GST fraud prevention to ensure business sustainability: a Malaysian case study

Purpose – This study provides in-depth explanation of Goods and Services Tax (GST) fraud prevention towards sustainability business. Design/methodology/approach – This study applies a qualitative research method, i.e. case study, to address the specific research objective. Findings – The finding revealed a GST prevention model towards sustainable business. The finding shows that it is pertinent for the government to set preventive strategies in order to retain sustainable income for the government. Two essential dimensions emerged in the findings to support preventive strategies, namely macroand micro-level measures. Practical implications – The findings of this study provide managers, investors and policymakers with evidence to what extent GST fraud could be minimize in order to safeguard government source of revenue and retain sustainable business in a country. As GST is an important source of revenue for the government, it is thus crucial to prevent fraud from occurring. Originality/value – Past studies have primarily focused on GST implementation from the perspective of service tax effectiveness and efficiency. However, this study examined the impact of GST fraud to determine measures that could ensure service tax sustainability using preventive strategies, in turn, introducing to the existing literature on indirect tax.


Introduction
Goods and Services Tax (GST) is commonly considered an effective way to raise government revenue (Ernst and Young Global Limited, 2019;Haron and Ayojimi, 2019;Shaari et al., 2015). As a matter of fact, studies have showed the significance of GST as a source of government income (Highfield et al., 2018;Haron and Ayojimi, 2019;Shaari et al., 2015;Sanusi et al., 2015;OECD, 2019). Due to this socioeconomic importance, many countries worldwide have implemented GST. To be exact, there are 19 countries in Asia, 7 countries in ASEAN, and 53 countries in Europe implemented GST (https://www.treasury.gov.my/pdf/gst/list_of_ countries.pdf).
However, the rising trend of GST fraud has become one of the global concerns. According to the KPMG Report (2017), although many countries have modified their GST systems to overcome fraud problems, GST frauds still persist. In Malaysia, for example, there were many reported GST frauds (Othman et al., 2019: Sanusi et al., 2018. Othman et al. (2019) found that during the GST implementation in Malaysia, from 2015 to 2018, GST frauds, not only increased significantly, rather the reported cases were complex and sophisticated. Malaysia has abolished GST and since September 2018, the previous government, under the leadership of Tun Mahathir Mohamed, has reinstated Sales and Services Tax (SST) and this continues with the newly elected government, under the leadership of Tan Sri Muhyiddin Yassin. Regardless of the change of the government, service tax fraud has always been a concern and the importance of mitigating measures could not be ignored. Ironically, little is known about this phenomenon, particularly from the aspect of mitigating measures.
From the academia perspective, little focus is given to investigating the preventive measures. Previous studies have commonly focused on the efficiency and the effectiveness of the system (Breen et al., 2002;Kasipillai et al., 2015). Due to the limited knowledge and the rising concerns on GST frauds in the Asian region and globally, we investigate how the Royal Malaysian Custom (RMC), established as enforcement agency by the government, prevent GST fraud. We were motivated by three key factors: first, the findings added value to the field of study as previous studies did not contemplate on GST fraud prevention. Thus, lesson learned from Malaysia contributes to advance the knowledge on mitigating GST fraud. Secondly, not many have explored fraud from the perspective of tax enforcement agency, indicating that this topic has not been researched sufficiently. Our inquisitive mind motivated us to inquire from the context of enforcement, extending the understanding of GST frauds from the insider perspective. Third, GST have been studied mostly through the quantitative methodology. We believe fraud studies required a qualitative approach to provide rich information on the complexity and sensitivity of the phenomenon. Conducing this phenomenon using qualitative methodology contributes towards extending the reasoning to interpret human action in the context of the study. Last but not least, the findings would be useful to other countries who adopted GST, particularly in the ASEAN countries, specifically and worldwide in general. Although Malaysia has abolished GST, the experience and knowledge presented by the voice of our respondents would be of great value to the practitioners and academia.
Based on the discussion above, this paper aims to discover the fraud prevention strategies, from the perspective of the RMC. This paper provides knowledge pertaining to service tax, particularly GST, where it explains the preventive measures. The integration of GST fraud prevention is the first line of defence towards sustainable businesses. GST prevention strategies help mitigate fraud and ensure business cost-effectiveness. Considering the significant of GST in many countries, particularly in Asia, it is pertinent for studies to be conducted to investigate the detrimental effect (such as fraud) towards sustainable business and economy growth. The experience of Malaysia could be the focal point of other countries worldwide. Thus, the findings of this study is useful to other GST implemented countries.
This study contributes to the literature by addressing four important issues: First, it should be noted that, while many studies have investigated service tax fraud, some areas are still insufficiently explored, especially fraud preventive aspects. Therefore, this study provides opportunities to build an in-depth understanding of the nature and extent of service tax fraud, contributing to the development of a broad prevention strategy. Second, this study used a socially constructive methodology that adheres closely to interactions with social reality to provide more insightful knowledge on the subject matter. Third, this study confirms the measures used to deter non-compliance and risks pertaining to service tax fraud, while asserting the importance of tax revenue in ensuring the economic growth and fiscal health of a country. Finally, this paper presents a useful approach for fraud detection and prevention using an analytical model that will help policymakers and regulators make informed decisions and take the most effective actions to combat fraud. Apart from providing in-depth knowledge on GST, the findings contribute to practitioners, particularly, enforcement authorities to solve the problem of government revenue loss and to retain a sustainable income for the nation. The findings benefit countries that have adopted not only GST, rather any nations that implement services tax such as SST, or VAT.
The rest of the paper is organized as follows: Section 2 provides a brief review of the literature on tax fraud, Section 3 explains the research design, and Section 4 summarizes and analyses the results. Meanwhile, Section 5 presents a discussion of the results. Finally, Section 6 concludes this paper.
2. Literature review 2.1 GST In Malaysia, per the statutory order published in the Federal Gazette, the GST Bill was introduced on 1 April 2015, mandating a GST level of 6%. Mooted by the Federal Government in 2011, the revenue-raising initiative was announced by the Malaysian Ministry of Finance in Malaysia's Budget 2010. It is important to mention that in Malaysia, GST was replaced with the SST in 2018.
