Determinants of Tax Non-compliance Intention among SME Owners in Malaysia: The Moderating Effect of Religiosity

The importance of small and medium enterprises (SMEs) as the backbone in contributing to national income in Malaysia is undoubted. However, the Inland Revenue Board of Malaysia (IRBM) had reported statistical evidences pertaining to tax non-compliance amon g SMEs’ owners. Despite its importance, their tax non-compliance behaviour has not been given enough attention, particularly on their owners. Therefore, this study empirically investigates the determining factors of SMEs’ owners on their tax non -compliance intention. This study extends the theory of planned behaviour (TPB) model by incorporating the theory of religious commitment in order to determine the relationship between the constructs of TPB and religiosity and tax non-compliance intention as well as to test the moderating effect of religiosity. Employing survey design, questionnaires were randomly distributed to 580 SME owners in Malaysia, with 72.8 percent response rate. The data collected were analysed by using PLS-SEM statistical tool. Findings reveal that attitude towards tax non-compliance and subjective norms positively influence tax non-compliance intention, whereas perceived behavioural control negatively influences tax non-compliance intention. Findings also indicate that the significant positive relationship between attitude towards tax non-compliance with tax non-compliance intention became weaker for highly religious SME owners. Moreover, the significant negative relationship between perceived behavioural control with tax non-compliance intention was stronger for SME owners with higher religiosity. These findings contribute theoretically and practically with imperative implications to academicians, government and tax authorities.


Introduction
SMEs have been known as the backbone of Malaysia and are important in generating national income. As released by the Small and Medium Enterprise Corporation Malaysia (known as SME Corp.) on 29 September 2017, based on the census of SMEs registered in Malaysia, there were a total of 920,624 SMEs in 2015. Such a huge number shows that SMEs could highly contribute to the national income if they comply with the tax regulations. However, based on the data provided by the IRBM, the audited cases resolved for SMEs in 2016 was 541,798 compared to 529,010 in 2015, indicating an increase of 2.4 percent. Despite the small increase, it may be considered as an indicator of a problem of tax non-compliance behaviour. Due to the SMEs' significant contribution to the Malaysia's economy, coupled with the fact that they represent the majority of business taxpayers, it is expected that more income taxes should be collected from this group of taxpayers. Nevertheless, the collection of national tax revenue is directly affected by the taxpayers' level of compliance. Tax authorities, academicians and non-governmental organisations all over the world are concerned about the serious problem of tax non-compliance among SME owners (Newman et al., 2018). Zakaria et al (2013) believe that tax non-compliance has been a worldwide phenomenon and has triggered such a huge loss of revenue to governments around the world. Surprisingly, a briefing paper presented by Murphy (2011) on the worldwide tax evasion cost reveals that out of 145 countries included in a survey, Malaysia was 44th worldwide and 11th in Asia. According to Ching (2013), tax non-compliance problems may cause shortfalls to a country's tax revenue and could be categorised as a white-collar crime.
According to Mohamad et al (2016), the characteristics and nature of business itself encourage SME owners not to comply with tax requirements. Non-compliance may arise because of the high opportunities to involve in cash business transactions and the possibilities of not to declare them. The SMEs' intention of not being truthful is undeniable since they have huge opportunities to deal with cash transactions, including sales and payments of purchases (Fatt et al., 2009); Morse, Karlinsky & Bankman, 2009). It is not easy for tax authorities to inspect businesses which are heavily based on cash transactions since there is a lack of proper documents to act as evidence (Yusof et al., 2014). Besides that, a study by Ali (2013) finds that the compliance level of business income taxpayers is at medium to low levels. This is because taxpayers have a high opportunity of avoiding taxes, which appeared to be mostly overlooked by the tax authorities. Mohamad et al (2016) also find that tax evasion to be highest among micro-and medium-sized SMEs located in suburban areas.
However, the issue should not be seen solely from the economic approach, but it should also be considered from the non-economic view. Attempts to explain individuals' behaviour have incorporated alternative perspectives from the non-economic approach, which include behavioural and psychological factors (Kirchler et al., 2012). Most of prior tax compliance studies focus on attitude, subjective norms and perceived behavioural control. However, Ali (2013) claims that religiosity should be further investigated in explaining the behaviour of taxpayers since individuals' religious values are likely to continuously inhibit negative attitudes and to encourage positive attitudes effectively. Thus, this may acknowledge the intrinsic factors of an individual in explaining the intention of taxpayers' non-compliance behaviour.
Religiosity is also found to be one of the moral forces to motivate taxpayers to pay taxes (Ali, 2013;Stack & Kposowa, 2006;Torgler, 2006). Despite producing inconsistent results of religiosity affecting compliance behaviour, many of the prior studies highlight that religiosity plays an important role in inhibiting the intention of non-complying with tax requirements (Muslichah, 2015). Religiosity is believed to explain the compliance behaviour of taxpayers since it may discourage negative attitudes and inspire positive attitudes of an individual (Eiya et al., 2016). Furthermore, Ali (2013); Mohdali & Pope (2014) suggest that future tax compliance studies which include religiosity as one of the contributing factors should include more self-employed taxpayers to improve the generalisation of findings to the Malaysian population. They suggest that religious values held by taxpayers may solve the tax noncompliance issue in Malaysia. Furthermore, prior studies recommend employing selfemployed taxpayers as a sample to offer a better picture of self-employed taxpayers' compliance behaviour since the chances of self-employed taxpayers to engage in tax noncompliance is higher than salaried taxpayers (Ali, 2013;Mohdali & Pope, 2014). Stack and Kposowa (2006) and Torgler (2006) also find that taxpayers' religious belief has become one of the main determinants explaining voluntary tax compliance. However, despite its importance in explaining taxpayers' behaviour, the role of religiosity as a moderator has not been widely discussed and brought to attention in discussing the SME owners' tax behaviour. Religiosity is expected to strengthen, weaken or alter the relationship between selected factors and the tax non-compliance intention among SME owners.
Given the gap of knowledge as discussed above, together with the importance of the contribution of SMEs and the need to restrain tax non-compliance, there appear considerable opportunities for further study. Besides observing the original TPB constructs, this study examines the influence of religiosity as regards to tax non-compliance intention as well as examines if religiosity moderates the relationship between the proposed determining factors and SME owners' tax non-compliance intention. Roth et al (1989) define tax non-compliance intention as an unwillingness to file the tax return, an underreporting of taxable income or an overstating of allowable deductions claimed from chargeable income. This definition will be adopted in this study since it provides a better explanation on the voluntary part of complying with tax because the success of selfassessment system is based on the taxpayers' level of voluntarism. The term 'intention' used as a proxy to represent SME owners' actual tax non-compliance behaviour. The use of the intended action instead of actual behaviour is acceptable in investigating sensitive issues, such as those involving actual unethical behaviours, in which researchers will face difficulties in obtaining those sensitive information (Alleyne & Harris, 2017). Intended actions have been applied by most researchers in predicting intention of people to engage in tax non-compliance (see Alleyne & Harris, 2017;Bobek & Hatfield, 2003;Hamid, 2013;Kuria & Omboi, 2016;Langham et al., 2012;Saad, 2011;Salleh, 2011;Smart, 2012;Sudiartana & Mendra, 2018;Wahyuni et al., 2019).

