1 Ms Pua Om Tee (“the accused”) was charged with several counts of wilfully evading Goods and Services Tax (“GST”) by making false entries in her GST F5 Return over several years, an offence under s 62(1)(b) of the Goods and Services Tax Act (Cap 117A, 2005 Rev Ed) (“the GSTA”). The accused pleaded guilty to three proceeded charges, and the judge (“the Judge”) sentenced the accused to 14 weeks’ imprisonment. During the sentencing hearing, the Prosecution had proposed a sentencing framework for the offence of wilfully evading GST which the Judge rejected. Both the accused and the Prosecution appealed against sentence. A young amicus curiae, Ms Cheryl Chong (“Ms Chong”), was appointed to give her independent opinion on the appropriate sentencing framework to be adopted.

2 I eventually adopted a framework broadly in line with Ms Chong’s proposed framework and sentenced the accused to 24 weeks’ imprisonment, allowing the Prosecution’s appeal and dismissing the accused’s appeal in the process. I gave detailed oral grounds then. I now give the full grounds of my decision.

3 At the time of committing the offences, the accused was the sole proprietor of Wah Ye Advertising and Little Box Event and Exhibition Printing which were in the business of manufacturing builders’ carpentry, joinery, and advertising printing. The accused had instructed her bookkeeper to exclude certain sales transactions undertaken by her businesses, thus omitting the GST output tax of those transactions from her GST F5 Returns with the intent to evade GST. This resulted in a total S$226,902.56 of GST being undercharged over several quarterly periods between April 2013 and September 2016.

4 11 charges were preferred against the accused – I set these out below:

5 The three proceeded charges which the accused pleaded guilty to were the 5th charge, 7th charge and the 8th charge. The collective amount of GST evaded as regards the three charges was $107,911.11. The accused consented to the remaining eight charges being taken into consideration for the purposes of sentencing (“the TIC Charges”).

6 Before the Judge, the Prosecution proposed a sentencing framework for offences under s 62 of the GSTA (“the Original Framework”). This framework was adapted from the five-step framework set out in Logachev Vladislav v Public Prosecutor [2018] 4 SLR 609, with one key modification: the traditional slight-moderate-severe harm categorization in the “harm axis” was eschewed, and instead three “levels of harm” based on the amount of GST evaded was proposed. The Original Framework used the following amounts of GST evaded to delineate each level of harm, that is: (a) up to S$75,000; (b) between S$75,000 and S$150,000; and (c) S$150,000 and beyond. I reproduce this below:

7 I take this opportunity to clarify a point made in my oral grounds of decision. I had stated that the Original Framework was based on the sentencing framework in Tan Song Cheng v Public Prosecutor and another appeal [2021] 5 SLR 789 (“the Tan Song Cheng Framework” and “Tan Song Cheng”) respectively) for offences under s 96 of the Income Tax Act (Cap 134, 2008 Rev Ed) (“the ITA”). This statement was premised on the Prosecution’s submissions on appeal that the Original Framework was “transposed” from the Tan Song Cheng Framework.

8 However, as I wrote these grounds, it became apparent that the Prosecution’s submission was not entirely accurate. When the Original Framework was proposed to the Judge, the decision in Tan Song Cheng had not been released: the Judge heard the parties’ submissions on sentencing on 16 October 2020 whilst the decision in Tan Song Cheng was only released on 9 June 2021. As there was no Tan Song Cheng Framework at the time the Original Framework was placed before the Judge, it would not be entirely accurate to state that the latter was based on the former at that time.

9 Moving on to the proceedings below, the Judge found the results from applying the Original Framework overly harsh: Public Prosecutor v Pua Om Tee [2021] SGMC 25 at [56]–[57]. On this basis, he declined to adopt the Original Framework. Instead, he examined the accused’s culpability and the harm caused by her actions in general terms.

(a) He found that the accused’s culpability was low due to the lack of sophistication in her modus operandi: at [59].

(b) He compared the accused’s offences to several case precedents, which he found to be similar in harm and culpability, where a sentence of two months’ imprisonment was imposed: at [59].

(c) He considered the fact that there were multiple TIC charges which amounted to a significant amount of GST evaded, reasoning that this warranted an uplift in sentence: at [60].

(d) In terms of mitigating factors, he considered the accused’s restitution and plea of guilt as being indicative of her remorse: at [61]

10 Ultimately, the Judge sentenced the accused to eight weeks’ imprisonment for the 8th charge; and six weeks’ imprisonment each for the 5th and 7th charges. He ordered the 5th and 8th charges to run consecutively, with the 7th charge to run concurrently, for a global sentence of 14 weeks’ imprisonment: at [62].

11 Both parties appealed the Judge’s decision. At the first hearing of the appeal on 23 August 2021, the Prosecution maintained the position it took before the Judge and sought the adoption of the Original Framework. On the basis of the Original Framework, the Prosecution sought a higher sentence of 24 weeks’ imprisonment. This was similar to the submission on sentence before the Judge. The Defence sought a lower sentence of 6 weeks’ imprisonment, arguing that the Original Framework should not be adopted, and that even if were adopted, it should not apply to the accused due to the doctrine of prospective overruling.