However, considering the statement made by former Prime Minister of Malaysia, Tun Mahathir Mohamad and the scholarly views on the benefits of GST, comparative to SST, there is possibility of reintroducing GST. The former Prime Minister of Malaysia, Tun Mahathir Mohamad, has indicated that GST may benefit the country, compared to SST, the government will conduct a study on the possibility of reviving it (Riar, 2019). In addition, some researchers opined GST is a good service tax system, as GST reportedly contributed MYR 41.2 billion in 2016 compared to the predicted MYR18 billion contributed by SST (Ching et al., 2017). Ching, Kasipillai and Sarker (2017) also claimed that GST would cast a wider revenue net by capturing the economic chain, including the shadow economy. Empirical evidence had showed a positive significant relationship between GST and economic growth (Vendakadasalam, 2014). For instance, Vendakadasalam (2014) found there is significant influenced between GST and national growth in the ASEAN countries. He found that the final consumption expenditure of households and the general consumption expenditure of the government were positively significantly related to GDP, implying GST as a service tax is essential in ensuring sustainable growth for businesses. Elsewhere, OECD (2019) study was conducted to assess the tax revenues of seventeen Asian and Pacific economies and it found that GST, is an important and increasingly promising source of revenue. The OECD (2019) report stated that GST contributed to more than twenty-eight per cent growth of Pacific economies. This statistic showed that GST is an important source of revenue for many countries and provide significant economy growth for many nations (Tan et al., 2018). This argument emphasizes on the important of the present study.
Other studies (such as Sanusi et al., 2015) also supported the move to re-establish GST in Malaysia, stating that GST was first established to mitigate transfer pricing manipulation. GST could still make a comeback partially because of the cascading cost and price effects on goods and services and the lack of clarity around the current consumption tax system in Malaysia, namely SST.
Nonetheless, regardless of the benefits of GST, studies have found it has inherent weakness. For instance, study has found that the scale of GST fraud is considerably widespread. Othman et al. (2019), in their work on GST fraud, found that GST fraud is detrimental to the economy and sustainable business. Keen and Smith (2006) also found that service tax such as GST was indeed vulnerable to fraud and had caused a massive loss in government revenue. Countries also reportedly experienced a sizable loss of revenue due to service tax (Koh, 2018). In a nutshell, GST is both blatant and stealthy. The emerging Asian economy, in particular, is experiencing unbridled growth in the scale of fraud. The fraud is becoming more widespread and insidious. The evasion of service tax such as GST has been a major obstacle to tax administrators around the world, Malaysia notwithstanding. The various types of violations include fictitious declaration, fraudulent claims, and evasion. In Malaysia case, more than 5,000 companies have reportedly been involved in GST frauds since it was implemented on 1 April 2015 till 2017 in Malaysia (Bernama, 2018). The impact of the hidden economy in Malaysia has led to substantial losses in tax revenue.
Early studies on the GST system were often too general, focusing particularly on tax implementation aspects (i.e. the awareness of and readiness for, GST implementation). Many previous studies on the Malaysian GST focused exclusively on the determinants of GST. For example, previous works investigated the awareness of and preparedness for, GST (Ching et al., 2017), GST governance (Sanusi et al., 2015;Chen and Taib, 2017) and the effectiveness of GST implementation (Ling et al., 2016;Mohd Yusof et al., 2014). So far, only a few studies have examined the prevention strategies for GST fraud. An overview of these studies, primarily in the Malaysian context, is given in the next section.
Past studies focused on taxpayers' non-compliant behaviour from the view of governance factors and economic factors (Kiow et al., 2017;Teng and Manual, 2016). For example, a fairly recent study conducted by Kiow et al. (2017) found that taxpayers' compliance behaviour increased with increased government transparency, indicating that the ethical perceptions of individual taxpayers influenced their decision to comply with paying tax and that public governance and government transparency influence their ethical perceptions of the tax (Kiow et al., 2017). In regard to economic factors, Teng and Manual (2016) found that the noncompliance behaviour of individual taxpayers had a significant effect on tax rate, and was insignificant for other economic factors such as tax rate, income level and inflation level, suggesting that these economic factors do not influence the non-compliance behaviour of individual taxpayers.
Generally, the scope of studies on GST was very much limited to compliance readiness. For instance, Ching et al. (2017) investigated the major challenges encountered by SMEs in Malaysia. Their study examined GST-compliance readiness, compliance costs and attitude. Using the mixed-method approach, the study found three main factors that affected taxpayer attitude towards GST, which are the taxpayers' perception of fairness, the GST readiness of businesses, and external factors that affect business operations. The study found mixed results regarding attitude towards GST implementation, where some SMEs expressed negative views while others viewed GST implementation as helping the authority to reduce illegal and unregistered businesses in Malaysia, while also improving transparency in businesses. Hence, reducing creative accounting and tax report manipulations. Nonetheless, their findings also indicated compliance costs as the factor that contributed the most to a negative attitude towards GST implementation, where it is seen to increase the cost of doing business in Malaysia besides being a burden.
Other studies, such as Mohd Yusof et al. (2014), have also highlighted the impact of individual taxpayer compliance behaviour and GST governance (Sanusi et al., 2015). However, past research findings on tax non-compliance in Malaysia have been inconclusive to such a degree that most studies that had broadly investigated service tax in Malaysia (i.e. GST) are still insufficient. According to Mohd Yusof et al. (2014), very few studies on tax noncompliance in Malaysia have examined the factors that could lead to tax fraud or noncompliance, such as penalty rate, marginal tax rate, foreign ownership, financial liquidity, company size and industry type. As for studies on GST compliance, studies such as Ling et al. (2016) examined GST compliance among Malaysian consumers. They found that the price of goods and services, government subsidies, and income inequality significantly affected GST compliance. Meanwhile, the price of goods and services had a negative correlation with GST compliance, suggesting that the public was responsive to the after-tax price, which in turn, negatively impacted GST compliance. Meanwhile, government subsidies were found positively related with GST compliance. The study also found an inverse relationship between income inequality and GST compliance. Mohd Yusof et al. (2014) also investigated non-compliance between Small and Medium Corporations in Malaysia. The study found that marginal tax rate, company size, and industry type significantly affected tax non-compliance. Also, the finding revealed that the construction industry was the most non-compliant, suggesting that the high tax losses due to non-compliance impacted the government's revenue from tax income.