Theory of Planned Behaviour
Theory of Planned Behaviour (TPB) proposes three constructs, namely attitude towards tax non-compliance, subjective norms and perceived behavioural control. These constructs may affect one's behavioural intention and thus one's actual behaviour. In addition, actual behaviour also being influenced by perceived behavioural control. However, the traditional TPB model has been criticised of not being able to thoroughly describe individuals' behaviour by solely depending on its original variable, thus Ajzen (1991) suggests that the model be extended to include other determinant variables in analysing the intention or the behaviour of individuals. Ajzen (1991) defines attitude towards a behaviour as "the degree to which a person has a favorable or unfavorable of an evaluation of a behavior". Beck and Ajzen (1991), further explain that the more favourable attitude towards behaviour, the stronger the individuals' intention to perform the behaviour under consideration should be. In other words, the more favourable attitude towards non-compliance behaviour, the stronger the intention of noncompliance to perform such behaviour. From the view of tax compliance behaviour, Loo, McKerchar and Hansford (2009) refer attitude towards tax as the level of taxpayers' confidence in managing or assessing their tax matters and this will reflect their attitude towards compliance behaviour. Attitude is found to have a significant positive relationship with behavioural intention towards tax compliance in Bobek and Hatfield (2003); Buchan (2005); Hai and See (2011); Hanno and Violette (1996); behavioural compliance in Trivedi et al (2005) and tax evasion in (Alleyne and Harris, 2017;Wahyuni et al., 2019). A cross-cultural study by Saad (2011);Hamid (2013) also find a positive significant relationship between attitude and behavioural intention towards compliance behaviour for both scenarios in Malaysia and New Zealand. Studies on SME owners also found that attitude and intention of compliance behaviour in Langham, Paulsen and Hartel (2012) and Sudiartana and Mendra (2018)  Attitude as a factor that influences non-compliance behaviour is well supported and the TPB also suggests that individuals are more likely to be more compliant if they perceive that they will receive a positive outcome from their compliance behaviour. However, they will be less compliant if they perceive an unfavourable outcome from their behaviour (Ajzen, 1991;Fishbein & Ajzen, 1975;Kirchler, 2007). Thus, the following hypothesis is proposed: H1: Attitude towards tax non-compliance positively influences the tax non-compliance intention of SME owners. Ajzen (1991) defines subjective norms as the influence of pressures from perceived referent others, such as family members, employers, friends, spouse, tax preparers, and peers in approving such decisions. According to Beck and Ajzen (1991), in line with the argument put forward in the TPB model, that if referent others approve or perform the behaviour, there is a high possibility that individuals may also perform the behaviour. In other words, it indicates that, if referent others approve of perform that non-compliance action, it will lead to an increase in the tax non-compliance action. Kirchler (2007) states that taxpayers are likely to follow the behaviour of other people that surround them. The signal received from the reference group may influence the compliance or non-compliance level (Jayawardane, 2015). This notion, which is in line with the TPB is supported by Hamid (2013) where it believes that individuals tend to perform such behaviour when the referent others also perform or approve that particular behaviour.