12 As I noted above at [7], on appeal, the Prosecution explained the Original Framework as being transposed from the Tan Song Cheng Framework. The latter was also adapted from the Logachev five-step framework and similarly eschewed the slight-moderate-severe harm categorization in favour of three “levels of harm” based on the amount of income tax evaded. For reference, I reproduce the Tan Song Cheng Framework below:

13 It is significant that the only difference between the Original Framework and the Tan Song Cheng Framework was that the Prosecution had proportionally raised the sentencing ranges in the latter to arrive at the sentencing ranges in the former, ie, the sentencing ranges in the Original Framework were increased by a factor of two-and-one-third as compared to the Tan Song Cheng Framework. For example, in the Tan Song Cheng Framework, the sentencing range for “moderate” culpability and Level 2 harm was 12 to 18 months’ imprisonment, whilst for the Original Framework it was 28 to 42 months’ imprisonment.

14 The Prosecution explained the increase on the basis of the difference in the maximum sentences found in s 62 of the GSTA and s 96 of the ITA. Section 96(1) of the ITA provides for a maximum imprisonment term of 3 years. On the other hand, s 62(1)(g) of the GSTA provides for a maximum imprisonment term of 7 years, ie, a maximum sentence two-and-one-third times that of s 96(1) of the ITA.

15 At this hearing, it became apparent to me that there were difficulties with the Original Framework, and that a satisfactory solution to the difficulties could not be achieved. The key difficulty was the Prosecution’s approach of a setting the sentencing ranges for the Original Framework by proportionally uplifting the sentencing ranges in the Tan Song Cheng Framework by a factor of two-and-one-third. As noted above at [13]–[14] the uplift was to account for the fact that the maximum imprisonment under s 62 of the GSTA was seven years, whilst that under s 96 of the ITA was only three years. The effect of this proportional uplift was that, all things being equal, an offender who evaded GST would be punished more severely under s 62 of the GSTA than an offender who evaded the same amount of income tax would under s 96 of the ITA. This, as I pointed out to the Prosecution, would only be justifiable if Parliament had intended a stronger deterrent against evasion of GST than income tax under s 62 of the GSTA and s 96 of the ITA respectively (see [29] below).

16 That this was Parliament’s intention was not clear on any of the materials put before me by either side. Neither the Prosecution nor the Defence could explain why the sentencing range in s 62 of the GSTA was more than double that in s 96 of the ITA. Thus, I was unable to accept the Original Framework based on the information that was before me. Accordingly, I decided that the appointment of a young amicus curiae would be of assistance to the court, and adjourned the appeal accordingly.

17 Ms Chong was appointed as a young amicus curiae to give her opinion on several questions that were framed by the court. These questions were based on the points discussed at the hearing on 23 August 2021, and were set out in her appointment letter dated 3 September 2021. Broadly, these were:

(a) Was it appropriate for the court to develop a sentencing framework for offences under s 62 of the GSTA?

(b) If it was, would it be appropriate to apply the Tan Song Cheng Framework? In answering this question, Ms Chong was invited to give her recommendations on the appropriate sentencing ranges, and whether the three “levels of harm” in the Tan Song Cheng Framework should be maintained.

(c) If it was not appropriate to apply the Tan Song Cheng Framework, what would be the appropriate sentencing framework for offences under s 62 of the GSTA?

In answering these questions, Ms Chong was requested to consider not only s 96 of the ITA and s 62 of the GSTA, but also s 96A of the ITA, which incidentally the court in Tan Song Cheng was not invited to consider in arriving at the Tan Song Cheng Framework for s 96 of the ITA.

18 On 4 October 2021, Ms Chong filed a detailed brief setting out her opinion and recommendations. She took the position that it was appropriate for the court to formulate a sentencing framework for s 62 of the GSTA as the full sentencing range had not been utilised in previous cases, and there was a lack of coherence and consistency in the past sentences imposed. However, she did not agree with the Original Framework. In particular, she disagreed with the Prosecution’s use of the Tan Song Cheng Framework as the basis for formulating a sentencing framework for s 62 of the GSTA for two reasons:

(a) First, Ms Chong was of the view that proportionally uplifting the sentencing range in the Tan Song Cheng Framework was incorrect as it did not take into account the higher sentences in s 96A of the ITA.

(b) Second, she was of the view that the three “levels of harm” approach of using the amount of tax evaded was (a) not consistent with the legislative intent behind the sentencing approach to GST evasion under s 62 of the GSTA, and (b) unsupported by the data submitted by the Prosecution.

19 Accordingly, Ms Chong proposed a framework based on the approach taken in Logachev by Menon CJ. Her proposed framework deviated from the Original Framework in that instead of the three “levels of harm” approach, it adopted the traditional “slight-moderate-severe” categorization for the harm axis. I reproduce her proposed framework below:

20 In Ms Chong’s opinion, the above framework would take into account and give appropriate weight to all the harm and culpability factors relevant to s 62 of the GSTA. Notably, despite disagreeing with the approach of a proportional uplift, in formulating her proposed framework, Ms Chong arrived at the same sentencing ranges as the Original Framework, albeit for different reasons. She explained that she derived the sentencing range in her framework by following the practice of pegging the highest category for harm/culpability at about two-thirds of the maximum prescribed punishment under s 62 of the GSTA: see Kow Keng Siong, Sentencing Principles in Singapore (Academy Publishing, 2019) at [13.052]–[13.053].