Tax fraud in Malaysia
Past studies on fraud are prone to examine direct tax. Rashid et al., 2014, for instance found, Shariah-compliant firms were found to use creative accounting to manipulate financial reports. The studies examined the influence of tax reporting red flags on the values of Shariah-compliant firms listed on Bursa Malaysia from 2001 to 2012. It appeared that these firms focused on various ways to commit tax fraud; however, in the case of creative accounting, tax avoidance was carried out in ways that did not violate the law. Although such investigations have reported many interesting results, additional research is necessary to provide important insight into the regressive nature of service tax, such as GST.
Other studies that focused on GST fraud was conducted by Othman et al. (2019). The study used a qualitative methodology to find GST fraud occurrences in Malaysia. They found many types of GST fraud, namely falsifying claims, sales manipulation, non-submission of GST, failure to register, GST avoidance, and carousel fraud. The study also found that GST fraud is becoming more and more sophisticated, GST fraud commonly occurs among medium-sized corporations and those with a mediocre educational background. Pressure, opportunity and rationalization were reportedly reasons for people to engage in GST fraud. The study also cited other reasons such as legal limitations, a lack of investigative skills in detecting fraud, limitations of fraud detection systems and the sophistication of the method, all of which create opportunities for GST fraud. The findings support the fraud theory, showing that pressure, opportunity and rationalization were the three factors that led to fraudster behaviour. Their study indicates that understanding GST fraud is important.

Fraud studies outside Malaysia
Many studies have investigated fraud prevention and detection with a focus on red flags (Bierstaker et al., 2006) and tax evasion (Stavja nov a, 2014). Cressey (1953) developed the three-factor model explaining red flags based on the theory underlying the Fraud Triangle Model by interviewing prison inmates in the United States. The study found that for a person to perpetrate fraud, he or she must perceive an opportunity; feel pressure; and then rationalize. The Fraud Triangle Model, prompting many academic studies on accounting fraud to refer to the same model to explain the factors that lead to fraud.
2.3.1 Prevention and detection of GST fraud. With respect to prevention measures, it is important to mention that the prevention and detection of fraud started gaining traction in accounting studies since the 1980s. Many studies (Loebbecke et al., 1989;Hackenbrack, 1993;Hylas and Ashton, 1982;Wright and Ashton, 1989) have investigated methods to prevent and detect fraud. Some found several mechanisms for detecting and preventing fraud such as analytical procedures and audit procedures. Internal control played a bigger role as a prevention measure three decades ago. Auditor reliance at this time was also greater. In the early 2000s, studies found traditional mechanisms to be good solutions to counter GST. Meanwhile, studies in this era focus more on the service tax system as a strong defence against fraud, therefore, implying that technology is not being used extensively to solve fraud. Such studies mainly focused on the administrative of the system (Ruffles et al., 2003;Borselli, 2008).
However, in later years, researchers such as Podlipnik (2012) found that administrative systems and technologies were some of the ways to prevent missing trader and carousel fraud. Podlipnik (2012) found that a more rigorous tax inspection would better catch the person participating in the fraud, as well as prevent the cooperation between tax authorities in different states and the refusal to deduct service tax. In addition, Podlipnik (2012) identified other fraud prevention strategies in the form of technological solutions such as real-time VAT, digital VAT or the elimination of the zero-rating of intra-community business to business trade, as well as through closing the tax gap by recovering VAT from the person participating in the transactions. Earlier researchers such as Slemrod (2007) also found a solution to prevent tax using efficient analytical tools and optimal taxation enforcement policies. A study was done in New Zealand to seek prevention through education. Krever and White (2007) found a low understanding about GST, implying that educating the people about GST is one way to prevent GST fraud. In Europe, Ainsworth (2010) found technology as a solution to GST or VAT fraud, particularly missing trader fraud. Their findings showed that technology such as real-time VAT and the VAT Locator Number were some of the most effective prevention measures for missing trader fraud.
In the context of tax fraud, the author's reading of the literature suggests that studies on tax fraud also focused on the behaviour of individual taxpayers. For example, Torgler (2008) conducted a study on taxpayer behaviour and found that individuals engaged in tax fraud to lower their taxes. Nonetheless, Torgler (2008) analysed ways to measure tax fraud, finding two mechanisms of detection, namely direct tax measures and indirect tax measures. These findings help the authorities to identify fraud related to tax that leads to revenue loss for the government. Using a survey, Torgler (2008) found that audit data obtained from the audit process could be used as a direct measure for detecting tax fraud. Previously, Bierstaker et al. (2006) found several common methods to combat and prevent fraud, such as firewalls, viruses, passwords and internal control, whereas continuous auditing, discovery sampling, data mining, forensic accounting, digital analysis and software were less often used. However, the study was done in 2006, hence explaining why the latter methods are not commonly used as prevention mechanisms.
Considering the benefits of GST for overall economic growth, the relatively complex nature of the GST system, and, in particular, the cascading tax effect, which include the advantages of certain grey areas and anomalies, it is interesting to note that a limited number of studies have been conducted on the prevention strategies aspects of GST.

