Tan Song Cheng v Public Prosecutor and another appeal [2021] SGHC 138

The appellants had pleaded guilty to charges under s 96(1)(b) of the Income Tax Act (Cap 134, 2008 Rev Ed) (“ITA”) alongside other charges under the Goods and Services Tax Act (Cap 117A, 2005 Rev Ed) (”GSTA”). They were convicted and sentenced by the same District Judge (“DJ”), sitting in his ex officio capacity as a Magistrate, in separate and unrelated proceedings. The Prosecution proposed a sentencing framework for s 96(1) ITA offences, which was adopted by the DJ. Dissatisfied with their sentences, both appellants filed their respective appeals. As the issues concerning both appeals were substantially the same, I heard both appeals together.

Having heard the parties’ submissions and considered the materials before me, it is clear that there has been no consistent or coherent sentencing trend in the sentencing decisions in respect of offences under s 96(1) of the ITA. Moreover, there has not been any sentencing guidance from this court since Chng Gim Huat v Public Prosecutor [2000] 2 SLR(R) 360 (“Chng Gim Huat”).

In this judgment, I address the need to develop a sentencing framework to guide sentencing for offences under s 96(1) of the ITA. I proceed to explain my reasons for substantially endorsing the Prosecution’s proposed framework and demonstrate the application of the framework to these appeals, to reach my conclusion that the sentences imposed were not manifestly excessive and that the appeals should therefore be dismissed.

The material facts

MA 9758/2020

In this appeal, the appellant, Tan Song Cheng (“the First Appellant”) was the director of TNT Cards & Silkscreen Pte Ltd (“TNTPL”) and the precedent partner of TNT Art & Silkscreen (“TAS”), which he ran together with one Lim Geok Mee (“Lim”). Between 2008 and 2014, whenever TNTPL’s sales revenue exceeded $1 million, Lim, with the approval of the First Appellant, would falsely reduce TNTPL’s report sales revenue, by transferring the revenue to three other business entities, one of which was TAS. For failing to report TNTPL’s liability to register for Goods and Services Tax (“GST”), TNTPL had committed an offence under s 61(a) of the GSTA. At the time of the commission of the aforesaid offence, the First Appellant was a director of TNTPL and hence, the First Appellant had committed an offence under s 61(a) read with s 74 of the GSTA. The First Appellant was fined $2,000 for the offence under the GSTA and is not appealing that sentence.

In addition to the above, in 2009 and 2011, the First Appellant agreed to Lim’s false reduction of TAS’s net profit, so that it would fall beneath an artificially-imposed cap of $100,000. This led to the under-reporting of the First Appellant’s share of TAS’s trade income. In addition, the First Appellant failed to declare holiday reimbursements received from TNTPL as a “performance reward”. This led to the under-reporting of the First Appellant’s employment income from TNTPL. The resulting amount of tax undercharged was $34,992.26 in 2009 and $34,444.18 in 2011. The mandatory penalties the First Appellant was liable for under s 96(1)(i) of the ITA were $104,976.78 and $103, 332.54 respectively.

The First Appellant pleaded guilty to the two proceeded charges under s 96(1)(b) of the ITA as well as the s 61(a) GSTA charge and was sentenced on 20 July 2020. Six similar s 96(1) ITA charges were taken into consideration for the purpose of sentencing. Through his nine offences, the First Appellant had evaded taxes totalling $221,938.01. At the proceedings below, the DJ observed that there was no consistency in the sentences imposed for s 96(1) ITA offences save that the sentences were always under two months’ imprisonment. He also observed that there was little or no correlation between the amount of tax evaded and the sentences imposed for cases decided post-Chng Gim Huat, and that the current sentencing norm did not utilise the full sentencing spectrum provided for by Parliament, which in turn undermined the seriousness of s 96(1) ITA offences. The DJ held that Chng Gim Huat did not establish a sentencing benchmark of a short custodial sentence of up to two months for s 96(1) ITA offences, and proceeded to adopt the Prosecution’s proposed five-step sentencing framework based on a harm-culpability analysis.

The DJ found that the harm occasioned, and the culpability of the First Appellant came within the low range. Taking into account the First Appellant’s plea of guilt and the six other charges taken into consideration, the DJ imposed a sentence of six weeks’ imprisonment in respect of each proceeded charge, and ordered the sentences for the two charges to run consecutively, resulting in a global sentence of 12 weeks’ imprisonment for the s 96(1) ITA offences. The DJ’s grounds of decision are reported at Public Prosecutor v Tan Song Cheng [2020] SGMC 50.