21 On 5 November 2021, the Prosecution filed reply submissions in response to Ms Chong’s recommendations. Notably, the Prosecution abandoned the Original Framework and proposed a revised framework which eschewed the three “levels of harm” approach, and was broadly in line with Ms Chong’s proposed framework, adopting the same sentencing ranges (as set out above at [19]). Accordingly, I will refer to both Ms Chong and the Prosecution’s revised framework as “the Revised Framework”. But while Ms Chong and the Prosecution concurred on the broad contours of the Revised Framework, they differed on several specific matters pertaining to the application of the factors within the Revised Framework. I deal with these below at [54]–[72].

22 Finally, on the appropriate sentence, despite resiling from the Original Framework, the Prosecution maintained its position that 24 weeks’ imprisonment was appropriate, applying the Revised Framework. The Defence did not file reply submissions, but addressed the Revised Framework in oral submissions.

23 After considering the written and oral submissions of the parties’ and Ms Chong’s brief, I was of the view that a sentencing framework for s 62 of the GSTA ought to be formulated. It was clear from the precedents submitted by the Prosecution that there has been a lack of consistency in sentencing offenders for offences under s 62 of the GSTA. To let such inconsistency fester would be unsatisfactory, and accordingly, I found it necessary to implement an approach to address this (a view that Ms Chong and the Defence shared).

24 With this threshold question answered, there were several issues to be resolved. The first was the broad framework that was to be adopted. The second was the specific factors to be considered in applying the broad framework. The third was the application of the framework in the present case.

25 In brief, my decision was as follows:

(a) First, that the broad framework to be adopted was the Revised Framework as the Original Framework was not justified in law or fact.

(b) Second, with regard to several specific issues pertaining to the application of the Revised Framework: (i) whilst the quantum of tax evaded is an important sentencing factor it cannot be considered the primary factor in every case; (ii) there was no difference between harm from tax evasion (in terms of underdeclaration of tax) and from refunds paid out to the accused; (iii) a breach of “professional responsibilities” was a relevant “offence-specific” culpability factor; (iv) the total amount of GST evaded is a matter for the “totality” analysis and (v) the imposition of a fine would be appropriate where the amount of tax evaded was small enough such that the effect of a fine would not be eclipsed by the mandatory penalty found in s 62(1)(f) of the GSTA.

(c) Third, the accused’s culpability was low and the harm caused was slight. Applying the Revised Framework, I sentenced her to a global sentence of 24 weeks’ imprisonment. This consisted of a sentence of 9 weeks’ imprisonment each for the 5th and 7th charges, and 15 weeks’ imprisonment for the 8th charge, with the 5th and 8th charges running consecutively.

26 I agreed with Ms Chong that the Original Framework should not be adopted. This was chiefly because of the method by which the Prosecution had derived the sentencing ranges in the Original Framework, ie, by proportionally uplifting the sentencing ranges in the Tan Song Cheng Framework, which I was of the view was inappropriate. Furthermore, I observed that generally, the “levels of harm” approach was not an appropriate basis for a sentencing framework for an offence under s 62 of the GSTA.

27 The defining feature of the Original Framework was that its sentencing ranges represented a proportional uplift of those found in the Tan Song Cheng Framework. As noted above at [13] the only difference between the two was that the Original Framework’s sentencing ranges had been proportionally increased by a factor of two-and-one-third to account for the higher maximum sentence provided for by s 62 of the GSTA.

28 However, this defining feature was also the significant flaw in the Original Framework. As I pointed out to the Prosecution, the proportional increase could only be justified if Parliament intended for GST evasion to be treated more seriously than income tax evasion, under s 62 of the GSTA and s 96 of the ITA respectively. An example may explain this. Imagine a case where an accused pleads guilty to evading $10,000 worth of GST; now imagine if that same accused had pled guilty to evading $10,000 worth of income tax using the exact same means. Applying the Original Framework (see [6] above) to the former, and the Tan Song Cheng Framework (see [12] above) to the latter, the offender who evaded GST would face a more severe imprisonment sentence (assuming of course, all other factors being equal).

29 This would only be justifiable if Parliament intended to punish offenders who evade GST more severely than those who evade income tax. The Prosecution’s original position was that “it is evident from the prescribed sentences in the relevant tax evasion provisions of the [GSTA] and the [ITA] … that Parliament appears to consider GST evasion to be more egregious than income tax evasion”. The Prosecution argued that it is “axiomatic that the higher maximum prescribed penalty … must signal the gravity with which Parliament views such crimes”. In support, the Prosecution had cited the case of Angliss Singapore Pte Ltd v Public Prosecutor [2006] 4 SLR(R) 653 at [84].

30 However, the passage cited stands only for the proposition that a maximum sentence signals the gravity with which Parliament views an offence – it does not deal with the situation that arose in the appeal where there were two different legislative provisions with different maximum sentences pertaining to offences for evasion of different types of tax. Accordingly, I did not think that the Prosecution could rely on the difference in the maximum sentences as evincing an intention by Parliament to treat GST evasion more seriously – there had to be some statement by Parliament signalling such an intention. As I noted above, during the first hearing of the appeal, the Prosecution could not point me to such a statement. This remained the same during the second hearing. There was good reason why this was the case, as pointed out by Ms Chong.

31 It was apparent that the higher maximum sentence in s 62 of the GSTA had nothing to do with Parliament intending to treat GST evasion more seriously than income tax evasion. Instead, the higher range in s 62 of the GSTA, as opposed to s 96 of the ITA, was explicable with reference to the sentencing range in s 96A of the ITA, which, as pointed out by Ms Chong, the Prosecution had not taken into account in formulating the Original Framework.

