Neural Network Analysis of Technology Adoption Intentions Among Womenpreneurs in Small and Medium Enterprises ^†

Abstract

Financial literacy has grown significantly in recent years, leading to increased accountability among individuals in managing their spending, investments, and financial planning. To effectively develop new technologies, how potential users respond to them beforehand must be understood. In this study, we developed a model based on the unified theory of acceptance and use of technology (UTAUT), encompassing five primary components: performance expectations, social influence, facilitating conditions, satisfaction, and continuation intentions. A total of 250 participants’ data were analyzed using an artificial neural network (ANN) to evaluate the integrated model. The results showed variables affecting lower-middle-class users of a digital financial literacy application’s acceptance of new technology.

1. Introduction

Indonesians have a solid understanding of financial institutions. Therefore, the financial literacy program needs to be implemented. The government has already educated the public about financial literacy in collaboration with a prominent Indonesian Islamic financial inclusion bank that prioritizes women’s empowerment [1]. However, the activities were impacted by the pandemic. This educational program was interrupted and halted due to the COVID-19 pandemic. Therefore, to revive it, innovation and technology are required. The Indonesian Service Authority’s National Survey of Financial Literacy and Inclusion estimates that 49.68% of Indonesians are financially literate [2]. Indonesians perceive entrepreneurship as essential for the growth of the Indonesian economy. However, according to the Ministry of State Secretariat of Indonesia, the ratio of new Indonesian entrepreneurship was 3.47%, which was lower than that of Singapore (8.5%) and Thailand and Malaysia (4.5%).

The number of entrepreneurs in Indonesia [3] was 55 million in micro, small, and medium enterprises (MSMEs), where women own 60% of those, or 33 million. According to data from the Global Entrepreneurship Monitor 2015, the population of productive-age women in Indonesia was 85 million, 26% of whom were womenpreneurs. The latest data published by the Central Statistics Agency in 2021 showed that 64.5% of 55 million MSMEs in Indonesia were owned and managed by women. More than 50% of the MSMEs were micro and small enterprises. The level of women’s business ownership increased faster than men’s because policies and culture in Indonesia began to encourage women to enter the business world. From the data released by the Canada–Indonesia Trade and Private Sector Assistance Project in 2019, MSMEs owned by Indonesian women operated in service rather than manufacturing, except for food production.

When creating new technology, it is essential to understand how well technology users accept it. The unified theory of acceptance and use of technology (UTAUT), the most popular model for assessing possible acceptance-influencing elements, is utilized to evaluate the prototype product [4]. The UTAUT model was used with the following hypothesis: (1) behavioral Intention (BI) is positively impacted by facilitating conditions (FCs), satisfaction (S), social influence (SI), and continuance intention (CI). According to the statistics, e-money boosts business performance. Focusing on financial literacy, we identified the significant factors influencing consumers in the middle-to-lower socioeconomic groups’ adoption of technology. Additionally, an acceptance model based on UTAUT was developed.

2. Literature Review

2.1. Financial Literacy

Financial literacy includes knowledge, skills, attitudes, and certainty around financial decision-making. The operational components of financial literacy encompass borrowing, budgeting, saving, and investing. Financial literacy is related to understanding fundamental economic concepts and the capacity to use that information [4,5]. Financial literacy positively impacts the decision-making process. A profound comprehension of financial concepts, such as risk, return levels, credit card payment patterns, and household budgeting, typically leads to more creative choices [6,7,8,9]. Every short- and long-term financial decision is influenced by financial literacy, which has effects on both individuals and society. Financial literacy favorably influences saving behavior [10,11]. Financial knowledge also boosts an individual’s self-esteem, productivity, and employability while also assisting them in avoiding financial troubles [11,12].

There is a wide range of elements that influence financial literacy. The level of financial literacy and financial behavior is significantly influenced by financial education. Additionally, income level impacts financial behavior and financial literacy [13,14]. MSMEs in India demonstrated that the quantity of gross income from a business was the element that most determines the financial literacy of MSME actors [14,15,16].

