Determinant of Behavioral Intention to Use Digital Zakat Payment: The Moderating Role of Knowledge of Zakat

: The potential for zakat in Indonesia is increasing over time. However, the value of the increase in zakat receipts is not proportional to the possibility of zakat. Research on zakat behavior using digital platforms is essential considering the lack of realization of zakat collection compared to the existing potential. This study develops the UTAUT model by adding knowledge of zakat as a moderated variable. A quantitative approach is used in this research. The number of samples in this study was 200, obtained through convenience sampling based on questionnaires distributed online. Data analysis in this research used partial least square-structural equation modeling (PLS-SEM). The finding of the study explains that performance expectancy, social influence, and facilitating conditions positively affected behavioral intention to use digital zakat payment. Meanwhile, the moderating effect shows that the relationship between performance expectancy and behavioral intentions is moderated by knowledge of zakat. These results have implications for the need to increase the effectiveness of digital zakat payments, have influencers who educate about zakat payments using digital and enrich the choice of zakat management institutions that accommodate digital payments so that the intention of muzakki to pay zakat using digital payments is rising. The moderating effect implies that for muzakki who have higher knowledge of zakat, the trust and effectiveness of digital payments can increase the desire to pay zakat through digital payments.


INTRODUCTION
research has discussed how muzzaki behavior uses digital media. The models used to explore the muzzaki behavior include the Technology Acceptance Model (TAM) (Purwanto et al., 2021;Utami et al., 2020) and the unified theory of acceptance and use of technology (UTAUT) (Kasri & Yuniar, 2021;Sulaeman & Ninglasari, 2020). This study wants to conform to the UTAUT theory in the context of the intention to use digital zakat fulfillment. UTAUT theory has been considerably applied in the previous study related to the response and application of technology. However, the study involves different findings for each variable that affect technology's intention and behavior. The findings indicate that the first variable, namely Performance expectancy, positively affects on behavioral intention (Hau et al., 2021;Musahidah & Sobari, 2021;Rahardjo et al., 2020;Samsudeen et al., 2020;Soodan & Rana, 2020). However, other findings state that performance expectancy does not affect behavioral intention (Izzati, 2020;Njenga & Salih, 2018;Purwanto & Loisa, 2020). The study results for the second variable, namely Effort expectancy, positively affect the behavioral intention (Intarot, 2018;Sah et al., 2021;Soomro, 2019;Sulaeman & Ninglasari, 2020;Teoh et al., 2020). However, an exciting finding states that Effort expectancy does not affect the behavioral intention (Soodan & Rana, 2020;Syifa & Tohang, 2020;Widodo et al., 2019). The third variable, namely Social influence, positively affects behavioral intention (Hau et al., 2021;Izzati, 2020;Musahidah & Sobari, 2021;Patil et al., 2020;Rahardjo et al., 2020). Another exciting finding states that social influence does not affect behavioral intention (Intarot, 2018;P & Manohar, 2021;Purwanto & Loisa, 2020;Raza et al., 2019;Soomro, 2019). The fourth variable, namely Facilitating condition, positively affects behavioral intention (P & Manohar, 2021;Patil et al., 2020;Purwanto & Loisa, 2020;Rahardjo et al., 2020;Samsudeen et al., 2020). Meanwhile, other results state that the Facilitating condition does not affect behavioral intention (Do et al., 2020;Musahidah & Sobari, 2021;Sulaeman & Ninglasari, 2020). Given the inappropriate results of the UTAUT model, it is important to construct the UTAUT model. This research adds knowledge of zakat as a moderating variable that can strengthen or weaken the relationship between the predictor and criterion variables. The addition of a moderating variable can be added if there are inconstant findings in a relationship between variables. In this study, the moderating variable is knowledge of zakat. Knowledge of zakat consists of essential knowledge and advanced knowledge. Basic knowledge consists of common knowledge, knowledge of asnaf, knowledge of the union to settle zakat, knowledge of the objects of zakat, and knowledge of zakat calculation. Meanwhile, advanced knowledge consists of knowledge of zakat institutions, zakat rule, the effect of zakat, the distribution of zakat, and digital settlement. The basic knowledge and advanced knowledge that BAZNAS includes in the component of the zakat literacy (Kasri & Yuniar, 2021;Yusfiarto et al., 2020). Adding knowledge of the zakat variable is considered appropriate because common knowledge and advanced knowledge are related to how humans use their awareness as a behavior change concept (Castro-González et al., 2020;Fujiki, 2020;Muñoz-Murillo et al., 2020).