Subjective Norms and Tax Non-compliance Intention
Studies by Saad (2011); Hai and See (2011); Hanno and Violette (1996); Smart (2012); Sudiartana and Mendra (2018) find subjective norms has a positively significant influence on tax compliance behaviour. Adimassu and Jerene (2016); Bobek and Hatfield (2003); Damayanti et al (2015) also find a consistent finding where subjective norms has a positively significant influence with tax compliance. The subjective norms and both behavioural intention and compliance in Trivedi et al (2005) are also find positively associated. On the other hand, the study shows insignificant effect of subjective norms and overstates expenses intention. This is in line with (Hamid 2013;Buchan, 2005). Alleyne and Harris (2017); Wahyuni et al (2019) which finds that the practice of tax evasion by individual taxpayers is not being influenced by subjective norms. Findings from Inasius (2015) finds an insignificant effect of subjective norms in influencing SME owners' tax compliance. The study believes that support from referent others does not impact the tax compliance decisions of SME taxpayers.
As a summary, mixed findings have been found regarding taxpayers' behaviour towards tax compliance. With regard to the TPB and interview findings of Hamid (2013) in which the respondents reveal that they also exchange ideas with their colleagues especially in complicated tax cases. They also believe that the level of expertise and experience of their colleagues might influence their decisions. Langham et al (2012) also believe that subjective norms who possess experiences such as tax agents, bookkeepers and accountants are among the important determinants that enhances the intention of compliance behaviour. It is expected that the tendency of SME owners to involve in tax non-compliance is higher when they obtain strong support from their referent others. Therefore, the tested hypothesis will be: H2: Subjective norms positively influence the tax non-compliance intention of SME owners. Ajzen (1991) defines perceived behavioural control as an individual control, which refers to "the degree of control of an individual perceives he/she has to engage in a particular behavior". According to Ajzen and Beck and Ajzen (1991), an intention to perform such behaviour could also be influenced by other causes such as the resources availableness and chances to perform such behaviour. From the point of view of tax compliance, Kirchler (2007) explains perceived behavioural control as taxpayers' self-confidence to assess their tax returns in a way which will bring advantage to them. Bobek and Hatfield (2003) explain that perceived behavioural control shows how strong a taxpayer's degree of control is in showing a certain behaviour, such as reporting a lower income, overstating expenses and other tax non-compliance behaviour. Bobek and Hatfield and Stone, Jawahar and Kisamore (2010) find that perceived behavioural control and cheating intention are positively related. This is consistent with tax evasion study by (Alleyne and Harris, 2017;Wahyuni et al., 2019).

Perceived Behavioural Control and Tax Non-compliance Intention
A cross-cultural study in New Zealand and Malaysia by Saad (2011) also finds that perceived behavioural control under the understating income case is negatively related to compliance behaviour. However, it is insignificant under the overstating expenses scenario. Hamid (2013) finds perceived behavioural control is positively related to intention to comply under the case of understating income in New Zealand. However, perceived behavioural control is unable to explain Malaysian taxpayers' compliance intention under the case of understating income. The same findings also found under the overstating expenses case in both countries. Findings of an interview reveal that the possibility of tax agents to have more control is higher in the case of understating cash income than in overstating expenses. This is because small businesses are rarely audited.
A study by Sudiartana and Mendra (2018) finds perceived behavioural control to have a positive (negative) significant association with intention to comply (non-compliance). This study reveals that the experience and obstacles encountered by SME owners in Indonesia affect their compliance intention. As behaviour becomes more controllable, the taxpayers' intention to comply becomes greater. In addition, the study finds that the behaviour of SME owners in Bali Province is mostly influenced by their emotion and scepticism towards tax authorities. These emotions and scepticisms may be managed to encourage their compliance behaviour.
In a study conducted by Bobek and Hatfield (2003), finds that the intention to comply is not being influenced by perceived behavioural control in 'home office expenses' scenario. This is in line with (Trivedi et al., 2005;Smart, 2012;Langham et al., 2012).
All in all, evidence of the ability of perceived behavioural control to influence tax compliance behaviour is mixed and not well-documented. Although some studies find that perceived behavioural control is significant in explaining cheating and non-compliance behaviour, some other studies find the relationship to be insignificant. Thus, it has been suggested that future investigations be conducted, especially in the tax compliance area (Hamid, 2013). The finding expected that the greater the control of SME owners towards non-compliance, the lower the tendency to involve in tax non-compliance will be. Following the above discussion, the tested hypothesis will be: H3: Perceived behavioural control negatively influences the tax non-compliance intention of SME owners.