MA 9768/2020

In this appeal, the appellant Lin Shaohua (“the Second Appellant”) was the precedent partner of two partnerships, Furniture Collection Centre (“FCC”) and Yang Hua Furniture Trading (“YHFT”). From 2009 to 2015, one Lim Sai Cheok (“Lucy”) prepared the accounts of both FCC and YHFT, and filed the income tax returns of FCC and YHFT, as well as the personal income tax returns of the Second Appellant. At the material times, whenever the actual sales figures of either FCC or YHFT exceeded the $1 million threshold for GST registration, the Second Appellant would instruct Lucy to reduce the reported sales to below the $1 million threshold. At the time of the commission of the aforesaid offences, the Second Appellant was a partner of FCC and YHFT and hence, the Second Appellant had committed an offence under s 61(a) read with s 74 of the GSTA. The Second Appellant was fined $4,000 for the offence under the GSTA and is not appealing that sentence.

In addition to the above, by instructing Lucy to reduce the reported sales figures for FCC and YHFT to below the $1 million threshold, the Second Appellant’s reported partnership income in 2016 was correspondingly reduced, leading to her personal income being under-reported. The resulting amount of tax undercharged was $79,142.13. The mandatory penalty the Second Appellant was liable for under s 96(1)(i) of the ITA was $237,426.39.

The Second Appellant pleaded guilty to one proceeded charge under s 96(1)(b) of the ITA as well as the s 61(a) GSTA charge and was sentenced on 7 August 2020. Two similar charges, one each under s 96(1) of the ITA and s 61(a) read with s 74(1) of the GSTA respectively, were taken into consideration for the purpose of sentencing. Through her four offences, the Second Appellant had evaded taxes totalling $536,379. At the proceedings below, the DJ made the same substantive observations of law as he had done in MA 9758/2020.

Once again, the DJ adopted the Prosecution’s proposed five-step sentencing framework based on a harm-culpability analysis, and found that the harm occasioned and culpability of the Second Appellant came within the low range. Taking into account the Second Appellant’s plea of guilt and the two charges taken into consideration for the purposes of sentencing, the DJ imposed a sentence of ten weeks’ imprisonment for the s 96(1) ITA offences. The DJ’s grounds of decision are reported at Public Prosecutor v Lin Shaohua [2020] SGMC 53.

The parties’ submissions

The First Appellant’s Case

On appeal, the First Appellant contends that while the aim of s 96(1) of the ITA is to deter tax evasion, the mandatory penalty of treble the quantum of tax evaded already constitutes sufficient deterrence, and that the amount involved in financial crimes should not be the sole or overriding metric for the assessment of harm.

In respect of the Prosecution’s proposed sentencing framework, the First Appellant submits that as there is no upper limit to the amount of tax evaded, the proposed sentencing bands which are based solely on the amount of tax evaded would run the risk of arbitrariness. In addition, it was submitted that the Prosecution’s proposed sentencing framework had omitted the sentence of a fine.

On the facts, the First Appellant contends that he was not an “active participant” of the tax evasion scheme and consequently he is less culpable than Lim, who received a sentence of four weeks’ imprisonment and a mandatory penalty of $63,422.82. The First Appellant also contends that the DJ had given insufficient weight to his plea of guilt, that the consecutive sentences had offended the one-transaction rule, and that the sentence of 12 weeks’ imprisonment in addition to the mandatory penalty was crushing.

The Second Appellant’s case

On appeal, the Second Appellant contends that under the Prosecution’s proposed sentencing framework, the quantum of tax evaded has a disproportionate influence on the length of the custodial sentence. As the quantum of tax is also the sole determinant of the mandatory treble penalty, using it as the determinant of harm caused would lend disproportionate emphasis to it. In addition, the Second Appellant contends that the Prosecution’s framework had not adequately accounted for the possibility of a non-custodial sentence.

On the facts, the Second Appellant argues that the DJ had not placed sufficient weight on her personal circumstances and that the sentence of ten weeks’ imprisonment was inconsistent with the sentencing precedents. An appropriate sentence would be a fine or four weeks’ imprisonment according to the alternative sentencing framework proposed in her submissions.