RESEARCH METHODS
This research applies a quantitative way with a survey method. The significant advantage of this approach is to explain the proposed model to predict behavioral intention to use digital zakat payment. This study adapts UTAUT model (Venkatesh et al., 2003). Performance expectancy is explained as to how individuals believe that applying a system can support them set jobs enforcement (Venkatesh et al., 2003). In this research, performance expectancy is how respondents believe that using digital zakat payments can help them achieve work efficiency. The previous study stated that performance expectancy affects behavioral intention (Hau et al., 2021;Musahidah & Sobari, 2021;Rahardjo et al., 2020). Accordingly, the ensuing hypothesis is established: H1: Performance expectancy can improve behavioral intention to use a digital zakat payment. Social influence describes the stage to which an individual's perception of the influence of others believes to apply a new system (Venkatesh et al., 2003). In this research, social influence is described as respondents' perceptions of the influence of others' beliefs to use digital zakat payments. The prior research stated that social influence affects behavior intention (Izzati, 2020;Patil et al., 2020). In consequence, the next hypothesis is established: H2: Social influence can improve behavioral intention to use a digital zakat payment. Facilitating condition describes how individuals believe that an organization and technical infrastructure have to assist the application of the system (Venkatesh et al., 2003). In this research, facilitating condition is defined as respondents' perception that an organization and technical infrastructure have to assist the application of digital zakat payment. The previous study stated that facilitating condition affects behavior intention (P & Manohar, 2021;Purwanto & Loisa, 2020;Samsudeen et al., 2020). In consequence, the following hypothesis is proposed: H3: Facilitating conditions can improve behavioral intention to use a digital zakat payment. Knowledge of zakat is described as respondents' knowledge about zakat, both basic and advanced knowledge. In this study, the variable knowledge of zakat was added as a moderating variable. According to prior research, knowledge is related to how humans use their awareness as a behavior change concept (Castro-González et al., 2020;Fujiki, 2020;Muñoz-Murillo et al., 2020). This probe proposed the hypothesis with knowledge of zakat as a moderator variable as follows: H4: Knowledge of zakat moderates the relationship between performance expectancy to behavioral intention to use a digital zakat payment. H5: Knowledge of zakat moderates the relationship between social influence on behavioral intention to use a digital zakat payment. H6: Knowledge of zakat moderates the relationship between facilitating conditions to behavioral intention to use a digital zakat payment.
The proposed model is pointed out in figure 1.

Figure 1. Research Model
The population in this study are Muslims who have the desire and have paid zakat using digital payments. The sample amounted to 200 people determined based on the convenience sampling technique (Galloway, 2005). This number has met the minimum sample requirements for PLS-SEM criteria, where the base sample size is tenfold the number of arrows hitting a latent variable in the PLS-SEM model (Hair et al., 2017). This number also meets the minimum requirements when considering statistical power and effect size. In this study, the most significant number of arrows hitting the variable was 3, and the expected minimum R 2 was 0.5, so the minimum sample was 38 (Kock & Hadaya, 2018). The questionnaires were distributed online on social media using google forms. The first section of the survey form is composed of questions about the demographic of respondents, as well as gender, age, job, and residency. The second part, which is proposed to assess the conceptual model, is composed of 21 items asking respondents to evaluate their intention to use digital zakat payment using the adapted UTAUT framework. The statements are adopted and modified from Sulaeman & Ninglasari (2020) and Venkatesh et al. (2003). Nevertheless, the statement related to knowledge of zakat is adapted from Kasri & Yuniar (2021). The five-point Likert scale measures all statements, ranging from 1 = strongly disagree to 5 = strongly agree. A pilot study was established to ensure the validity and reliability of the tool for data collection and to ensure that the respondents clearly understood the words. Validity and reliability tests are conducted with 30 respondents using IBM SPSS version 23 before examining the whole model. Data analysis was performed using PLS-SEM. There were three stages carried out, namely the assessment of the measurement (outer) model, the assessment of the structural (inner) model, and the analysis of moderation. There are three criteria at the evaluation stage of the measurement model: first, composite reliability is used for measured internal consistency. Second, outer loading and the average variance extracted (AVE) are used for measured convergent validity. Third, Fornell-Larcker criteria are used for the measured discriminant validity (Sholihin & Ratmono, 2021). The next stage is the evaluation of the structural model, which is composed of five steps, namely assessing the problem of collinearity, evaluating the significance and relevance of the structural model relationship, evaluating the level of the coefficient of determination, evaluating the effect size, and evaluating predictive relevance (Hair et al., 2017). The last stage in data analysis in this investigation is moderation analysis. Moderation analysis in this study uses a twostage approach because this approach can be used both to measure exogenous variables and reflective and formative moderating variables. This approach also has higher statistical power than other approaches (Sholihin & Ratmono, 2021).