Actor-network theory
An Actor-Network Theory (ANT) is a theory that explains the complexity of a social order. As Cresswell et al. (2010) explained, ANT has its own epistemological and ontological position, consisting of networks that include humans, things, ideas and concepts. ANT referred to these networks as "actors". Cresswell et al. (2010) interpretation of ANT, as a social order that contains multiple networks, that are inter-related, is most appropriate in explaining the assumptions of our study, there are: first, Cresswell et al. (2010) associated the social process with human and non-human, as they believed that there is an inter-relatedness between human and non-human in shaping social processes. Second, Cresswell et al. (2010) explained ANT from a wider perspective, taking information technology as the concept of social processes, as such helps to conceptualize the dynamic relationships between humans and technology. Third, Cresswell et al. (2010) defined actor as the "source of an action regardless of its status as a human or non-human" (p. 2). They attest that inanimate things (e.g. such as technology) can also have agency, thus suitable to be call actors. We concur with Cresswell et al. (2010) that reality can only materialize when there is inter-relatedness between the inanimate things (technology) as agents and humans as actors. Without one, the other could not function, as the combination of both creates the actors the possibility to act. We refer to Cresswell et al. (2010) as their rationale supported the inter-relatedness between the two agents described in our study (technology and human). Congruent with Creswell, the association between technology and human, represents the social reality in mitigating process. ANT accepts that technology is part of the social reality rather than as impact factor of the social process and that technology emerged from social interests and shape social interactions. In this paper, we reflect on the potential contributions of ANT and its applicability to investigating GST fraud mitigating processes. We illustrate the key issues through reflecting on our investigation of the inter-relatedness of combining human and technology as the process of mitigation.

Methodology
The qualitative methodology was found appropriate for this study because empirical research in this field is still in its infancy. Levi (2015) claimed that fraud studies find a more fertile ground in qualitative methodology.

Purposive sampling: case study
The case study design is appropriate for the current research, as it aims to obtain an in-depth understanding of GST fraud from the perspective of the enforcement agency, focusing on the prevention strategies. A single case study design was deemed appropriate, as RMC is the authority mandated to combat service tax fraud in Malaysia. Yin (2014) explained that a single case study design is appropriate when examining critical or typical case. This study is seeking a depth of understanding of GST fraud prevention strategies from the perspective of RMC, fit into the description of critical or typical case as described by Yin (2014). Kohn (1997), on the other hand, posited that the case study method could be employed to explore new areas and issues for which little theory is available or measurement is unclear. As such, the case study method was found the most appropriate.
The scope of this case study was limited to methods of preventing service GST fraud. This study aimed to examine the process of prevention strategies from the perspective of an enforcement agency, RMC. The selection criteria for the sample was based on the expertise of the respondents. These specific selection criteria were established to seek sources that are familiar with the phenomenon of GST fraud.
According to Patton (2002), purposeful sampling requires access to key informants in the field who can help identify information-rich cases. RMC was chosen as the purposive sampling target because it is the main enforcement agency in Malaysia that deals with service tax previously and SST presently. Rapley (2014) explained that it is crucial for qualitative studies to select the sample case to ease the analysis process. He elaborated that the central resource for which sampling decisions are made is based on the specific people, situation, or site, as the sample offers specific biased or information-rich data.

Triangulation of data sources
Sampling involves selecting cases and selecting data sources "that best help the researcher understand the case" (Stake, 1995, p. 56). Guided by Stake's (1995) recommendation, three main data sources were used to help the authors understand how to prevent GST fraud, namely (1) a focus group, (2) documents and (3) face-to-face interview.