The Prosecution’s case

In response to both appeals, the Prosecution submits that a sentencing framework is called for in respect of s 96(1) ITA offences. According to the Prosecution, the full sentencing range has not been utilised in the precedent cases, resulting in a lack of correlation in sentences to the severity of the offences, and the sentences clustering at the lower end of the sentencing range. To this end, the Prosecution proposed a modified five-step sentencing framework (“the proposed sentencing framework”) adapted from Logachev Vladislav v Public Prosecutor [2018] 4 SLR 609 (“Logachev”) for the offences under s 96(1) of the ITA. In the proposed sentencing framework, the Prosecution pegged the various levels of harm primarily on the quantum of tax evaded, alongside the additional factors of involvement of a syndicate or a transnational element. In this regard, the Prosecution submits that the proposed ranges of tax evaded for each level of harm are not arbitrary, and that the mandatory penalty should not have an impact on the determination of the sentence.

On the facts, the Prosecution takes the position that the DJ had been correct in his assessment of the Appellants’ culpability and the harm caused, as well as their respective offender-specific factors, in calibrating the sentences.

The amicus curiae’s submissions

Mr Kok Yee Keong (“the amicus curiae”) was appointed under the Supreme Court’s young amicus curiae scheme to assist the court with his opinion as to whether it was appropriate for the court to develop a sentencing framework for s 96(1) ITA offences. The amicus curiae was also requested to consider what an appropriate framework might be, and to provide his views on the Prosecution’s proposed sentencing framework. In addition, he was requested to consider whether the quantum of tax evaded should be the primary determinant of harm in calibrating the appropriate sentence and the impact of the mandatory penalty, if any, on the determination of the sentence.

The amicus curiae’s submissions are broadly in line with the Prosecution’s position. He agrees that a sentencing framework ought to be developed as there is a paucity of reasoned decisions on s 96(1) ITA offences, the sentences cluster around the minimum sentence, and the sentences appear inconsistent despite the fact that the cases are not that factually diverse.

In this regard, the amicus curiae also agrees with the Prosecution that the modified Logachev sentencing framework should be adopted, although he proposes modifying the framework in various ways, such as using the quantum of tax evaded as the sole determinant of the level of harm, and bringing the other harm considerations within the ambit of culpability.

I note that the amicus curiae’s submissions attempt to take the proposed sentencing framework a step further by proposing that the court should also use this opportunity to endorse the Logachev framework as a universal approach towards determining the sentence for any given offence, with suitable modifications. I appreciate that this suggestion is well-intentioned but with respect, I decline to adopt it. It may well merit reconsideration at a future point, but it is unnecessary for the purpose of dealing with the present appeals. Moreover, it is unclear whether a universal approach would be adequate to address the breadth and variance of criminal offences in Singapore.

The need to develop a sentencing framework

The sentencing precedents

Hitherto, Chng Gim Huat has been the sole High Court decision regarding the sentencing approach for s 96(1) of the ITA offences. In that case, the appellant failed to report income amounting to $1,314,000 and had claimed trial to two charges under s 96(1)(a) of the Income Tax Act (Cap 134, 1994 Ed). At first instance, he was sentenced to imprisonment of two and four months on the two charges, with both sentences running concurrently, in addition to the mandatory penalty of $1,069,936.90. Yong Pung How CJ dismissed the appeal against conviction but allowed the appeal against sentence, and reduced the period of imprisonment to one month and two months on the two charges with both sentences ordered to run concurrently. Yong CJ took into account the appellant’s swift action in notifying the Inland Revenue Authority of Singapore (“IRAS”), his contributions to the community, and his age. In doing so, Yong CJ also stated that the imposition of the treble penalty was mandatory, and that in view of the need for general deterrence, a custodial sentence would normally be warranted: see [105] and [107].

Since the decision in Chng Gim Huat, there has been a relative paucity of reasoned decisions regarding offences under s 96(1) of the ITA. As the amicus curiae has pointed out, apart from the DJ’s two judgments in the present appeals, there are apparently only two other judgments:

(a) Public Prosecutor v Onn Ping Lan [2005] SGMC 8, where the offender had pleaded guilty in the midst of trial to 20 counts of tax evasion. He was sentenced to a mandatory minimum aggregate sentence of six months’ imprisonment by virtue of s 96(2)(a) of the ITA. The mandatory penalty imposed was $546,471.01.