RESULTS AND DISCUSSION 1. Respondent' Profile
The description of the respondents in this research is described by the criteria of gender, age, jobs, and residency. Based on table 1 profile of research respondents, it is explained that the number of male respondents is 61 people or (30.5%), and the number is 139 people or (69.5%) of the amount sample. In terms of age, respondents have varying ages ranging from ages 15 to 25, amount 145 people or (72.5%), ages 26-35 years amount, 45 people or (22.5%), and 36-45 years, amount ten people or (5%). Apart from the demographics related to gender and age that will be used as constructs, the analysis is also carried out in terms of occupation and place of residence. Most widely respondents in this study worked as students amounted to 110 people (55%) and at least one respondent in state-owned enterprises and farmers, severally one person (0.5%). Respondents who work as entrepreneurs amounted to 33 people (16.5%), private employees amounted to 25 people (12.5%), teachers amounted to 19 people (9.5), lecturers were eight people (4%), and civil servants were three people (1.5%). Then the respondents who lived the most were North Sumatra with 96 people (48%), Yogyakarta with 33 people (16,5%), West Java with 11 people (5.5%), West Java and South Sulawesi each amounted to 9 people (4.5%), Riau amounted to 8 people (4%), East Java totaling six people (3%), Bengkulu and Jakarta totaling four people (2%), West Nusa Tenggara, and South Sumatra counting three people (1.5%), West Sumatra amounted to 2 people (1%) and Bali, Banten, Gorontalo, Jambi, West Kalimantan, South Kalimantan, West Kalimantan, Riau Islands, Lampung, East Nusa Tenggara, Papua, and West Sulawesi respectively, each amounted to 1 person (0.5%).

Measurement Model Internal Consistency Reliability
The construct reliability can be assessed by composite reliability. A construct is announced reliable if the combined reliability value is > 0.70, as shown in table 2. These results show the composite reliability of each construct, namely performance expectancy with a value (0.924), social influence with a value (0.919), facilitating conditions with a value (0.913), behavioral intention with a value (0.938), and knowledge of zakat with a value (0.923). It can be concluded that the overall composite reliability value variable is > 0.70 and meets the composite reliability criteria (Hair et al., 2017).

Convergent Validity
The convergent validity of the outer model can be measured by the correlation between the indicator score and the loading factor. The standard for validity loading factor value of each indicator is greater than 0.70 (Sholihin & Ratmono, 2021

Discriminant Validity
The results of the Fornell-Lercker Criterion calculation in this research are shown in Table 3. Based on discriminant validity testing through Fornell-Lacker criteria, it can be seen that each variable is more than the correlation of each construct with other constructs, which is more than 0.70. Accordingly, the constructor variable in this study has a good discriminant validity value (Sholihin & Ratmono, 2021).

Structural Model Collinearity (VIF)
The total collinearity VIF value results from a full multicollinearity test that includes vertical and lateral multicollinearity Fields (Kock & Lynn, 2012). The output of the multicollinearity examination in this study showed VIF values of 1.722, 1.674, 1.965, 2.671, and 1.570, which overall value is smaller than 3.3. This means that the model in this study is free of vertical and lateral multicollinearity and the Cammon method bias because the VIF value is < 3.3.

Path Coefficient
The results of the inner model analysis in this research which describe the magnitude of the path coefficient and p-value, are in Table 4. Building upon the results of the structural (inner) model in Table 4, it can be known that the P-Value value of 0.00 is smaller than 0.05, with the resulting coefficient value of 0.28. H1 can be accepted because performance expectancy can improve behavioral intention to use digital zakat payments. This study confirms previous findings that performance expectancy affects behavioral intention (Hau et al., 2021;Musahidah & Sobari, 2021;P & Manohar, 2021;Rahardjo et al., 2020;Samsudeen et al., 2020). H2 in this study is that social influence can improve behavioral intentions to use digital zakat payments. Building upon the results of the structural (inner) model in Table 4, it can be known that the P-Value value of 0.00 is smaller than 0.05, with the resulting coefficient value of 0.28. H2 can be accepted because social influence can improve behavioral intention to use digital zakat payments. This study confirms previous findings that social influence positively affects behavioral intention (Izzati, 2020;Patil et al., 2020;Soodan & Rana, 2020;Sulaeman & Ninglasari, 2020;Teoh et al., 2020). H3 in this study is facilitating conditions that can improve behavioral intentions to use digital zakat payments. Building upon the results of the structural (inner) model in Table 4, it can be known that the P-Value value of 0.00 is smaller than 0.05, with the resulting coefficient value of 0.33. This means that H3 can be accepted because facilitating conditions can improve behavioral intention to use digital zakat payments. This study confirms previous findings, stating that facilitating conditions positively affect behavioral intention (Purwanto & Loisa, 2020;Raza et al., 2019;Soodan & Rana, 2020;Soomro, 2019;Widodo et al., 2019).