32 For context, I reproduce both ss 96 and 96A of the ITA:

33 As is apparent from the emphasised portions above, the fundamental difference between ss 96 and 96A of the ITA is the means or modus operandi by which income tax is evaded. The latter is targeted at serious fraudulent conduct in evading income tax or obtaining a PIC bonus, defining such conduct as either: (a) the falsification of books of account or records/the preparation or maintenance of false books of account or records (as well as the authorisation of such actions); and (b) the use of any fraud, art or contrivance (as well as the authorisation of such actions). That this is the operative difference between the two provisions was made clear during the introduction of s 96A by the then second Minister for Finance, Mr Lim Hng Kiang (see Singapore Parliamentary Debates, Official Report (11 November 2003) vol 76 at col 3510 (Mr Lim Hng Kiang, second Minister for Finance)):

34 As a result, the maximum sentence for offences under s 96A is higher than for offences under s 96 of the ITA. In short, the ITA differentiates between tax evasion and serious fraudulent tax evasion by having separate provisions for each type with different sentencing ranges, with the latter having a higher maximum sentence to reflect the more egregious manner of evasion.

35 Such a division is not present in s 62 of the GSTA. This was significant in my view. I reproduce the section below:

36 From the emphasised portion, it is clear that, unlike the ITA, the GSTA does not segregate tax evasion and serious fraudulent tax evasion into separate provisions with their own individual sentencing ranges. Instead, s 62(1) of the GSTA, encapsulates, in one section, provisions similar to those found in both ss 96(1) and 96A(1) of the ITA. Section 62(1)(a) to (c) of the GSTA mirrors s 96(1)(a) to (c) of the ITA while s 62(1)(d) and (e) of the GSTA mirrors s 96A(1)(a) and (b) of the ITA. Thus, the equivalent of “serious fraudulent tax evasion” in s 96A of the ITA is provided for in s 62(1) of the GSTA. This to me explained why the sentencing range in s 62 of the GSTA is significantly higher than that in s 96 of the ITA, ie, the presence of more serious methods of tax evasion necessitated a higher sentencing range. It was not due to Parliament regarding GST evasion per se to be more serious than income tax evasion.

37 This being the case, it was incorrect to derive the sentencing range for offences under s 62 of the GSTA by proportionally uplifting the sentencing ranges in the Tan Song Cheng Framework (which was based on and meant solely for s 96 of the ITA). Thus, I did not accept the Original Framework.

38 Aside from the above, there were also several issues with the “levels of harm” approach utilised in the Original Framework, as pointed out by Ms Chong. In its reply and oral submissions, the Prosecution fairly accepted Ms Chong’s criticism. I explain these criticisms and why I agreed with them.

39 To begin with, it was apparent that in determining the three “levels of harm” in the Original Framework, the data and approach used by the Prosecution were incorrect. This was pointed out by Ms Chong in support of her position that the Original Framework was wrong “in fact”. I understood this to mean that the data did not support the “levels of harm” proposed in the Original Framework. In this regard, two of Ms Chong’s points stood out for me.

40 First, based on calculations she provided, Ms Chong pointed out that the typical GST return filed per quarter was approximately S$25,000. Yet the Prosecution’s “Level 1” monetary threshold was S$75,000 – three times the average provided by Ms Chong. Accordingly, viewed from this perspective it was not clear to me whether evading S$75,000 worth of GST could be said to be a “low level type of harm”.

41 Second, and more significantly, in using past cases as an empirical basis to set the thresholds for the three “levels of harm” in the Original Framework, it appeared that the Prosecution conflated (a) the total amount of GST evaded for all the proceeded charges with (b) the GST evaded for an individual charge. In other words, instead of using the GST avoided per proceeded charge as the reference data, the Prosecution used the GST avoided for all proceeded charges as the relevant data point, in each case. This was clear from an examination of the Prosecution’s first set of submissions, in particular, Annex A of those submissions (“Annex A”).

42 Annex A consisted of a table of 53 unreported cases of offences under s 62 of the GSTA. As was pointed out by Ms Chong in her brief, for 37 of the 53 cases, or 69.8%, the total amount of GST evaded was below S$75,000 (the “total amount” meaning the aggregate amount for all the proceeded charges). In six of the 53 cases, or 11.3%, the total amount of GST evaded fell between S$75,000 and S$150,000. Finally, in ten of the 53 cases, or 18.9%, the total amount of GST evaded was above $150,000.

43 These figures were used by the Prosecution to arrive at and justify the thresholds for the three “levels of harm” in the Original Framework. I reproduce the relevant portion of the Prosecution’s first set of submissions below:

44 The percentages calculated by Ms Chong (see [42] above) correlated with those cited by the Prosecution in their submissions (see [43] above). It was clear from the Prosecution’s submissions that the percentages they had used were based on GST evaded for all proceeded charges and not the GST evaded for the individual proceeded charges in each case. In other words, it set the “levels of harm” in the Original Framework based on the total amount of GST evaded for all proceeded charges. This obviously was incorrect as the Original Framework was meant to be applied to individual charges – this much is clear from the fact that the fifth step of the Original Framework provided for the application of the totality principle where an offender has been convicted of multiple charges. Thus, the Prosecution had used erroneous figures to justify the different thresholds for the “levels of harm” in the Original Framework. This was contrasted to Tan Song Cheng where the Prosecution had based its “levels of harm” on the amount of income tax evaded for the individual charges.