2.2. UTAUT

The UTAUT model has been widely employed to measure the reception of healthy technologies [17]. Reference [18] examined eight models from earlier studies on the adoption of information technology, including the theory of reasoned action (TRA), the theory of planned behavior (TPB), the motivational model (MM), the model of PC utilization (MPCU), the model of combining TAM and TPB (C–TAM–TPB), the innovation diffusion theory (IDT), the technical adaptation model (TAM), and the social cognitive theory (SCT). UTAUT is a more thorough paradigm (Figure 1). UTAUT reintegrates the earlier models to separate variables into two categories. The first category consists of four variables: social influence, performance expectancy, effort expectancy, and facilitating circumstances. The second category consists of moderating variables, such as gender, age, voluntariness, and experience, which majorly influence core variables [18].

UTAUT uses four constructs to explain how people accept new technologies: performance expectancy, which measures the advantages of using the technology under study; effort expectancy, which captures how easy or difficult it is to use it; and acceptance expectancy [18]. Social influence and enabling conditions are analyzed at the level of technological infrastructure that supports and makes using the technology accessible. Social influence is examined to understand how the use of technology by social groups may affect user acceptability [19]. For example, chatbots were evaluated using the UTAUT model [20]. The UTAUT model is described in Figure 1.

3. Methodology

We investigated the relationship between performance expectancy, social influence, facilitating condition, satisfaction, and continuance learning. The reliability and validity of data were tested to analyze the impact of expectancy and facilitating conditions on continuance intention through the mediation of satisfaction.

When creating new technology, it is essential to understand how it is accepted by potential users. UTAUT is the most widely used model to assess factors impacting user acceptance and test the prototype. The UTAUT model was used with the following hypothesis: (1) behavioral Intention (BI) is positively impacted by facilitating conditions (FCs), satisfaction (S), social influence (SI), and continuance intention (CI). We assessed the level of technology adoption by MSMEs in Indonesia to determine middle-class and lower-class users’ use of technology and their financial literacy. Five primary constructs were used in this model with the following details (

Table 1. Construct and measurement variables in this study.

Construct	Variable	Measure	Reference
PE	PEE1	Using the e-learning platform to gain financial literacy can be beneficial daily.	[21]
	PEE2	I can rapidly understand financial literacy by using the e-learning platform.
	PEE3	My MSME productivity may benefit from using all the financial literacy learning resources available on the e-learning platform.
FC	FCI	I know the fundamentals of using technology in my daily life	[22]
	FC2	I have access to tools like smartphones and the internet to help me.
	FC3	The e-learning platform can still be accessed using reasonably priced mobile internet.
	FC4	The cost of using e-learning technologies (the internet and mobile devices) is affordable.
S	SI	I quickly found the knowledge I needed because the e-learning application’s content is well-structured.	[23]
	S2	All the knowledge required for financial literacy is available on the e-learning platform.
	S3	The e-learning platform has every aspect I need to improve my understanding of financial literacy.
	S4	All in all, I am happy with the app’s interface.
SI	SII	Nowadays, people in my community would assume that I should learn about financial literacy through technology.	[23]
	S12	My family believes I should use technology to boost my productivity as an MSME.
	S13	I could get help from people I care about using e-learning sites to improve my financial literacy.
CI	CII	When the complete version of this e-learning platform is made available, I’ll use it.	[23,24]
	CI2	I advise my family members to improve their financial literacy using the e-learning platform.
	CI3	At some time, using technology will help my business.
	CI4	In the future, I’ll use the prototype to increase my understanding of financial literacy.
	CI5	Compared to other options, I intend to use the program considerably.

). The theoretical research framework of this study is shown in Figure 2.

• PE refers to the perception that technology helps MSMEs’ owners increase their business performance and measure how easy the system is to use.
• FC is defined as a belief of the individuals that an organizational and technical infrastructure can support the use of the technology.
• S refers to how the users’ overall impression can affect their acceptance of technology.
• SI is used to evaluate how the usage of technology by relatives and acquaintances affects the user’s acceptance of it.
• CI is defined as an individual’s subjective probability that they would engage with the technology.

A questionnaire was used to collect data to analyze the respondents’ opinions. A total of 250 respondents’ demographics, including age, possession of a mobile phone, and technological aptitude, were analyzed. Womenpreneurs from Banda Aceh, Jakarta, Bogor, Depok, Tangerang, and Bekasi City, Indonesia, participated in the survey. The survey data were analyzed using ANN and SPSS version 26.