Theory of Religious Commitment
Drawing from Glock (1959Glock ( , 1962 where the discussion was merely on the religious situation during his time and within the American society, Finney (1978) then developed the Theory of Religious Commitment, which is based on the following five factors: (1) belief or ideological, (2) practice or ritualistic, (3) feeling or experiential, (4) intellectual or knowledge, and (5) devotion or consequential, which relates to the irreligious influence towards the above four factors. Worthington (1988) then developed the Religious Commitment Inventory (RCI) on theories of religious value. Initially, the measurement of RCI was based on 62 items. Later, Worthington et al (2003) modified it to suit with major religions around the world, known as Religious Commitment Inventory-10 (RCI-10) which comprised of 10 items and the samples were taken from major religions around the world, drawn from religious, non-religious and anti-religious groups.
The main element in RCI is religious commitment, defined as "the degree to which a person adheres to his or her religious values, beliefs, and practices and uses them in daily living" (Worthington et al., 2003). Worthington et al (2003) hypothesise that highly religious committed individuals (including Christians, Buddhists, Muslims, Hindus and non-religious group) are positively influenced by their religion. Religious commitment looks at how individuals are internally committed to their religion. Religious commitment has been explained and practiced in daily lives, and the commitment is not only towards beliefs in traditional religious faiths but also towards individuals' membership, participation and experience of a religious organisation (Worthington et al., 2003).

Religiosity and Tax Non-compliance Intention
According to Benk et al (2016); Eiya et al (2016), religiosity can be discussed from two different approaches: (1) religious affiliation and (2) religious commitment. Religious affiliation can be observed from one's attendance or from whether one is an active member of a religious organisation, such as church. It reflects the self-identification of an individual's religion, for instance being a Muslim, Christian, Buddhist or believer of other religions (Benk et al., 2016;Eiya et al., 2016). On the other hand, religious commitments are unobservable since they refer to beliefs of being religious and how individuals acquire absolute guidance and practice it in their daily lives (Benk et al., 2016;Eiya et al., 2016). Ali (2013) use RCI-10 in examining the religiosity influence on taxpayers' compliance behaviour and finds a significant positive (negative) relationship between religiosity and tax compliance (tax non-compliance) behaviour. This has been supported by (Benk et al., 2016;Hanifah and Yudianto, 2019;Mohdali and Pope, 2014;Muslichah, 2015;Nagac, 2016;Nazaruddin, 2019;Richardson, 2008). A cross-cultural study carries out by Mohdali, Benk, Budak, MohdIsa and Yussof (2017) finds a positive (negative) significant correlation between religiosity and tax compliance (non-compliance) behaviour in Turkey, for both voluntary and enforced scenarios. However, the Malaysian study shows that religiosity is only significant under the voluntary scenario, but when being enforced, religiosity will not have any influence on taxpayers' compliance behaviour. This is supported by Eiya et al (2016);McKerchar et al (2013); Welch et al (2005) where studies find that religious value is not the influencing factor in increasing tax compliance level since religiosity is not able to explain tax compliance behaviour.
All in all, religiosity as a factor that influences non-compliance behaviour is not well supported. Although some studies find that religiosity is important in inhibiting cheating and non-compliance behaviour, others show insignificant results. With regard to the Theory of Religious Commitment, it suggests that the tendency of SME owners to engage in tax noncompliance is expected to be higher if they are of low religiosity. Additionally, religiosity is expected to be able to curb the tax non-compliance intention since criminology studies also find that religious values are able to restrict the level of rule-breakings (Torgler & Schneider, 2007). Therefore, the hypothesis developed is: H4: Religiosity negatively influences the tax non-compliance intention of SME owners.