45 The Prosecution in their further submissions fairly recognised and accepted Ms Chong’s criticism. They also produced a revised Annex A this time with a fresh analysis of 56 precedents. The revised Annex A focused on the GST evaded for the individual charges. This analysis yielded different results, with 96.7% of the individual charges involving amounts less than $75,000, 0.89% involving sums between $75,000 and $150,000 and 2.45% involving sums above $150,000. It was clear from these new figures that the thresholds proposed in the Original Framework were not tenable.

46 A further reason for rejecting the Original Framework was that I found there to be conceptual difficulties with the “levels of harm” approach in general. This was a view shared by Ms Chong.

47 As a preliminary point, GST rates are subject to change, which would lead to higher amounts of tax evaded and potentially skewed sentencing outcomes as the “levels of harm” would be derived from cases where the tax evaded was computed on previous tax rates. It is common knowledge that there are plans for the GST rate to be raised in the near future – a fact recognised by the Prosecution. As a consequence of these higher rates, the amount of GST evaded would be larger. This would lead to different results in the future if monetary thresholds such as those in the Original Framework were used. There would also be a need to update the thresholds whenever the GST rate was changed.

48 The more fundamental problem with the “levels of harm” approach is that there is no upper limit to the amount of GST that could be evaded. Absent such a limit, there is no reference point to determine what amount is “slight”, “moderate” or “severe”. It is conceptually flawed to then divide the harm into three different levels based on arbitrary monetary thresholds. This view was also expressed by Menon CJ in Logachev in the context of cheating offences under the Casino Control Act (Cap 33A, 2007 Rev Ed) (“the Casino Control Act”):

49 Menon CJ’s observations cited above were clearly not premised on any consideration unique or specific to cheating under the Casino Control Act. Instead, they were premised on a logic of general application. As such, they would apply with equal force to offences under s 62(1) of the GSTA. It should be noted that in Logachev, the Prosecution had proposed a similar sentencing framework premised on the amount of money cheated: see [26] of Logachev. Menon CJ ultimately rejected it for various reasons, including the reason stated above: Logachev at [43]. There were other reasons cited by Menon CJ that were relevant to the present case; however these will be dealt with in the context of the related issue of why the quantum of GST evaded should not be considered the default primary determinant of harm (see [55]–[64] below).

50 Having not accepted the Original Framework, I considered the Revised Framework proposed by Ms Chong and the Prosecution. I preferred the broad contours of the Revised Framework which was as follows:

(a) The first step requires the court to identify the harm and culpability factors. With regard to harm, the amount of GST evaded, and the state resources spent on investigation are examples of relevant factors. With regard to culpability, examples of relevant considerations include the degree of planning and premeditation, the sophistication of the methods of evasion, and the offender’s role in the offence.

(b) The second step requires the court to determine the indicative starting range for sentencing based on a three-by-three harm-culpability matrix. The harm is categorised into slight, moderate and severe, whilst the culpability is categorised as low, medium and high.

(c) The third step requires the court to determine the indicative starting sentence within the range. This requires a more granular analysis of the facts of the case to come to a specific sentence.

(d) The fourth step requires the court to consider offender-specific factors to calibrate the indicative starting sentence up or down. Examples of aggravating factors would include any TIC charges, relevant antecedents, or an evident lack of remorse. On the other hand, mitigating factors would include a plea of guilt and voluntary restitution.

(e) The fifth and final step requires the court to determine how the sentences should run, with reference to the usual principles such as the one transaction rule and the totality principle.

51 In my opinion, the Revised Framework did not have the same flaws of arbitrariness and rigidity that the Original Framework suffered from. Whilst, arguably, the Revised Framework would be more difficult to apply consistently, it must be remembered that consistency in sentencing is not an inflexible or overriding principle – the different degrees of culpability and the unique circumstances of each case play an equally if not more important role: Rahman Pachan Pillai Prasana v Public Prosecutor [2003] SGHC 52 at [23].

52 Furthermore, it must be remembered that sentencing is not a mathematical exercise, a view expressed by Yong Pung How CJ in Soong Hee Sin v Public Prosecutor [2001] 1 SLR(R) 475. In that case, the accused faced a charge of criminal breach of trust, having taken a sum of $10,485.22. Counsel had led Yong CJ through a line of sentencing precedents, seeking to draw “some sort of mathematical formula from which the proper sentence in each case could be calculated with scientific accuracy”: at [11]. In response to this exercise, Yong CJ stated emphatically:

53 In my opinion, the Revised Framework was more in line with the fundamental principles and nature of sentencing in criminal law. Accordingly, I accepted it to be the appropriate sentencing framework.

54 However, while I accepted the broad contours of the Revised Framework, several details in its application were contested between the Prosecution, the Defence, and Ms Chong. I now turn to my decision on these specific issues.

55 The most significant point of tension was whether the amount of tax evaded was the “primary” factor in determining the harm caused by an offender. The Prosecution’s position was that the quantum of GST evaded should be treated as the “primary, but not the sole, determinant of harm” [emphasis in original]. Ms Chong on the other hand submitted that it was not appropriate “to place too much emphasis on the quantum of GST evaded”.