(b) Public Prosecutor v Chew Tiong Wei [2016] SGDC 59, where the offender had pleaded guilty to 28 charges which included three counts for tax-related offences. He was sentenced to an aggregate sentence of six months’ imprisonment for two charges under s 96(1)(b) and one charge under s 37J(3)(a) of the ITA, due to the requirement in s 37J(4)(a) of the ITA stipulating that a person who is convicted for three or more offences under the relevant provisions would be subject to a mandatory minimum imprisonment term of six months. The mandatory penalty imposed was $152,893.95. His appeal was dismissed but no reasons were given for the High Court’s decision.

Bearing in mind the need for caution in relying on sentencing precedents without fully-reasoned grounds or explanations as to how the sentencing judge reached a sentencing decision (see Janardana Jayasankarr v Public Prosecutor [2016] 4 SLR 1288 at [13(b)]), I turn next to consider the range of sentences meted out in tax evasion cases which are unreported. From the cases cited by the Prosecution and amicus curiae, it is clear that while imprisonment terms have been the norm for tax evasion cases, the sentences meted out have ranged between one- to six-weeks’ imprisonment (see also Darren Koh, Poh Eng Hin & Tang Siau Yan, The Law and Practice of Singapore Income Tax (LexisNexis, 3rd Ed, 2020) (“Singapore Income Tax”) at paras 19.212–19.213).

As observed by the authors of Singapore Income Tax at paras 19.213–19.214, this range does not utilise the full sentencing range prescribed by Parliament, which allows for a maximum sentence of up to three years’ imprisonment under s 96(1) of the ITA. As a result, the sentences have clustered at the lower end of the permissible custodial sentencing range in s 96(1) of the ITA. This appears to reflect an anchoring effect which may have been the unintended upshot of the decision in Chng Gim Huat. As emphasised by the Court of Appeal in Suventher Shanmugam v Public Prosecutor [2017] 2 SLR 115 (“Suventher”) at [29] (citing Vasentha d/o Joseph v Public Prosecutor [2015] 5 SLR 122 at [45]–[46]), the courts should generally seek to utilise the full spectrum of possible sentences. To do so otherwise would run the risk of promulgating a sentencing trend inconsistent with Parliament’s stance with respect to each particular offence (see Suventher at [26]). On the facts of the present appeals, this would mean that any sentencing framework specified by this court would necessarily have to encompass sentences up to three years’ imprisonment, as well as non-custodial sentences.

Consistency in sentencing

As Chan Seng Onn J had observed in Takaaki Masui v Public Prosecutor and another appeal [2020] SGHC 265 at [91]–[92], consistency in sentencing encompasses both the issue of adopting a consistent methodology as well as reaching consistent sentencing outcomes. The use of a sentencing framework squarely deals with the former issue (see Public Prosecutor v Wong Chee Meng and another appeal [2020] 5 SLR 807 (“Wong Chee Meng”) at [55] –[57]). However, the latter issue is equally important in that when a court is faced with two very similar cases, it should arrive at broadly similar outcomes. This accords with the principle of consistency namely, that “like cases should be treated alike” (see Chang Kar Meng v Public Prosecutor [2017] 2 SLR 68 at [73], citing Public Prosecutor v UI [2008] 4 SLR(R) 500 at [17]), and by extension, that unlike cases should be treated differently.

It is clear from the sentencing precedents for s 96(1) ITA offences that the precedents do not evince any consistent sentencing trend. For example, it was highlighted by the Prosecution in the hearing below that an offender who had evaded tax of over $400,000 had been sentenced to the same sentence of one month’s imprisonment as an offender who had evaded tax of over $20,000. While I accept that every sentence reflects a complex aggregation of factors which has to be considered with respect to the particular facts of each individual case, the fact that an offence may encompass a “wide variety of circumstances does not, in and of itself, preclude the adoption of a sentencing framework” (see Wong Chee Meng at [56]).

However, from a review of the sentencing precedents tendered, it appears that offences under s 96(1) of the ITA do not in fact encompass a wide-ranging variety of circumstances. The cases, with the exception of Chng Gim Huat, tend to fall generally into two categories. In the first, the offender deliberately fails to declare the full income he has received. In the second, the offender falsifies records to denote a lower income. Accordingly, my observation that the cases concerning s 96(1) of the ITA are not that factually diverse reinforces my view that a sentencing framework would be appropriate, and can be capable of general application.

Did Chng Gim Huat set a benchmark?