Focus group.
A focus group was conducted as part of the data collection method to triangulate and provide information-rich data. A focus group helps the researcher acquire and understand the experiences of the administrators in their surroundings. In this case, the administrators are the RMC officers who oversee the GST compliance operations. Thus, RMC administrators from Kedah and Perlis were invited to participate in this study. The criterion for the sample selection was based on the criteria that the participants must be willing to participate in the study as described by Patton (2002). The criteria used for the sampling selection was based on Patton's (2002) reasoning: The logic and power of purposeful sampling lie in selecting information-rich cases for in-depth study. (Patton, 2002, p. 230).
The aim of the focus group was to obtain more insights into the administration of taxes, especially the strategies, systems, processes and practices used to prevent GST fraud. The RMC officers from both Kedah and Perlis were selected based on the relevance of the topic of study and their willingness to participate in the study. As mentioned, attempts were made to include other states (altogether there are twelve) but to no avail.
Each RMC state administration was divided into discussion groups of six to seven officers per group. In total, 26 officers/administrators from various GST unit, i.e. enforcement, compliance and operations, were involved in the focus group discussion. A leader was assigned to each group, and questions were posed to each leader. Additionally, each group had a moderator to initiate the discussion. These moderators were given the task of opening the discussion with questions such as: What are the most commonly perpetrated types of GST fraud? What strategies do you use to prevent and investigate GST fraud? How do you gather evidence? The discussions lasted for about two hours and were recorded on tape. The taperecorded discussions were transcribed verbatim and analysed to allow patterns and themes to emerge; hidden information uncovered from the otherwise confusing data is very useful for refining and redefining GST fraud mitigation strategies and action plans. Based on the data, a model that integrates GST fraud prevention, detection and mitigation was developed.
3.2.2 Documents. This study also relied on documents to obtain reported cases of GST fraud. Documents on GST fraud perpetrated from 2015 to 2016 were gathered from the RMC. In addition to the interviews, the data from these documents gave the researcher insights into the common types of GST fraud in Kedah and Perlis.
3.2.3 Interviews. To obtain both retrospective and real-time accounts, primary data were also collected through face-to-face interviews with GST investigation officers (IO) and the director of the GST unit of the RMC in three states in Malaysia, namely Kedah, Perlis and Putrajaya. The respondents were selected based on their suitability to answer the questions. As suggested by Silverman (2005), it is important to select the right individual (in this case, officers from RMC) for the interviews because engaging with the right individual will provide depth of data and a rich understanding of GST fraud prevention measures. Moreover, there are two main logics associated with these decisions in primary research as well as research synthesis: data saturation and data sufficiency (Suri, 2011, p. 72). Reaching saturation in data collection will decrease the risk of misinterpretation (Stake, 1995).
The profile of the respondents of the face-to-face interviews is listed in Table 1. Each interview lasted approximately two hours. Respondents selections based on the position of the individual since a person's position in an organization determines his skills and expertise in the area. The officers interviewed were from the enforcement division and the administrative division. It is important to note that the division of RMC did not change when the system changed from GST to SST. The officers in charge of the fraud division, as depicted in Table 1, retained. Patton (2002) advocated that purposeful sampling requires access to key respondents in the field who can help unearth information-rich cases. The selection of individual interviews was based on Patton's (2002) homogenous sampling strategy, where the sample was selected based on commonality of the respondent. According to Suri (2011), homogenous samples (practitioners) are appropriate for research involving a phenomenon that has direct implications on their practice as explained in the sample selection section, which fit into the description of our study.
To gain more insight into the research question (How to prevent GST fraud?), the questions posed to the investigation officers and administrative officers were "How do you trace GST fraud?" "What are the strategies used to detect fraud?" and "How do you prevent or anticipate GST fraud?" Then, the conversations were tape-recorded and each interview was subsequently transcribed verbatim. This study followed Kowal and O'Connell's (2014) suggestion in that the same person transcribed the interview data and analysed the data. Seidman (2006) was reluctant to establish an exact sample size; he offered criteria to guide the qualitative researcher to determine the point at which enough interview data have been attained, sufficiency of data and data saturation. The study adopted the said criteria, where sufficiency of the interview data, was reflected through a saturation of the information, evidenced by the repetition of textual themes obtained from the research respondents. In the context of data saturation and sufficiency, following the focus group, a face-to-face interview was conducted with respondents from Putrajaya, which is the RMC Head Office. The data was found sufficient and was deemed saturated when a similar pattern of description emerged with the last few respondents (R6 and R7). At that point, the data was deemed saturated. Strauss and Corbin (1994) have described data analysis as a continuous interplay between data collection and data analysis. Additionally, Stake (1995) stated that qualitative research capitalizes on "ordinary ways of making sense". Thus, the ultimate reason is to make sense of the data. This study adopted Stake (1995), who advocates the ordinary way of making sense of the data. However, in respect of the process of analysis, this study adopted Gbrich (2012) qualitative data analysis, which is based on block and file technique of analysing the data. Each process involves the stages of abstraction in which, at each stage, the data is further reduced to form smaller groups based on similarities. The process is to reduce the data to meaningful groupings of themes.

Results
The data analysis showed that RMC employed multi-method strategies to prevent GST fraud. Section 4.1 discusses the GST fraud prevention strategies to ensure sustainable revenue for the government.

