the moderation effect based on the interaction effects of contracts, and the outcome (Figure 3 and Table 7) revealed that the facilitating condition (β = 0.127, t = 2.469, p < 0.05) positively and significantly moderates the relationship between performance expectancy and willingness to adopt the IoT among rural agri-entrepreneurs, thereby supporting hypothesis 14 empirically.

5. Discussion

The study investigated the antecedents of the willingness of rural farmers in the agricultural sector in Bangladesh to pay for and use the IoT. Based on the UTAUT 2 model, the study empirically tested the factors and revealed that all the factors are well suited, as all the antecedents of ‘willingness to pay’ are significant, except for the performance expectancy.

The study assumed (H1) that performance expectancy significantly affects the willingness to use the IoT in an agricultural setting, thereby supporting previous studies [45,46]. This study reiterates that performance expectancy positively influences the willingness to use the IoT, and it is the second most predictor in the model. This implies that the greater performance expectancy or perceived benefit of using the IoT in agricultural production activities will result in a correspondingly greater chance of accepting the technology. On the other hand, results from the study highlighted that the performance expectancy does not affect the willingness to pay for IoT in the Agricultural industry (H2). This is the exact opposite of the previously conducted study by Gu and Liu [47], where it was stated that, irrespective of the level of benefit provided by a system, it still will not correspond to the amount that is charged for it.

As expected in H3, government support has a positive effect on the willingness to utilize the IoT. This outcome is consistent with past research [19] by Marakarkandy et al. [50] and Tan and Teo [109], and inconsistent with the study conducted by Setiawan et al. [110], where it was stated that government policy support can greatly make a difference in the acceptance of the IoT in the agricultural industry. Hypothesis 4 predicted that facilitating condition significantly influences the willingness to adopt the IoT, and the study outcome confirms this prediction. The obtained result reaffirms previous research works [20,53] and contradicts other studies [19]. Because the IoT is the latest technology, it requires infrastructure, which is still difficult for the majority of farmers to avail. Therefore, the higher the facilitating condition, the greater the chances of IoT adoption in agricultural entrepreneurship.

External influences such as social influence by peers and relatives have a great influence on system adoption (H5), and present research works have proven it again. People’s environment acts as a referral and supports utilizing a system. This result is in line with past literature [19,56], which highlighted that a higher social influence in favor of using the IoT will result in a greater possibility of utilizing it for agriculture. Similarly, it was assumed (in H6) that hedonic motivation influences the willingness to adopt an IoT system. This study supports the hypothesis significantly and positively. The result is in accordance with previous research works [19,55], implying that the degree of fun enjoyed by a user favorably contributes to their choice of a system.

Another factor contributing to the willingness of people to utilize the IoT is the level of effort they feel it will take them to use the technology, which is solidly corroborated in this study. This is in tandem with previous works of literature [19,45,63,64], and indicates that the greater the ease of use of the IoT technology, the greater the possibility of its use from the user end. Likewise, trust in system security was postulated to influence the use of IoT in agriculture (H8), which is consistent with the obtained results of this study. The outcome of this result also supports past results [6,19,66,68,111,112] and negates studies by Alalwan et al. [63] and Khalilzadeh et al. [67]. Thus, greater consumer trust in the security of the IoT during usage will instigate its further use, particularly in rural agricultural settings (H9). Nonetheless, trust has also been proven to be a predictor of the willingness of rural agri-entrepreneurs to pay for the IoT. This result corroborates the previously conducted study by Augusto et al. [70].

As predicted, price value influences the willingness to adopt the IoT, which satisfies earlier research [59,71,72]. The result of this study revealed that price value contracts had the greatest influence on the willingness to use the IoT. Similarly, the relationship between personal innovativeness and trust was hypothesized, which turned out to be significant in this study. The study conducted by Cass and Carlson [80] was observed to be consistent with this result, implying that the innovativeness of agri-entrepreneurs helps to reinforce the trust of IoT users. In addition, this study revealed that creative people are more likely to utilize the new technology, supporting Hypothesis 12. This is also in line with previous studies [78,113], which established the link between personal innovative attributes and the adoption of technologies. Thus, the higher the innovativeness of an individual, the greater the intention of agri-entrepreneurs to utilize the IoT.

The greatest antecedent of willingness to pay is the willingness to adopt the IoT, which supports Hypothesis 13. The significant and positive relationship between the willingness to adopt and the willingness to pay satisfies previous results [81,83], implying that the willingness to adopt the IoT greatly influences the willingness to pay for the technology. With respect to the moderation role, it was assumed that facilitating condition significantly moderates the association between performance expectancy and willingness to adopt the IoT. The present study confirms H14, which postulates that the moderating role of facilitating conditions is consistent with the previously conducted study by Humida et al. [88]. Specifically, higher performance expectancy results in a greater willingness to accept the IoT if the user is equipped with a greater facilitating condition such as an Internet facility, instrumental price support, etc.

6. Conclusions

The objective of this paper was to investigate the motivation of people in the agricultural industry toward the payment and use of the IoT. The study confirms that effort expectancy, performance expectancy, facilitating condition, hedonic motivation, government support, price value, personal innovativeness, and trust influence the willingness to adopt the IoT. Predictors such as trust and willingness towards the adoption of the IoT affect the willingness to pay, while the construct of performance expectancy failed to similar effect.