4. Results and Discussion

Age is one of the factors affecting the acceptance and use of information and communication technology (ICT). The baby boomer generation, generations X, Y, and Z are the four age-generational categories. The age ranges of the four generations vary. Generation Z is the youngest, and baby boomers are the oldest. A total of 48.92% of womenpreneurs were from Generation X. Generation X showed the highest rate of technological acceptance, whereas baby boomers had the lowest rate (

Table 2. Respondents’ age.

Generation	Age	Number	Percentage
Generation Z	10–25 years old	6	3.23
Generation Y	26–41 years old	84	45.16
Generation X	42–57 years old	91	48.92
Baby Boomers	58–76 years old	5	2.69

). Generation X needs to be prioritized because they are the most intended consumers of digital financial literacy applications.

In this study, smartphone ownership was analyzed to observe how simple it was for womenpreneurs to use technology. The difficulties of using a smartphone impact the ability to access technology. Most womenpreneurs owned smartphones. They showed a large independence on smartphones to obtain information. The ease of use of the financial literacy e-learning smartphone application was examined. Compared with womenpreneurs who owned smartphones, those without had fewer opportunities to acquire financial literacy expertise (

Table 3. Respondents’ smartphone ownership.

Smartphone Ownership	Number	Percentage
Family-owned	12	6.45
Sharing	13	6.99
Self-owned	161	86.56

).

We identified dependent variables related to satisfaction and continuance intention. The established model used 80% of the data as a training dataset and 20% as a testing dataset, following Pareto rules. We used Sigmoid and hyperbolic tangent in the layer. Figure 3 and Figure 4 and

Table 4. ANN accuracy.

Variable	ANN Training	ANN Testing
Continuance intention	0.384	0.417
Satisfaction	0.445	0.518

show the configuration information and architecture of the ANN.

The accuracy of the three developed ANN models was evaluated using the root-mean-square error (RMSE). RMSE represents the errors made during the training and testing phases. The models’ average RMSEs were 0.384 and 0.417 for the training data and 0.445 and 0.518 for the testing data. The models were accurate in predicting a range of endogenous categories. The ANN models developed in this study were validated to be reliable and precise. The normalized importance was calculated by comparing the average of each predictor to the highest mean value, which is expressed as a percentage.

Table 5. Independent variable for S.

Variable	Importance	Normalized Importance
PEE1	0.113	78.40%
PEE2	0.093	64.20%
PEE3	0.096	66.20%
PEE4	0.063	43.50%
PEE5	0.035	23.90%
PEE6	0.118	81.30%
FC1	0.051	35.40%
FC2	0.032	21.90%
FC3	0.033	22.70%
FC4	0.145	100%
SI1	0.098	67.80%
SI2	0.091	63.20%
SI3	0.032	22.40%

and

Table 6. Independent variable for CI.

Variable	Importance	Normalized Importance
PEE1	0.081	36.90%
PEE2	0.081	36.90%
PEE3	0.059	26.80%
PEE4	0.047	21.40%
PEE5	0.035	15.80%
PEE6	0.044	20.10%
FC1	0.055	25%
FC2	0.063	28.50%
FC3	0.025	11.40%
FC4	0.032	14.50%
S1	0.029	13.20%
S2	0.081	36.70%
S3	0.145	65.80%
S4	0.221	100%

show each predictor’s mean and normalized relevance during the ANN modeling procedure. Independent variable importance results showed that FC4 was the most significant variable related to S (0.145). At the same time, S4 was the essential variable associated with CI (0.221). The average RMSEs for the two neural network models were 0.445 and 0.518 for the testing data and 0.384 and 0.417 for the training data. This illustrated how well the models predicted a variety of endogenous categories. The normalized importance was determined by comparing each predictor’s average to the highest mean value, expressed as a percentage. According to the results of independent variable importance, FC4 was the most critical variable for S (0.145). S4 was the crucial variable linked to CI (0.221).

5. Conclusions

The study’s results provide an important reference for creating mobile applications for womenpreneurs. Because of its limited demographics, it is necessary to consider its restrictions. More social difficulties and factors from other theories and models need to be included, and the sample size also must be increased to generalize the results of this study. Future studies are necessary to look at an enterprise at a macro level using non-technological approaches (such as in-person instruction) to help womenpreneurs in Indonesia or other areas become more financially literate.