Religiosity as a Moderator
Undoubtedly, religiosity has been regarded as one of the influential elements of individuals in breaking a law. Several studies have taken a challenge to explain the role of religiosity as a moderator in influencing the impact of selected determinants on tax compliance behaviour. For example, Palil et al (2013) study reveals that religiosity significantly moderates the influence of tax education and knowledge in explaining tax compliance. The study stresses that religiosity is important in influencing the level of tax compliance and it believes that noncompliance behaviour should be avoided since the value of moral and ethics may distinguish between sinful and non-sinful acts. In Eiya et al (2016) study, religiosity is found to moderate the relationship between education of taxpayers and tax compliance behaviour among Nigerian taxpayers. Saad (2010) uses religiosity as a moderator between the TPB constructs and behavioural intention in complying with zakat payments, among Muslim sole proprietors and partnerships found that religiosity strengthens the influence of subjective norms on the intention to pay zakat. A study by Bobek and Hatfield (2003) extends the work of Reckers, Sanders and Roark (1994) indicate that moral obligation, as a moderator, significantly explains tax compliance behaviour. Lubian and Zarri (2011) also find that taxes are paid on the basis of people's inclination towards moral obligations. Lau et al (2013) reveals that intrinsic religiosity could explain the relationship between money ethics and tax evasion but extrinsic religiosity orientation could not explain the relationship between money ethics and tax evasion.
The above discussion shows that religiosity has been examined to find out if it could explain tax compliance behaviour in Malaysia and other countries, particularly if it can act as a moderator in explaining the relationship between some selected determinants and tax compliance behaviour. These studies are without their limitations, and some suggestions have been put forward. First, there is a suggestion from Abdul-Razak and Adafula (2013) to incorporate religious elements in exploring tax compliance. This may be worth to explore due to the absence of research in analysing religious denominations of respondents. Second, Sower and Sower (2004) argue that religiosity may be used as a moderator in testing the relationship between each of cultural and individual values and the ethical intention and behaviour of individuals. Third, Riahi-Belkaoui (2004) suggests exploring the issue of religiosity in having a clear picture of tax compliance problem. Finally, Stack and Kposowa (2006) recommend to include adults as a sample in religiosity studies because the findings on deviant behaviour would be stronger if adults are used instead of students.
Following the review of previous studies, mixed findings are found. Some studies provide evidence of a weak relationship between the determinants and the dependent variable in an attempt to explain the influence of some selected variables on the tax non-compliance intention. Thus, this study intends to take the challenge by testing religiosity as a moderator variable in explaining the tax non-compliance intention among SME owners. This is consistent with Baron and Kenny (1986) suggestion to introduce moderator variable(s) if the relationship of a predictor and a criterion variable is weak or inconsistent. They believe that a moderator may strengthen, weaken or alter the relationship between a predictor and a criterion variable. Moreover, as per the knowledge of the researcher, a limited studies in Malaysia have found to investigate the impact of religiosity as a moderator in explaining the relationship between some selected determinants and tax non-compliance among SMEs owners. Thus, the tested hypotheses will be: H1a: The positive relationship between attitude towards tax non-compliance and the tax noncompliance intention is stronger when SME owners' religiosity is low.
H2a: The positive relationship between subjective norms and the tax non-compliance intention is stronger when SME owners' religiosity is low.
H3a: The negative relationship between perceived behavioural control and the tax noncompliance intention is stronger when SME owners' religiosity is high. Figure 1 illustrates the research model of this study, based on TPB as underpinning theory and Theory of Religious Commitment as supporting theory. The TPB model has been extended in this study to incorporate religiosity as independent variables, as well as religiosity as a moderating variable. These variables are relevant in explaining Malaysian SME owners' tax non-compliance intention

Research Design
Due to the unavailability of a complete list of SME taxpayers in Malaysia, the list of firms registered under the National SME Development Council (NSDC) is used. As released by SME Corp. on 29 September 2017, the total number of SMEs registered as of 2015 was 920,624. By excluding firms that fall under large firms (13,559), SMEs located in East Malaysia (Sabah, Sarawak and the Labuan Federal Territory -119,305), and private limited companies and other legal organisations (including public limited companies, co-operatives, public corporations and private non-profit organisations -291,334) the population is 496,426 firms. With the assistance of computer application to select respondents, firms are randomly selected from the identified list of SMEs. Following the rule of thumb of Krejcie and Morgan (1970) and Barclay, Higgins and Thompson (1995), therefore, the sample sizes of 580 SME owners can be considered sufficiently large and distributed throughout Peninsular Malaysia. Out of this, the researcher manages to collect 450, however, 28 of the questionnaires are