GST preventive strategies
Based on the data analysis, three key components of GST fraud prevention strategies employed by the RMC were identified. These strategies involved: (1)  the Face-to-face interview a cohesive culture of honesty, (2) conducting GST awareness and education programmes and (3) initiating proactive and reactive measures. Each strategy is discussed further in the following sections. 4.1.1 Creating and sustaining a cohesive culture of honesty. Our analysis indicates that a cohesive culture is crucial, especially at the early stage of GST implementation. Our respondents voiced that cohesive culture built public trust and this is essential as public trust can improve the perception of the RMC. Our analysis revealed that RMC is committed to acting in the public interest and has taken several measures to cultivate an atmosphere that promotes transparency, accountability and integrity as part of it creating cohesive culture of honesty. Our analysis revealed that GST compliance activities such as policies and programmes, not only improved the RMC's practices but also created its culture of honesty. For example, GST legislation, such as the GST Act 2014, GST Orders, GST Regulation 2014, and GST Guides, including the RMC's General Guides and Industry Guides, among others, are some specific examples of how RMC control and facilitate compliance and minimize the risk of opportunistic GST fraud. Codified law is a form of mitigation strategies, create GST compliance, thus creating cohesive culture. Apart from that, laws and regulations create expectations of punishment, as such, deterring dishonest behaviour and a permissive environment. For example, offenders are punishable by fine or jail under the GST Act 2014.
4.1.2 GST education and awareness. Since its implementation, the RMC has conducted many education programmes designed to provide the general public with exposure to and knowledge of the salient features of GST. The focus is not only on improving the taxpayers' knowledge of GST policies and related procedures, such as changes in legislation and noncompliance issues, but ideally to gain public support for tax policies and to foster taxpayers' willingness to comply with the tax rules. The "crime of non-payment' is often viewed as an act perpetrated by ignorant taxpayers. For example, some businesses are not GST registered, possibly due to a lack of understanding or knowledge on how to claim input tax credit on GST incurred, and this, in turn, leads to increased cases of non-compliance. Concisely, taxpayer education is the bridge linking voluntary tax compliance and tax culture. Realizing this, the RMC has increased its strategies for designing and providing the best possible approach for disseminating accurate and up-to-date taxation information for the purpose of taxpayer education (e.g. awareness campaigns through regular seminars, training and via its GST ambassador programme).
4.1.3 Proactive and reactive measures. Two main themes emerged; proactive and reactive measures, as preventive strategies which could further be divided into technology-driven methods and human-driven methods. The technology-driven method includes the use of technology such as tax management tools, while the human-driven method involves covert examination, internal controls and analytical analysis. 4.1.3.1 Use of technology. Technology use appeared as a theme that describes the process of preventing GST fraud. The use of technology represents the tax management tools used by the enforcement officers to prevent GST fraud. This study identified tax management tools as a technology-driven method. 4.1.3.2 Tax management tools. Tax management tools appeared as a theme that was associated with technology-driven method of mitigating GST fraud. The RMC used technology-driven method to prevent tax fraud. According to our respondents, apart from a cost-effective mechanism, technology-based tools assisted the officers to detect anomalies or unusual transactions. For example, the RMC used GENTAX and ACL software as technology tax management tools. GENTAX is tax-processing system designed to support the RMC to combat fraud. GENTAX facilitates compliance through proactive fraud auditing, allowing RMC officers to identify and analyse trends or anomalies via critical observations. According to our respondents, anomalies can be detected using a data collection system (DCS), which is a computer application that facilitates the process of data collection. This allows structured information to be gathered in a systematic fashion, subsequently enabling the analysis of the data obtained. For instance, one of the respondents described unusual patterns typically reflecting anomalies and possible ambiguity, so the trends could be an indicator of (possible) suspicious activities. In such a case, officers can efficiently flag suspicious transactions and address immediate discrepancies. ACL is another tool that was commonly used.
Operating a similar function as GENTAX, ACL is a data analytic software that is effective as it offers analytics data pattern, thus form as a control mechanism. The officer in-charged observed the pattern appeared in the system and any unusual pattern will be investigated further through inquiry. Our respondents claimed that ACL works on a large number of data using minimal time, making data processing effective and efficient. According to our respondents, ACL is extensively used at the RMC, as it is very effective in preventing and detecting fraud. Our finding indicates that ACL is used to identify anomalies (see Table A1: Quote no. 1, Quote no. 2). 4.1.3.3 Covert Examination. Covert examination appeared as a theme that describes RMC's process of mitigating GST fraud. Covert examination is a human driven method. Our analysis indicates that cover examination consists of test shopping or mystery shopping. Test shopping or mystery shopping is the most common method used by RMC (see Table A1: Quote no. 6). The study data revealed that, during the initial stage of implementation, the RMC would use test shopping. This process of prevention involves randomly selected samples from businesses. Several investigation officers will be present at the premise and conduct covert operations (see Table A1: Quote no. 7, Quote no. 8). This study found that the success of covert examinations was based on the capability and ability of the covert officers to prevent fraud. 4.1.3.4 Internal Control: audit strategy. Internal control is another human-driven method. The RMC officers use three internal control strategies to prevent and detect fraud: verification audit, in situ audit, and field audit. The interview data and the focus group data delineate auditing techniques such as verification audit, while others focus on field audit and audit in situ as preventive measures (see Table A1: Quote no. 8-11). Field audit or fieldwork refers to hands-on activities in which investigation officers or compliance officers complete their duties or activities in the field. The RMC uses this mechanism to reduce GST fraud based on monitoring tools or activities. This method is also used proactively to seek information or detect potential GST fraud among traders. The aim is to prevent fraud before it happens or to detect fraud before it becomes worse. This method is one of the most efficient discovery mechanisms because it is conducted randomly. Another detection mechanism adopted by the RMC is the surprise audit or audit in situ (see Table A1: Quote no. 9-11).

Analytical symptoms
There are three main procedures of analytical symptoms: 1) analytical analysis, 2) risk management and 3) behavioural pattern.

Analytical analysis
The RMC applies an important procedure to detect GST fraud. Analytical analysis can be used to detect red flags in risky areas, for instance, accounts and records of businesses. Basically, the RMC adopts two types of analytical analysis: surveys and statistical analysis. Statistical analysis relies on audit tools, such as patterns in graphs (see Table A1: Quote no. 12, Quote no. 13, quote no. 14, Quote no. 15, Quote no. 16).

Conduct risk management
Risk management has emerged as a theme that represents the RMC's mitigating fraud (see Table A1: Quote no. 17-20). According to our respondents, the purpose of risk management is to review GST compliance risks. Our data disclosed three stages of risk management, there were: the identification of risk, the assessment of risk and the evaluation of risk. At the identification stage, RMC officers analysed the nature of risk in the businesses that were registered in the system. In the assessment stage, the RMC officers focused on assessing whether the risks are significant or otherwise. For example, a company that deals with import and export would be categorized as a high risk. If found the company is of high risk, the system would then proceed to the third stage, which is the evaluation stage. At the third stage, the RMC officer would conduct an audit investigation. At this stage, the officers would proceed with interviews, aimed towards detecting anomalies if any.