56 I agreed with Ms Chong and held that the amount of tax evaded ought not to be taken as the primary consideration (although I recognised that it was an important factor that went towards the broader question of harm). This was for many of the same reasons I had considered in rejecting the three “levels of harm” approach in the Original Framework (see [46]–[49] above). In addition, there was a further reason why the quantum of GST evaded ought not to be the primary harm factor: to do so would run the risk of ignoring other sentencing factors completely, ie, to make the quantum of GST evaded the primary factor would inevitably lead to it becoming the sole determinant.

57 A similar view was expressed by Menon CJ in Logachev. He noted at [46] of that judgment that the particular mischief targeted by s 172A(2) of the Casino Control Act was criminal activity in general in casinos. Accordingly, “singling out the amount cheated [had] the potential to divert attention away from the other relevant sentencing considerations that go towards the harm caused” and “also has the potential to divert attention away from the relevant sentencing considerations that go towards the offender’s culpability”: at [44] and [47]. Ultimately, Menon CJ declined to single out the amount cheated as the primary harm factor: at [51].

58 These observations were particularly apposite in the present case. Whilst I recognised that using an easily quantifiable factor such as the amount of GST evaded was attractive in its simplicity, it risked creating tunnel vision that might result in culpability factors and other harm factors being overshadowed. This would be particularly problematic given the mischief that s 62 of the GSTA seeks to address.

59 As noted by Ms Chong, the particular mischief that s 62 of the GSTA targets is the wilful, fraudulent GST evasion through any of the prescribed means. Whilst this is not apparent from a plain reading of the provision, it became abundantly clear that this was the case from a contextual and purposive interpretation: Tan Cheng Bock v Attorney-General [2017] 2 SLR 850 at [37].

60 With regard to the contextual approach, it was noteworthy that the provisions surrounding s 62 of the GSTA are also geared towards deterring the evasion of tax simpliciter: ss 58–61 of the GSTA also create offences for the evasion of tax.

(a) First, s 58 is a gap filling provision that provides for a general penalty where penalties are not provided for in offence creating provisions in the GSTA.

(b) Second, s 59 creates an offence for making an incorrect return, and importantly, provides for more onerous penalties where an incorrect return is made “without reasonable excuse or through negligence”.

(c) Third, s 60 provides for penalties where an offender fails to pay or make returns within a prescribed period.

(d) Fourth, s 61 makes failure to register in accordance with the First Schedule of the GSTA an offence.

61 Leaving aside s 58, it was clear to me that these provisions are targeted at inadvertent or negligent conduct. The culpability is clearly lower than s 62 which expressly requires “wilful intent”. Thus, the distinguishing feature in s 62 is that the offender must have intended to evade tax by any of the constituent methods found in s 62(1). This suggests that the key mischief that was intended to be addressed by the enactment of s 62 is the wilful evasion of GST. This point finds support from the title to the section: “Penalty provisions relating to fraud, etc.” [emphasis added].

62 This interpretation is also supported by the Parliamentary debates when amendments to the GSTA were introduced. During these debates, it was stated that the penalty provisions had been “amended to make a distinction between innocent errors and fraudulent conduct” [emphasis added]: Singapore Parliamentary Debates, Official Report (12 October 1993) vol 61 at col 584 (Dr Richard Hu Tsu Tau, Minister for Finance).

63 Thus, accepting that the distinct mischief targeted by s 62 is wilful GST evasion, the method or mode of evasion, and not just the amount of GST evaded, must be a key consideration in the sentencing exercise. However, if the amount of GST evaded was made the primary factor, it might eclipse or dilute the other considerations that were relevant to not only the harm analysis, but also the culpability analysis (see [56]–[58] above).

64 However, I noted that whilst the amount of GST evaded was not necessarily the primary factor to consider in determining harm, there might be cases where it did come to the forefront in the harm analysis. For example, where other harm factors were absent, or where the accused’s method of evasion was simple, the only factor that would stand out would be the amount of GST evaded. In such a case, by reason of the absence or relative insignificance of the other factors related to harm, the amount of GST evaded would become the de facto primary factor. This was particularly relevant in the present case.

65 The next point in contention arose from Ms Chong’s submission that an aggravating factor would be where the accused, through fraudulent means, obtained payments, credits or refunds. She argued that this would result in a “higher net loss”. I understood this submission to essentially mean that, if one offender had simply evaded $X, and another had received $X in refunds from the state, the latter’s conduct would be regarded as more serious. The Prosecution’s position was that this should not by itself be an aggravating factor.

66 I agreed with the Prosecution. Ms Chong’s argument was that refunds ought to be treated more seriously than tax evaded, rationalizing that tax evaded was “input”, whilst refunds paid out to the accused was “output”. There was thus a conceptual difference between the two. However, in my opinion, they both involved loss to the state, and it did not matter whether the loss was occasioned by the evasion of tax, or by the payment of a refund. As the Prosecution submitted, in either case, “there is a shortfall in the pool of public funds caused by the tax evasion offence”. In oral submissions, Ms Chong accepted this point. Thus, the fact that an offender had obtained a refund or payment from the state was not an additional aggravating factor.

67 Next, Ms Chong submitted that an additional offender-specific factor to consider ought to be whether an accused has wilfully evaded GST in breach of “professional responsibilities”. The Prosecution agreed that this was a relevant factor, but took the position that it was an offence-specific culpability factor.