Model Constructs and Measures
The model consists of 6 constructs with 44 items, are developed as reflective measurement models since the items for each construct are mutually interchangeable. Hypothetical scenarios comprise of 6 items solely focus on the willingness of taxpayers to declare their expenses or omit their sales used to measure tax non-compliance intention, adopted and modified by Hamid (2013); Saad (2011), which are based on Beck and Ajzen (1991) and guidelines suggested by (Ajzen, 2006). As for TPB constructs, it adopts and modifies the instruments used by Hamid (2013) and Saad (2011), which are based on Beck and Ajzen (1991) and guidelines suggested by Ajzen (2006) in measuring 12 items of attitude towards tax noncompliance, 6 items of subjective norms and 10 items of perceived behavioural control. All constructs contain a number of items or scenarios that require respondents to state their agreement towards the scenarios using a scale of 1 to 5 (1 = "strongly disagree" and 5 = "strongly agree"). With regard to religiosity, the RCI-10 measurement developed by Worthington et al (2003) solely concerned with the application of religious faith in guiding taxpayers in all aspects of life is adopted and modified in this study to measure religiosity. A five-point Likert scale is used to measure religiosity, ranging from 1 ("not at all true of me") and 5 ("totally true of me").

Result and Findings
The data collected were analysed by using PLS-SEM statistical tool. The model specification in PLS-SEM comprises two sub-models: (1) measurement model and (2) structural model, both of which are interrelated.

Assessment of Significant of the Measurement Model
In assessing the reliability and validity of the reflective measurement models, Hair, Hult, Ringle and Marko (2017) suggest criteria as shown in Table 1. The values of indicator reliability for all items meet the threshold value of 0.708 as recommended by (Chin, 1998). The values of indicator reliability that do not meet the threshold are subsequently revised and items loading below 0.708 are removed. The results of the revised model are presented in Table 1.

Table 1 Results Summary for Indicator Reliability, Composite Reliability and Average Variance Extracted
The final model (see Table 1) reports that only 29 items of the measurement model are retained. It also suggests the constructs' internal consistency reliability is satisfactorily achieved since the value of composite reliability meeting the minimum requirement of 0.708, as suggested by (Hair et al., 2017). A range of AVE values as shown in Table 1 suggests that all constructs meet a threshold value of 0.50, following suggestion by (Hair et al., 2017).
In assessing the discriminant validity, Hair et al (2017) suggest criteria as shown in Table 2 and  Table 3 The Fornell-Larcker criterion is met when the square root of the AVE is greater than the correlations as shown in Table 2. The Heterotrait-Monotrait ratio of correlation (HTMT) criterion as shown in Table 3 is met when its value is lower than 0.90 as suggested by (Gold et al., 2001). The requirement of discriminant validity is therefore met.

Assessment of Significance of the Structural Model -Direct Effect
The R-square value of the model is 0.62 which indicates that attitude towards tax noncompliance, subjective norms, perceived behavioural control and religiosity can explain 62.0 percent of the variance of the tax non-compliance intention. The R-square is considered moderate as suggested by (Chin, 1998). Table 4 shows the analysis results of the structural model for direct effect. It reveals that attitude towards tax non-compliance (β = 0.633, t = 15.913, p < 0.01) and subjective norms (β = 0.069, t = 1.856, p < 0.05) significantly and positively influences tax noncompliance intention, whereas perceived behavioural control (β = -0.171, t = 4.557, p < 0.01) significantly and negatively influences tax non-compliance intention. It shows that H1, H2 and H3 are supported. However, the path coefficient shows that religiosity has no significant influence on the tax non-compliance intention (β = -0.033, t = 0.952, p > 0.10), which shows that H4 is not supported.

Assessment of Significance of the Structural Model -Moderating Effect
The R-square after the interaction effects of religiosity as a moderator is 0.633 which indicates that attitude towards tax non-compliance, subjective norms, perceived behavioural control and religiosity collectively explain 63.30 percent of the variance in the tax non-compliance intention. The R-square is considered moderate as suggested by (Chin, 1998). Table 5 shows the analysis results of the structural model for moderating effect. It reveals that the relationship between attitude towards tax non-compliance and tax noncompliance intention is moderated significantly by the interaction of religiosity (β = -0.085, t = 1.834, p < 0.05), showing that the positive relationship between these two variables become weaker with the influence of religiosity, thus H1a is supported.
Next, it also reveals that religiosity has a significant moderating effect on the relationship between perceived behavioural control and tax non-compliance intention (β = -0.113, t = 2.926, p < 0.01). H3a, which proposed that high religiosity of SME owners strengthen the negative relationship between perceived behavioural control and tax non-compliance intention, hence is supported.
However, the statistical results reveal that religiosity does not moderate the relationship between subjective norms and the tax non-compliance intention (β = 0.051, t = 1.224, p > 0.10). Therefore, H2a is not supported.