Behavioural pattern
The RMC observes behavioural patterns to detect GST fraudsters. There were two core behavioural patterns noted. Firstly, through the data gathered from the system. Secondly, through the personality behavioural patterns gained based on the RMC officers' observation of the fraudster. The former involved the RMC officers worked with the available documents (i.e. the records compiled by officers), to detect if there is any unusual predictive pattern that emerged of the fraudster business. The later involved surveillance and observation by the RMC officers. During the process of surveillance and observation, the RMC officer noted unusual behaviour or sudden unusual business transactions. Subsequently, the investigation officers performed their investigation based on their analytical observation of the behavioural pattern from the two sources.
4.5.1 Reactive measure. Our analysis revealed some dynamics between the enforcement division, the compliance division, and stakeholder engagement, implying that human driven method is part of the essential mitigating strategies in preventing GST fraud. Upon receiving tips and complaints from stakeholders, the RMC enforcement division will conduct an initial investigation, following that, the enforcement division officers will collaborate with the compliance unit. Following section explain the process of how the reactive measure in preventing GST fraud is conducted. 4.5.1.1 Tips and Complaint as a prevention measure. Tips and Complains are one of the effective reactive preventive measures. Tips and complaints commonly stem from stakeholders (see Table A1: Quote 3 and Quote no. 4). The data in this study revealed that the RMC occasionally rely on anonymous tip-offs and whistle-blowing. According to the respondents, however, during the initial implementation stage, not many tips and complaints were received. Nevertheless, over a period of time, the RMC noted an increase in the reporting of GST fraud and anomalies.
The reasons for the above case include: (1) More tip-offs and user-friendly channels for whistle-blowers, such as hotlines and websites, which provide quick access for reporting suspicions of GST fraud.
(2) The public is now aware of the consequences of tax manipulation related to GST and recognize the symptoms of manipulation.
(3) Enforcement agencies are taking on more active roles to disseminate GST information, especially in regard to non-compliance, to the general public.
It is interesting to note that, after several years of GST implementation, tip-offs and whistleblower complaints in cases of GST fraud are on the rise. This traditional reactive approach has long been used to detect fraud. Tip-offs reported by whistle-blowers are the most effective source of information to combat GST fraud because it reduces the time necessary to unearth fraud (see Table A1: Quote 5). In fact, investigations could be carried out immediately following complaints from the public. The current findings indicate that GST awareness not only increases public literacy and acceptance but also instils public confidence, leading them to perceive the authorities as fair and equitable. Hence, infused by strong moral and ethical values, the public is encouraged to step forward and lodge complaints with the RMC.

Discussion
This article aimed to discuss and interpret how RMC mitigate GST frauds. Previous studies (such as Halbouni et al., 2016;Zakaria et al., 2016), stated that fraud could be curtailed through prevention measure. To be more specific, previous studies such as Bolton and Hand (2002) claimed that fraud prevention consists of strategies that are developed to prevent GST fraud. In this section, this study highlighted three main GST prevention strategies employed by the RMC, which are: (1) a cohesive culture of honesty, (2) GST awareness and education programmes, (3) implementation of proactive measures. These strategies can be grouped into macro-level and micro-level, where the former includes cohesive culture of honesty, GST awareness and education programmes, while the later consists of implementation of proactive measures. This finding is congruent with Gunasegaran et al. (2018). Gunasegaran et al. (2018) found creating a culture of honesty and proactive measures resulted in preventing fraud. Further, RMC uses human-driven method and technology-driven method to counter GST fraud.

Macro-level measures and micro-level measures
Two essential dimensions emerged in the findings to support RMC's preventive strategies, namely macro-level and micro-level measures. Macro-level indicates general measures adopted in the process of preventing GST fraud. To be more specific, the macro-level dimension deliberates on the process of creating and sustaining an honest environment such as creating and sustaining a cohesive culture of honesty, implementing education and awareness programmes, and establishing proactive measures. Meanwhile, the micro-level dimension entails specific prevention strategies that deliberate on the process of preventive control systems using human-driven factors and technology-driven factors to counter GST fraud. Considering the complexity of GST, the sheer volume of GST transactions, and the lack of resources RMC was found to rely on the human capacity to prevent and detect GST fraud. The findings revealed that human capacity, such as mystery shoppers, tip-offs and complaints, field operations, street surveys, proactive and reactive audits, analytical analysis, fraud risk management, and behavioural patterns, were all used as preventive strategies. Equally important, is the technology-driven method, which focuses on the internal system to prevent and detect fraud. Reliable technology-driven systems include datadriven software such as tax processing systems (e.g. GENTAX). Technology-driven measures are efficient and reliable in countering GST fraud. The current analyses of the data indicates that GENTAX is an efficient system that provides behavioural patterns, used to detect GST fraud. Therefore, data-driven software effectively supports enforcement agencies to prevent GST fraud. For instance, GENTAX facilitates compliance through proactive fraud auditing, allowing the RMC officers to identify and analyse trends or anomalies by way of critical observations. The use of technology is supported in the literature. This finding is congruent with previous study (Bolton andHand, 2002: Halbouni et al. (2016), that prevention technologies are the best ways to reduce fraud. Halbouni et al. (2016) for example, found that technology techniques and fraud software are essential to prevent fraud. Further, they found that advance analytical techniques are also an essential mitigating mechanism.
Although the findings of this study found that the RMC adopts more conventional ways of mitigating GST fraud, the use of technology is apparent. The use of technology is paired with human in the process of preventing and detecting fraud. Overall, the findings elaborate on the interplay between humans and technology, which outcome provides area of which Actor-Network Theory (ANT) commonly applied. ANT explains that human and non-human come together and demonstrate agency environment. ANT helps to explain an environment that comprised of humans and technologies as actors (Lupicini, 2014). To be more specific, ANT elaborates on the interplay of humans and technologies. This theory supports our explanation of the macro and micro level factors where relationship between human and non-human (technology driven method) in the context of preventing GST fraud are obvious, thus ANT offers a promising theoretical explanation to the study of tax fraud.
Creating an honest environment is one of the human factors that contribute to effective preventive measure. This finding is congruent with Zimbelman and Albrecht (2012) suggestion that create a positive environment, such as training program, creates a low-fraud atmosphere. For example, the RMC introduced the GST Ambassador Programme. The programme was aimed to educate businesses using a customer-friendly strategy or, in other words, compliance through education, rather than enforcement and punitive actions. This preventive measure helped the public understand the GST process better and thus reduced non-compliance and unintended fraud activities despite reported cases of unintended fraudulent activities. In addition, RMC also highlighted the importance of having internal and external verification. This finding is aligned with previous studies (Persons, 2009;Kaplan et al., 2009) that an effective governance and internal control would results in effective preventive fraud measure. Our findings indicated that internal control such as audit procedure is crucial to prevent GST fraud.