I note that the First Appellant had initially adopted the position that Chng Gim Huat sets the sentencing benchmark for s 96(1) ITA offences, but now accepts that Chng Gim Huat does not represent a sentencing benchmark. The Second Appellant has not made any submissions to the effect of whether Chng Gim Huat sets a sentencing benchmark. She has however advocated both here and below that the court should adhere to the overall trend of sentencing precedents.

As stated in Ng Kean Meng Terence v Public Prosecutor [2017] 2 SLR 449 (“Terence Ng”) at [31]–[32], a “benchmark” approach to sentencing is characterised by the following features:

(a) The sentence is imposed in respect of an archetypal case (or a series of archetypal cases) (Terence Ng at [31]);

(b) The notional case must be defined in terms of the factual matrix of the case as well as the sentencing considerations (Terence Ng at [31]);

(c) It provides a focal point against which sentences in subsequent cases with differing degrees of criminal culpability can be accurately determined (Terence Ng at [31]);

(d) It lays out the parameters of its reasoning to allow future judges to determine what falls within the scope of the ‘norm’ (Terence Ng at [31]);

(e) The benchmark approach is particularly suited for offences which overwhelmingly manifest in a particular way or where the manner of offending is extremely common (Terence Ng at [32]).

While Chng Gim Huat had explicitly set out the primary sentencing consideration of deterrence for s 96(1) ITA offences, as well as some other sentencing considerations, such as the time span of the offences, the systems and methods used to evade tax, whether there was restitution of the tax evaded, the amount of tax evaded, and the culpability and circumstances of the defendant, it does not appear to have provided a focal point for comparison or calibration of sentences in subsequent cases. In addition, as alluded to above at [29], the facts of Chng Gim Huat do not appear to be typical of the usual modus of tax evasion – the appellant had wilfully omitted to declare interest income he had obtained from repayments that were made towards personal loans he had extended. I agree with the DJ’s view that Chng Gim Huat did not lay down a benchmark sentence for s 96(1) ITA offences, and should not have been regarded as a sentencing guideline judgment as such.

In summary, in view of the relative paucity of reasoned decisions, lack of consistency in sentencing decisions, as well as the fact that the full sentencing range set out by Parliament has not been utilised, I agree with the DJ that a sentencing framework is called for with respect to offences under s 96(1) of the ITA.

The rationale underlying offences under s 96(1) of the ITA

In Chng Gim Huat, Yong CJ held that offences under s 96(1) of the ITA could be construed as a “deliberate fraud on the State”, and as such offences are difficult to detect or investigate, a custodial sentence is normally warranted on the grounds of general deterrence (see Chng Gim Huat at [107]).

This is also illustrated by the statement of Mr Hon Sui Sen, the Minister for Finance, when presenting the second reading of the Income Tax (Amendment) Bill in Parliament in 1975 (Singapore Parliamentary Debates, Official Report (27 March 1975) vol 34 at cols 1100–1101):

While no evidence was placed before me as to whether income tax remains the mainstay of the State’s revenue, this does not displace the fact that the primary mischief sought to be addressed by criminalisation of tax evasion under s 96(1) of the ITA is the failure of the offender to hand over what is due to the State. As Yong CJ noted in Chng Gim Huat at [107], tax evasion offences affect society as a whole since any resulting deficiencies in revenue would have to be made up by other law-abiding taxpayers. Additionally, prior to the amendments to the ITA in 1975, under s 96(1) of the Income Tax Act (Cap 141, 1970 Rev Ed), apart from the mandatory treble penalty, the punishment for tax evasion was a fine not exceeding ten thousand dollars, or imprisonment for a term of not less than six months and not exceeding three years, or both fine and imprisonment. As a consequence of Parliament’s removal of the minimum of six months’ imprisonment, I am reinforced in my view that Parliament had intended for the entire sentencing spectrum to be utilised.