Discussion TPB Constructs and Tax Non-compliance Intention
First, the path coefficient and the p-value indicate that attitude towards tax non-compliance positively and significantly influences the tax non-compliance intention of SME owners in Malaysia. Thus, H1 is supported. This suggests that SME owners with a favourable attitude towards manipulating income and expenses have a strong intention to engage in tax noncompliance. This is in line with the TPB as explained by Ajzen (1991), where the likelihood to perform such behaviour is higher if taxpayers believe that they will receive positive outcomes, instead of the negative ones. Prior literature reports that attitude could directly or indirectly influence the tax non-compliance intention. The finding of this study is expected to be in line with those of previous studies such as (Alleyne and Harris, 2017;Bobek and Hatfield, 2003;Buchan, 2005;Hai and See, 2011;Hamid, 2013;Langham et al., 2012;Loo et al., 2009;Marti et al., 2010;Nkwe, 2013;Saad, 2011;Smart, 2012;Sudiartana and Mendra, 2018;Trivedi et al., 2005;Wahyuni et al., 2019).
Second, with respect to subjective norms, the path coefficient and p-value suggest that subjective norms in determining the tax non-compliance intention have been found to be positively significant. Thus, H2 is supported. This suggests that SME owners with supportive significant others such as peers, families and friends who approve their actions in manipulating income and expenses would have a stronger intention to involve in tax noncompliance. This is consistent with Ajzen (1991) in explaining the TPB where the likelihood to perform non-compliance behaviour is higher among taxpayers if such behaviour is approved by their referent others. One possible explanation is that taxpayers often seek the advice of their immediate family members or friends regarding tax matters instead of the tax experts since this will minimise their compliance cost (Palil, 2010). The likelihood of non-compliance with taxes is lower in a situation where taxpayers refer to the compliant ones. On the other hand, the taxpayers can become a non-compliant too if they refer their tax matters to noncompliant taxpayers. The finding is supported by prior studies which report that the more negative views the taxpayers receive from the reference group, the more positive the intention of non-compliance would be and vice versa. The findings that there is a significant influence of subjective norms on the tax non-compliance intention among SME owners support the previous findings such as that of the study by Bobek and Hatfield (2003) where referent others are found to approve taxpayers' non-compliance. The importance of subjective norms in influencing taxpayers' behaviour is also highlighted in prior studies by (Adimassu and Jerene, 2016;Damayanti et al., 2015;Hai and See, 2011;Hanno and Violette, 1996;Saad, 2011;Smart, 2012;Trivedi et al., 2005).
Third, with respect to perceived behavioural control, the path coefficient and p-value suggest that perceived behavioural control is negatively significant in determining the tax noncompliance intention, thus, resulting in H3 being supported. The negative sign of the path coefficient indicates that lower non-compliance behaviour becomes apparent as taxpayers highly control their tax avoidance or evasion. This suggests that the higher the control the SME owners have on tax avoidance and evasion, the less likely is their intention to engage in tax non-compliance. However, this finding contradicts the view of the TPB. As explained by Ajzen (1991), taxpayers whose control over tax avoidance or evasion is high are more likely to have an intention to engage in tax non-compliance. It also appears that the finding of this study contradicts the findings of some other studies such as those of (Alleyne and Harris, 2017;Beck and Ajzen, 1991;Bobek and Hatfield, 2003;Saad, 2011;Stone et al., 2010). In their studies, the researchers find that perceived behavioural control positively influence the intention of non-compliance. Interestingly, the current finding is in line with those of Hamid (2013) and Sudiartana and Mendra (2018) which show that the more the control and the ability the taxpayers have, the more they become compliant in dealing with tax matters, thus reducing non-compliance actions. One explanation is that SME owners tend to seek good advice from tax agents in preparing their tax returns. Once they acquire the good advice from the tax agents, they will confidently assess their tax returns in a way which will bring an advantage to them such as not being severely penalised by tax authorities (Kirchler, 2007). This is in line with Kraft et al (2005) who explain that the behaviour of individuals themselves will depend on their beliefs about having control on that behaviour. Another explanation is that the type of respondents in each study varies. For example, Alleyne and Harris (2017); Bobek and Hatfield (2003) focus on individual taxpayers and Saad (2011) uses a combination of salaried and self-employed taxpayers. While in this study, the focus is on SME owners. As far as Malaysia is concerned, salaried taxpayers are subject to the monthly tax deduction regulation whereas SME owners can look for tax agents' advice on tax matters.