Conclusion
This paper aimed to explain the GST fraud prevention macro-level and micro-level strategies adopted by the Royal Malaysian Customs Department (RMC). The results showed that the RMC adopts various strategies to prevent GST fraud. The data revealed that human capacity and technology are essential in preventing GST fraud. This study provides one key realization: that preventing GST fraud requires an understanding of the system, the process, and, most importantly, the behavioural patterns of fraudsters, i.e. the people. The findings appear to indicate that ANT does play a key role in explaining theoretical aspect of fraud. This can be attributed to several factors. First, ANT deals with interplay of human technological relationships. Second, ANT is a theory that acknowledges the complex interplay of humans and non-humans as actors, simultaneously act upon each other. This mirrors the practical context of preventing GST fraud environment, where involvement of human driven and technology driven are paramount.
This study has several implications: First, the GST prevention strategies emerged from the data, providing effective ways for the RMC and other related agencies to curb fraud. Second, the findings provide guidance for the RMC, as well as agencies that deal with fraud, to better identify fraudsters. Third, examining prevention strategies involving practitioners provided better understanding of current practices to mitigate GST fraud, regardless of the government's decision to implement GST or SST, as both are under the same service tax type. With this understanding, RMC can build upon current practices to create effective service tax fraud prevention model. Enforcement agencies in Asian countries or in any country, particularly those adopted GST, or SST or VAT, could develop a mitigating fraud model. Further, although Malaysia has replaced GST with SST, the experience of Malaysia dealing with GST fraud during its implementation provides a deep understanding about the issue, particularly when the issue is under study. Considering many countries in Asia are adopting GST, it is important for the others to learn from Malaysian experience and sharing our results would benefit these countries to prepare for GST fraud.
The current findings are also subject to certain limitations. This study focused on the RMC as a case study, so the findings are therefore context based; therefore, future research could conduct surveys to generalize the findings. The current study is limited to the prevention and detection of GST fraud in Malaysia, and thus lacks the perspective of the behaviour or profiling of fraudsters. Therefore, it is suggested that future research explore the profiling of fraudsters in more depth.
. . . is using a manual accounting system, maintained and written by hand 4 For example, I am auditing a poultry business; the company has 13 farms and has a monthly claim of 2 million; a 2 million claim in tax, but the company only sells chickens! But this is a zero-rated category! For 3 million chickens, I divided the number by 13 farms: 1 farm should have about 25 thousand chickens. When I examined all the farms, 25 thousand chickens is impossible: perhaps only 2 thousand. We believe that the declaration should be zero rated; this is not right -the input is just not right! Sample of how the tax management tools assist in preventing and detecting GST fraud 5 Complaints are made through email, telephone, or direct to the customs offices. However, the most common is from competitors We go to a restaurant; we call this an audit in situ. We observe, in terms of sales, the concept of the restaurant, customers' response, by our own efforts; for example, we sit there, eat there, etc. From there, we can perform the estimation from the figures, and we add another 10%.
From our findings, we will issue a letter; for example, we will inform restaurant A, after the assessment, that the restaurant has paid less income tax according to the assessment report, let's say . . . restaurant A has to add another 5 thousand, we will issue a letter (of inquiry), within 2 weeks. If the restaurant does not agree, a show case letter is required. If the restaurant does not reply, we will release "the" letter Audit In situ 10 By right, the input-output ratio cannot be more than 10%, because raw materials are zero rated (exempted supplies), unless he pays no taxes, except for canned drinks, cigarettes, and sardines, while others are tax free 11 There was one case, handled by the Compliance Management Unit. The unit conducted a verification audit and interviewed the owner. The owner said that his business was not doing well. The owner submitted a GST statement and declared zero rated. When I took the statement from the compliance officer, the company was registered under caf e, construction. Thereby, I had to hold his documents, to see the fair value of the sale; that was indeed a fraud Verification Audit (continued ) 12 For example, in terms of construction, unlike retail, sometimes you close a sale and sometimes you don't; in the retail industry, we can read graphs; input and output should be in line . . . we can read patterns in graphs. By this means we can observe and focus (on trend lines), and we can even analyse patterns over a period of time. In general, we can detect unusual patterns through graphs; secondly, we go by industry, market level, and we can compare patterns in terms of the market Analytical Analysis

Analytical Analysis
Analytical Symptoms 13 We focus on taxpayers' declarations; taxpayers declare once every 3 months, but in 3 months, a taxpayer only declared a hundred thousand. Thereby, we calculate the mean value 14 When it comes to graphs, we can look at the market, location, and traffic 15 To detect fraud, customs use a comparison strategy; customs will compare the same size of company, using a ratio. If they are not normal, an investigation will be conducted 16 We went to conduct an assessment, then make a comparison within the same industry, i.e. input and output. Thereby we got the ratio. When we checked, it was not normal and we sent someone to investigate 17 Information is obtained based on registered declarations, information such as how much refund did we get And how much tax output is paid by a taxpayer? These are all hard data, while we do an analysis to see if the risk is high Identification of risk

Conduct Risk Management
18 We ask for the documents and we will do a verification if the company is a high-risk company. For a company with a high indicator (based on colour), our officer will be notified; the officer will then request the documents, which we will verify Assessment of risk Evaluation of risk 19 Red (the highest risk), yellow, and green 20 If it is red and high, the system will give an instruction to the lead desk audit. And we will verify it again and ask the company to respond 21 We have a system that indicates high risk Table A1.