R-Square
Building upon the test results, the R-Square (R 2 ) value is 0.631 (63.1%). This means that the performance expectancy, social influence, and facilitating conditions can be defined as the behavioral intention to use digital zakat payments by 63.1%, classified as moderate because the R 2 value is < 0.75. At the same time, the other 36.9% were explained by other variables outwards the model.

Effect Size (f square)
The output showed that the value of the effect size of performance expectancy on behavioral intention was 0.179 > 0.15. This value means performance expectancy has a moderate relationship to digital zakat payments' behavioral intentions. The value of the effect size of social influence on behavioral intention is 0.161 > 0.15. This means that performance expectancy has a moderate relationship to digital zakat payments' behavioral intentions. The value of effect size facilitating conditions for behavioral intention is 0.215 > 0.15. This means that performance expectancy has a moderate relationship to digital zakat payments' behavioral intentions.

Relevant Prediction (Q square)
Building upon the test results, the predictive relevance value (Q 2 ) is 0.622. It can be interpreted that the structural model used in this study can estimate the parameters and has an observation value of 0.522.

Moderating Analysis
The output of the moderation analysis in this study is pointed out in Table 5.  Table 5 explains that the pvalue is 0.019 < from 0.05 with the resulting coefficient value of 0.144. This means that hypothesis 4 can be accepted because knowledge of zakat can increase performance expectancy on behavioral intention to use digital zakat payments. It implies that the efficiency of digital zakat payment improves intention to use, especially for people who have better knowledge about zakat. People who have basic and advanced knowledge about zakat will understand that paying zakat using a digital platform will increase payment efficiency. Afterward, the higher the payment efficiency understanding, the higher their intention to use digital zakat payment.
The fifth hypothesis of this study is that knowledge of zakat can moderate social influence on behavioral intention to use digital zakat payments. Based on the moderation analysis test results in table 5, it is explained that the p-value is 0.418, which means >0.05 with the resulting path coefficient value of 0.015. This means that hypothesis 5 is rejected because knowledge of zakat cannot increase the effect of social influence on behavioral intention to use digital zakat payments. From these findings, it can be understood that friends, relatives, and influential people around them might cancel individual understanding of the digital zakat payment. Their belief in their knowledge may not be as strong as their belief in the importance of others.
The sixth hypothesis in this study is that knowledge of zakat can moderate facilitating conditions for behavioral intentions to use digital zakat payments. Building upon the moderation analysis output in Table 5, it is explained that the pvalue is 0.464, which means > 0.05, with the resulting path coefficient value of 0.006. This means that hypothesis 6 is rejected because knowledge of zakat cannot increase the influence of facilitating conditions on behavioral intentions to use digital zakat payments. It implies that one's support and resources have not been able to advance understanding of digital zakat payment. Their belief in their knowledge maybe not be as strong as their belief in the resources and infrastructures related to digital zakat payment. It can happen because zakat literacy in Indonesia is relatively low (Canggih & Indrarini, 2021).

CONCLUSION
This study indicates that performance expectancy, social influence, and facilitating conditions can improve behavior intention to use digital zakat payment. These results have implications for the need to increase the effectiveness of digital payments, have influencers who educate about zakat payments using digital payments and enrich the choice of zakat management institutions that accommodate digital payments so that the intention of muzakki to pay zakat using digital payments is rising. The output of the moderating effect examination pointed out that knowledge of zakat only improves the quality of the relationship between performance expectancy and behavior intention. Interestingly, the relationship between social influence and facilitating conditions with behavioral intention does not moderate by knowledge of zakat. This means that for muzakki, who have a higher understanding of zakat, the trust and effectiveness of digital payments can increase the desire to pay zakat through digital payments. The limitation of this study only examines the intention to use digital zakat payment. Further research can explore how the behavior of muzakki using digital zakat payments. Future research also needs to explore the antecedents of muzakki satisfaction in using digital zakat payments.