Religiosity and Tax Non-compliance Intention
The path coefficient and p-value of the structural model suggest that religiosity is insignificantly at influencing tax non-compliance intention among SME owners. Thus, H4 is not supported. This demonstrates that religiosity is not an important factor to be taken into consideration in making tax decisions. The finding does not support the Theory of Religious Commitment, under which there are five factors, namely belief or ideological, practice or ritualistic, feeling or experiential, intellectual or knowledge, and devotion or consequential. According to Finney (1978), these factors explain that highly religious committed individuals are those who are positively influenced by their religion. The findings of Benk et al (2016) However, the finding of this study is in line with (McKerchar et al., 2013;Palil et al., 2013;Welch et al., 2005) who find that religiosity is not able to explain tax compliance behaviour.
Only religious people with a high level of income and education tend to comply with tax compliance behaviour. As proposed by Kieffer and Sloan III (2009), the finding can be explained from the fact that morality and legality are two different principles. As far as this study is concerned, complying with tax laws and requirements is more about legality, rather than morality. On the other hand, religiosity is more related to morality. Although SME owners believe that their strong religious faith can guide them in all aspects of life, Eiya et al. (2016) argue that some people are being hypocrite by not giving an honest response. They also find that religiosity and tax compliance relationship is insignificant.

The Moderating Effect of Religiosity on the Relationship between TPB Constructs and Tax Non-compliance Intention
First, the path coefficient and p-value of attitude towards tax non-compliance (H1a) provide evidence that religiosity significantly moderates the relationship between attitude towards tax non-compliance and the tax non-compliance intention. The finding shows that the presence of religiosity has changed the direction of the relationship between attitude towards tax non-compliance and tax non-compliance intention from positive to negative. It shows that the attitude towards tax non-compliance towards the tax non-compliance intention has changed for SME owners who have high religious values. It points out, therefore, that a high level of religiosity will discourage the tax non-compliance intention of SME owners. This may be explained by the strong religious values held by many Malaysians (Mohdali and Pope, 2014). The behaviour of SME owners is a reflection of their religious beliefs and faiths which encourage them to carry out civic duties to the country. This is supported by Mohdali and Pope who explain the importance of religiosity in curbing tax non-compliance attitude.
Second, with regards to subjective norms, the path coefficient and p-value of H2a suggest that SME owners' religiosity does not significantly moderate the relationship between subjective norms and their tax non-compliance intention. Thus, this finding fails to support H2a. The insignificant results indicate that Malaysian SME owners' religiosity does not affect the influence of the subjective norms on the tax non-compliance intention.
Third, as for perceived behavioural control, the path coefficient and p-value show that religiosity significantly moderates the perceived behavioural control and tax non-compliance intention relationship, thus, suggesting support for H3a. The finding indicates that the presence of religiosity in the model has strengthened the previous negative relationship between perceived behavioural control and the tax non-compliance intention. It shows that when the level of religiosity is high, the relationship between perceived behavioural control and the tax non-compliance intention of SME owners becomes stronger. This implies that in the presence of religiosity, an increase in the power of control in a non-compliance action will discourage the tax non-compliance intention of SME owners. Interestingly, the finding is in line with that of Kashif et al (2017) which find that religiosity strengthens the positive relationship between perceived behavioural control and the intention to behave ethically. In the context of this study, religiosity is found to strengthen the negative relationship between perceived behavioural control and the tax non-compliance intention. This can be justified by the argument that individuals highly religious values tend to demonstrate a high level of perceived behavioural control (Walker et al., 2012) because individuals with a high level of perceived behavioural control are tend to be motivated by religious teachings and practices (Welch et al., 2006). Based on the argument, SME owners with high religious values, who normally believe that they are accountable for their actions, would thus assess their tax returns to their advantage so that there is no severe penalty charged by the tax authority (Kirchler, 2007).

Conclusion
Tax non-compliance has been investigated and debated for decades. Although attempts have been made by various researchers, voluntary tax behaviour issues have yet to be resolved. There is still a need to further investigate and find ways to solve the issue. Thus, this study provides empirical evidence of extending the TPB model by incorporating Theory of Religious Commitment. As a result, hypothetical scenarios are formulated to provide answers to achieve the research objective. However, this study is also subject to several limitations. The constructs determining the tax non-compliance that have been incorporated in this study are not all-inclusive. Other influencing constructs or techniques may be relevant but not incorporated in the theoretical framework of this study. Future research should also consider economic factors as well as other non-economic factors which are not covered in this study in solving the tax non-compliance issue, particularly by focusing on the behaviour of SME owners in dealing with tax matters. Next, future research should also consider the mixedmethod approach in enhancing the understanding of tax non-compliance by providing complementary explanations since the use of mixed-method is limited and should be managed cautiously since respondents will feel uncomfortable due to the anonymity issue. Although this study encounters some limitations, it offers evidence of the main effects and the moderating effects of religiosity in explaining and enhancing the understanding of SME owners' tax non-compliance behaviour. Therefore, this study is expected to provide meaningful contributions to the government and tax authorities at enhancing the tax administration and increasing taxpayers' level of compliance.