6.1. Contribution and Implications

6.1.1. Theoretical Contribution

Theoretically, this research has contributed significantly in three ways, in terms of a new context, a new model with new constructs, and new results. First, this research provides new quantitative knowledge of the antecedents of willingness to adopt the IoT in a developing country. This research has covered suggestions (adoption research in the innovation and diffusion stages) highlighted in the conducted study of Jayashankar et al. [15], particularly in the arena of IoT adoption in agricultural industry research, in a developing country such as Bangladesh. So far, from the researcher’s knowledge, no empirical research has been identified in the area of IoT adoption in the agricultural industry in Bangladesh. Second, several studies have utilized the UTAUT model from several industry perspectives, but the use of this model in IoT adoption in the agricultural section is still lacking. Jayashankar et al. [15] did not employ any theoretical model, as opposed to Pillai and Sivathanu [13], who utilized the Behavioral Reasoning Theory in the Indian context. The present study contributes theoretically to the application of the UTAUT 2 model in an agricultural setting.

Third, this research added three new constructs to the existing UTAUT 2 model to fit into the agricultural industry. For instance, the willingness to adopt a particular technology is deemed to be insignificant, except in a bid to discover how much they are willing to pay for it, as the adoption of the IoT is a private investment for agri-entrepreneurs. In many cases, a missing link exists between the intention to use and the willingness to pay for IoT technology, and in contrast, research on the willingness to pay for the technology has mostly been conducted in other industries. To the best of the authors’ knowledge, no research has taken the initiative to infuse the willingness to pay into the UTAUT2 model to explain the use of the IoT in developing countries. Thus, the present study is a novel attempt to bridge this gap and form a comprehensive model for information system use.

Fourth, with the addition of trust and government support, as well as the basic components of the UTAUT2, this empirical study extends beyond the path relationship recommended by Venkatesh et al. [32] in the basic UTAUT2 model but was ignored in earlier literature works. This study proposed new causal path relationships between the main determinants of WTA and WTP (GS → WTA, TT → WTP, PE → WTP, WTA → WTP, and TT → WTA). On the other hand, PE → WTP was deemed insignificant in the context of IoT adoption. The facilitating condition was proposed and established as a moderating variable in this empirical research, which is rare, particularly in agricultural industry research.

The integration of four additional components with the UTAUT2 is scarce in the literature, especially in the context of a resource-constrained nation like Bangladesh. According to researchers, this study outlines the most important drivers in a developing nation context and recorded reliable results that could be applied to the people of this region. With the use of SEM analysis, this empirical study attempted to bridge the theoretical research gap by testing and verifying a new extended model.

6.1.2. Practical Implication

The outcomes of this research have policy consequences. First and foremost, the empirical findings of this study have demonstrated that facilitating conditions significantly influence the willingness to adopt and pay for the IoT directly and indirectly. In addition, government backing has been proven to be an important predictor of IoT adoption, and as a result, the government should invest in the development of suitable and necessary infrastructure to enable IoT activities. This might be accomplished by implementing a dependable backend system and extending the network to accommodate a greater number of users within the network. In the same way, the government should enact laws and rules to assist IoT businesses to function effectively. Citizens will realize that the IoT is a technology approved by the government that will be used to transform agricultural production in the future.

Second, the price value was revealed to be statistically significant, implying that consumers compare the costs and benefits of utilizing an IoT system. As a result, the monetary cost of procuring IoT instruments for citizens should be minimized or eliminated through government subsidies on the import of IoT components and systems. This is because most rural farmers are perceived to be poor, and providing high-cost IoT services to them might deter their uptake and utilization of IoT technologies.

Third, policymakers should ensure that the IoT system is simple to operate. System designers and developers should create systems that are easier to use and require less effort to operate, and the provision of clear directions, navigational stages, and a “Help” button might help. By including the general public in the development of the IoT, system designers can make it even more user-friendly. System designers, on the other hand, should work to increase their utility. This might be accomplished by making the necessary information available on social media or other forms of mass media (radio, television, newspaper, etc.). As a result, leaders of rural communities, as well as reputed farmers and other prominent figures can serve as change agents. The Union Parishad, a local government agency, can play a supportive role in promoting the system’s use and acceptance.

Fourth, the results have revealed that trust is a predictor of willingness to use and pay, and that building trust on security issues requires proper system design by the designer, along with an awareness campaign. Therefore, policymakers should focus more on educating the public on the aforementioned issues. The adoption of IoT services may be improved if these issues are addressed in the design process. Broadcasting personalized text messages, training, and holding meetings in rural areas can assist in spreading the word about the effective usage of the IoT.

6.2. Limitations and Further Research Direction

Although the primary goals of this research were attained, some study limitations still exist; therefore, the outcomes of this study should be interpreted with caution, as the information was only gathered once. This suggests that rural farmers may have learned more expertise over time, which might have an impact on the utilization of the data. This opens the door to even more in-depth studies, such as longitudinal ones, as such research may yield more accurate data that better represent the evolution of farming practices throughout the years. As the IoT in agriculture is still a new field, we encourage a greater investigation into how contextual variables specific to digital agriculture influence technology adoption. For example, academics may examine how consolidation among agricultural enterprises and value co-creation through crowdsourcing could affect the adoption of the IoT. This study highlighted the willingness to utilize the IoT from the perspective of customers. It did not consider the actual usage of IoT technology, and this could be examined in future research to acquire insights from an existing client’s context. Additionally, future studies could address the intention–behavior gap in IoT use in the agricultural sector by integrating the mediating variables. As the present study only utilizes facilitating conditions as moderating variables, future studies could experiment with some other moderating variables such as perceived risk, technology readiness, etc.