The relevant sentencing considerations for offences punishable under section 96(1) of the ITA

I turn now to the relevant sentencing considerations that apply in the context of offences punishable under s 96(1) of the ITA. In Terence Ng at [39], the Court of Appeal drew a distinction between offence-specific and offender-specific factors, with the former relating to the manner and mode in which the offence was committed, while the latter related to aggravating and mitigating factors which were personal to the offender. In Public Prosecutor v Koh Thiam Huat [2017] 4 SLR 1099 at [41], I had held that in evaluating the seriousness of a crime, the court has to consider the harm caused by the offence and the culpability of the offender. In Logachev at [36], having considered the two aforementioned cases, Menon CJ held:

The categorisation set out in Logachev neatly encapsulates and delineates the various sentencing considerations that the court should consider in exercising its sentencing discretion. In addition, while there appeared to have been some hesitation on the part of Menon CJ in Logachev to comparing casino cheating cases with cheating simplicter cases under s 420 of the Penal Code (Cap 224, 2008 Rev Ed), I note that this hesitation may have arisen having regard to the differences in sentencing ranges in the offences concerned, and not because the offences bore significant differences (see Logachev at [32]–[33]).

In my view, many of the offence-specific factors listed in Logachev are also present and relevant to offences involving tax evasion. Accordingly, I have incorporated the relevant sentencing considerations from Logachev, that would apply with equal rigour in the context of offences punishable under s 96(1) of the ITA:

Offence-specific factors

Factors going to harm

(1) Quantum of tax evaded

The First Appellant argues that the quantum of tax should not be a determinant of harm for s 96(1) ITA offences, as it is already accounted for by virtue of the mandatory penalty of treble the quantum of tax evaded, which has a strong deterrent effect. On the other hand, the Prosecution argues that notwithstanding the fact that the quantum of tax evaded is considered in the mandatory penalty, relying on the same factor as the primary determinant of harm would not amount to double-counting.

In this regard, it is my view that the objective of the mandatory penalty is not simply to reclaim the revenue defrauded since the multiplier bears no correlation with the time-value of money. Rather, the primary purpose of the high prescribed penalty is to deter tax evasion. This is entirely in keeping with Yong CJ’s opinion in Chng Gim Huat at [105] that the treble penalty is of a different genus from the usual sentencing option of a fine, and is additional to the sentencing options prescribed under s 96(1) of the ITA.

Following from the discussion above, it is apparent that despite the fact that the quantum of income tax evaded necessarily determines the sum of the mandatory penalty imposed, that same quantum can be a primary, albeit not sole, determinant of the harm caused to the State.

(2) Involvement of a syndicate

The fact of involvement of an offender in a criminal syndicate, as well as the size and scale of the syndicate concerned, is an independent aggravating factor that exists quite apart from the element of premeditation, sophistication and planning (see Logachev at [52]–[53]).

In my view, the existence of such an element justifies enhancing the sentence. At the second reading of the Income Tax (Amendment) Bill, the Second Minister for Finance, Mr Lawrence Wong, said as follows (see Singapore Parliamentary Debates, Official Report (2 October 2018) vol 94):

It is plain from the above statements that a tough stance against criminal syndication is warranted. Parliament had made clear its concerns over the involvement of criminal syndicates in tax evasion schemes.

(3) Involvement of a transnational element

In Logachev, Menon CJ held that the involvement of a transnational element is an aggravating factor, because of the increased difficulty of detection as well as the need to prevent Singapore from becoming an essential cog in criminal enterprises (see Logachev at [54]–[55]).

The Prosecution has suggested that in considering this element, the court should consider issues such as whether there was the use of cross-border transactions or offshore companies and trusts to obfuscate the purpose of fund transfers, and conceal the beneficial ownership of taxable assets. Bearing in mind the fine line that can exist between tax avoidance and tax evasion, I am of the view that the approach that the Prosecution submits is sensible, especially in light of the difficulties that underpin the detection of tax evasion when offshore corporate structures are utilised. Such difficulties were among the reasons why Parliament introduced amendments to the Income Tax Act in 2016 to allow for the automatic exchange of financial account information (“AEOI”). At the second reading of the Income Tax (Amendment No. 2) Bill, the Senior Minister of State for Finance, Ms Indranee Rajah, explained as follows (see Singapore Parliamentary Debates, Official Report (9 May 2016) vol 94):

In addition, the Prosecution has suggested an additional harm indicium of “undermining of government schemes or harm done to the confidence in public administration”. However, in my judgment, such a harm factor would run the risk of double-counting, as the premise of tax evasion is the subversion of the governmental scheme of taxation, and the offence attracts a deterrent sentence in view of the impact those very acts will have on public administration and public confidence. Nevertheless, as the categories are non-exhaustive, I accept that the occasion may arise where it can be shown that an act of tax evasion indeed has the additional effect of undermining governmental schemes or damaging the confidence reposed in public administration.

Factors going towards the offender’s culpability

(1) The degree of planning and premeditation.