68 I agreed with Ms Chong and the Prosecution that a breach of a professional responsibility was aggravating for the same reason as an abuse of one’s position, or a breach of trust – the accused has been put in a situation where he or she was expected to act with integrity, but did not. However, I agreed with the Prosecution that it ought to be an offence-specific factor. There was an additional point. It was not readily apparent to me why Ms Chong had drawn a distinction between a breach of professional responsibilities and a breach of trust (which was already an offence-specific factor she had listed in the Revised Framework).

69 In oral submissions, Ms Chong explained that a breach of trust related to situations such as where the accused was a director of a company, whilst an example of a breach of professional responsibilities would be where the accused has a professional accreditation such as being a chartered accountant. In my opinion, this distinction can be adequately subsumed under the general offence-specific factor of a breach of duty. As noted above, the key point with regard to this factor is that the accused has been placed in a situation where he is expected to act with higher standards of integrity, but had failed to do so. This would be the case regardless of whether the duty was professional or otherwise. There was no necessity for a specific carve-out for “professional responsibilities”.

70 I also consider two further points raised by Ms Chong that I agreed with. The Prosecution did not contest these points.

71 First, I agreed with Ms Chong’s submission that the total amount of GST evaded should be analysed as part of the totality principle under the fifth and final step. The question that this submission raised was: would considering the total amount of GST evaded at the final stage of the framework overlap with the consideration of the TIC charges in the fourth stage? The main concern was that of double counting. In my opinion, this concern was unwarranted. TIC charges are aggravating factors, whilst the total amount of tax evaded would serve as a marker for the sentencing judge to “check” that the final sentence is in line with the offender’s overall criminality. These are two separate inquiries and I thus agreed with Ms Chong’s position.

72 Second, as to when the imposition of fines for offences under s 62 of the GSTA was appropriate, Ms Chong aligned herself with the position taken by See Kee Oon J in [73] of Tan Song Cheng. There, he held that a fine might be imposed for offences where the deterrent effect of the fine would not be eclipsed by the imposition of the mandatory penalty. As such, generally, where the amount of tax evaded would result in a mandatory penalty that outstripped the maximum fine, a custodial sentence would be more appropriate. I agreed with the logic in this position and held that it applied equally to GST evasion.

73 With these details clarified, I applied the Revised Framework and sentenced the accused to 24 weeks’ imprisonment.

74 However, before applying the Revised Framework, I dealt with the Defence’s submission that the doctrine of prospective overruling ought to be invoked, and thus the Revised Framework ought not to be applied.

75 The Defence’s argument was premised on two alleged sentencing benchmarks: Chng Gim Huat v Public Prosecutor [2000] 2 SLR(R) 360 (“Chng Gim Huat”), and Loon Wai Yang v Public Prosecutor [2020] SGHC 34 (“Loon Wai Yang”). The Defence argued that the introduction of a new sentencing framework represented a significant change that would prejudice the accused. Thus, the doctrine of prospective overruling ought to be invoked.

76 In determining whether the doctrine of prospective overruling was to apply, the court would consider several factors as set out in the case of Public Prosecutor v Hue An Li [2014] 4 SLR 661 (“Hue An Li”) at [124]. These were: (a) the extent to which the law or relevant principle concerned was entrenched; (b) the extent of the change to the law; (c) the extent to which the change to the law was foreseeable; and (d) the extent of the reliance on the law or legal principle concerned. In my opinion, in the present case, the factors clearly weighed against the application of the doctrine.

77 First, neither of the two cases were entrenched, as they were not “sentencing benchmarks”. A “sentencing benchmark” is a case which identifies an “archetypal case … and the sentence which should be imposed in respect of such a case”. Such an archetype must be identified with some specificity, both in terms of the facts of the case, and the sentencing considerations that informed the sentence meted out: Ng Kean Meng Terence v Public Prosecutor [2017] 2 SLR 449 (“Terence Ng”) at [31]. For reference, an example of such a case is Wong Hoi Len v Public Prosecutor [2009] 1 SLR(R) 115 which set out a benchmark sentence of four weeks’ imprisonment for cases involving the voluntary causing of hurt to a public transport worker.

78 With regard to the two cases cited by the accused, neither met the characteristics of a “sentencing benchmark” for an offence under s 62 of the GSTA. First, the case of Chng Gim Huat involved offences under the ITA, not the GSTA. Second, the case of Loon Wai Yang, whilst concerning an offence under s 62 of the GSTA, did not bear any of the characteristics noted above – there was no identification of specific facts or sentencing considerations that future courts could use as touchstones: Terence Ng at [32].

79 Next, the introduction of the Revised Framework did not represent a significant change. A sentencing framework merely seeks to clarify and promote consistency in the existing state of the law, ie, it does not bring about a distinct change in the law. Related to this, I did not think that such a change was unforeseeable. This was especially so given the recent move by our courts towards developing and implementing more consistent sentencing approaches. Ultimately, accused persons generally could not argue that they had a “legitimate expectation” as to the sentencing framework that would be applicable to them. Little weight ought to be given to the expectations of a person who flouts the law and later finds out that the expected costs or consequences are worse than anticipated. The only legitimate expectation that accused persons could rely on here was that they would be sentenced within the statutorily prescribed range: see Adri Anton Kalangie v Public Prosecutor [2018] 2 SLR 557 at [57].

80 Accordingly, the doctrine of prospective overruling did not apply.

81 Moving on to the application of the Revised Framework, with regard to the first step, I first set out the relevant offence-specific factors, having canvassed some above at [50(a)], and dealt with others from [54]–[72]:

82 With regard to the “harm” factors, the accused was not part of a syndicate and there was no transnational element present. Further, there was no suggestion from the Prosecution that significant resources had been spent in investigating her offences. Thus, the only harm factor relevant to each charge was the amount of GST evaded. With regard to the “culpability” factors, the accused’s methods of evasion were not sophisticated, and there were no other relevant culpability factors.

83 Turning to the second step, I was of the opinion that the accused’s culpability was low, and the harm caused by her action was slight. Applying this to the sentencing table (reproduced below), this gave an indicative sentencing range of a fine, up to 14 months’ imprisonment:

84 The third step presented the most difficulty. To begin, the amounts evaded in all three of the charges (around S$30,000 to S$50,000) were not insignificant, and were enough to push the individual sentences over the custodial threshold (a position that the parties were in agreement on). The difficulty was in determining the length of the appropriate custodial sentence.

85 To aid in the exercise, I divided the indicative sentencing range of up to 14 months’ imprisonment into thirds. This resulted in the following “sub-ranges”:

86 The Prosecution proposed that for the 5th and 7th charges (where the amount of GST evaded was just shy of $30,000) the indicative starting sentence ought to be in the highest end of Sub-range 1, ie, around four and two-thirds months’ imprisonment. Converted into weeks, this would be about 18 weeks’ imprisonment for each charge. For the 8th charge (which involved just under S$50,000 of GST evaded) the Prosecution’s position was that this ought to warrant a sentence in the very middle of the entire 14-month range, ie, seven months’ imprisonment. Converted into weeks, this would be about 30 weeks’ imprisonment.

87 The Defence’s position on the other hand, was that the sentence ought to be the same for all three proceeded charges, and that this would be about two and one-third months’ imprisonment. This position was forwarded on the basis that only one of the harm factors was present, and in terms of culpability, the court ought not consider the long period of offending as this should only be considered at the fourth stage when the TIC charges were taken into account.

88 In my opinion, the culpability of the accused was at the lower end of the low category as her methods of evasion were not sophisticated. Further, I agreed with the Defence’s submission and did not take into account the long period of offending at this point as this would amount to double counting the TIC Charges (which I considered later at the fourth stage of the sentencing exercise).

89 However, I did not agree with the Defence’s submission that there should be no difference between the 5th and 7th charges on the one hand, and the 8th charge on the other. While the quantum of GST evaded is not the primary harm factor, it is nonetheless important. The 5th and 7th charges involved amounts of around S$30,000, whilst the 8th charge involved almost S$50,000. In my opinion, the difference of around S$20,000 was not insignificant, and thus the 8th charge warranted a higher sentence.

90 Furthermore, whilst it was correct that only one harm factor was present, this did not lead to a reduction in sentence as the Defence seemed to suggest. One must consider the extent of the harm caused as a whole, and in this case, the only marker which I could reference was the amount of GST evaded.

91 In my opinion the sentence for the 5th and 7th charges fell in the high end of Sub-range 1, whilst the 8th charge fell somewhere close to the middle of the entire range. Accordingly, I held that the indicative starting sentences for each charge was as follows: for the 5th and 7th charges, 14 weeks’ imprisonment each; and for the 8th charge, 24 weeks’ imprisonment.

92 Moving on to the fourth step, I considered the offender-specific factors for offences under s 62 of the GST Act. These are summarised below:

93 First, there were eight TIC charges which represented a further S$118,991.45 of GST evaded. For two of the TIC Charges, the 6th and 9th, the amount of GST evaded was low, and would likely not attract a custodial sentence if they had been proceeded with. As such, I did not factor these into any enhancement of the sentences for the proceeded charges. For the rest of the TIC charges, given the more substantial amount of tax evaded, I was of the opinion that the custodial threshold would have been crossed had they been proceeded on. Accordingly, I attached the following uplifts for each of them:

94 However, I gave credit to the full restitution made by the accused and only enhanced the proceeded charges on the basis of the 1st, 2nd and 4th charges, thereby increasing the sentences for all three proceeded charges by two weeks each. Thus, the individual sentences for the proceeded charges were 16 weeks’ imprisonment each for the 5th and 7th charges, and 26 weeks’ imprisonment for the 8th charge.

95 Finally, I also considered the fact that the accused pleaded guilty at an early stage and her overall remorse for her actions. Taking this into account, I reduced the sentences by about 40%, giving the following individual sentences: 9 weeks’ imprisonment each for the 5th and 7th charges, and 15 weeks’ imprisonment for the 8th charge.

96 Finally, with regard to the fifth and final stage, since the accused faced three proceeded charges, two of the sentences had to run consecutively as per s 307 of the Criminal Procedure Code (Cap 68, 2012 Rev Ed). I ran the 5th and 8th charges consecutively for a total of 24 weeks’ imprisonment. Considering the accused’s overall criminality in the present case, I did not view this as being crushing or out of proportion. Accordingly I did not further vary her sentence.

97 I allowed the Prosecution’s appeal and sentenced the accused to a total of 24 weeks’ imprisonment. I dismissed the accused’s appeal.

98 I would also like to take the opportunity to commend Ms Chong on her detailed and cogent work – the assistance she rendered to this court and the parties was immensely valuable. Her hard work and professionalism are a credit to the legal profession.