It is settled law that the presence of planning and premeditation is an aggravating factor, and reflects greater criminality, compared to an offence committed opportunistically or on impulse (see Logachev at [56]). The Prosecution has suggested that examples of such planning and premeditation include the creation of false transactions to reduce the offender’s tax liability. I agree that, in general, evidence that the offender had attempted to hide his income from the scrutiny of the authorities is an aggravating factor. However, I do note that there can be significant overlap between the various factors that go towards the offender’s culpability, and the court should be careful not to double-count the aggravating factors.

(2) Sophistication of the means used to evade income tax or to avoid detection

An offence committed through the use of sophisticated methods will be an aggravating factor in sentencing compared to one done more simply (see Logachev at [57]). I agree with the Prosecution that an indicator of sophistication would be the level of obfuscation employed by the offender, which would possibly encompass the use of complex corporate structures. However, I would again make the observation that the court should be cognisant of the risk of double-counting as there is potential overlap with the aggravating factor of planning and premeditation.

(3) Evidence of a sustained period of offending

While an offence being perpetrated over a sustained period of time is an aggravating factor which could be subject to the sentencing principle of specific deterrence (see Logachev at [59]), the court has to guard against the risk of double-counting especially where the offender has other charges concerning similar offending behaviour which are taken into consideration for sentencing. In this regard, I would also make the observation that offences under s 96(1) of the ITA are unique in the sense that each offence or charge engendered necessarily relates to the offender’s conduct in evading tax for a specific calendar year (ie, year of assessment). Consequently, this offence-specific factor may have only a limited impact on the overall sentence, barring exceptional circumstances.

(4) Offender’s role

The offender’s role may be a relevant aggravating factor, where the facts indicate the involvement of a criminal syndicate. The more influential the offender or the larger the role he has in that syndicate, the stronger the inference that the offender is more culpable and consequently deserving of heavier punishment (see Logachev at [60]).

(5) Abuse of position and breach of trust

In the event that the offender is found to have abused a position of trust or authority or breached the trust reposed in him/her, this would be a relevant aggravating factor (see Logachev at [62]). An example of this would include where the offender has utilised client accounts to evade taxes.

Offender-specific factors

Aggravating factors

The offender-specific aggravating factors of having offences taken into consideration for sentencing purposes, relevant antecedents, and evident lack of remorse, are all well-established in case law and I do not propose to revisit them (see Logachev at [63]–[66]; Terence Ng at [64]).

Mitigating factors

(1) A guilty plea

A guilty plea is a clear mitigating factor in sentencing, on the ground that it would have saved the criminal justice system resources, and that it represents genuine remorse on the part of the offender to accept the consequences of his conduct (see Logachev at [67]). However, with regard to tax evasion offences under s 96(1) of the ITA, there may be cases where the documentary evidence in the form of a “paper trail” is incontrovertible. Coupled with the presumption found in s 96(3) of the ITA, it may be argued that offenders in such circumstances would have been caught red-handed. As noted in Wong Kai Chuen Philip v Public Prosecutor [1990] 2 SLR(R) 361 at [14] and in Public Prosecutor v Tan Fook Sum [1999] 1 SLR(R) 1022 at [33], a guilty plea may therefore be quite inevitable and therefore of little or no mitigating value.

(2) Voluntary restitution

It is generally accepted that restitution made by the offender, without the party cheated having to resort to attempts at recovery, is generally accepted as a mitigating factor on the ground that it reflects the offender’s remorse (see Logachev at [69]). This would similarly apply in the context of s 96(1) of the ITA, as the mischief sought to be addressed, as mentioned above at [36], is the loss of revenue caused to the State.

(3) Co-operation with the authorities

While co-operation with the authorities can be a mitigating factor, the weight to be given to it would very much depend on the specific circumstances of the case (see Logachev at [70]). In the context of offences under s 96(1) of the ITA, this is all the more important in view of the difficulty in detecting such offences, and the fact that income tax returns are heavily dependent on self-reporting.

The sentencing framework for offences punishable under section 96(1) of the ITA

Turning next to the sentencing framework, following from my view at [39] above that the sentencing considerations in Logachev are applicable, I am also of the view that the DJ had correctly accepted that the five-step sentencing framework in Logachev can be transposed to offences punishable under s 96(1) of the ITA, which would apply to offenders who claim trial. The Logachev framework itself has already been adapted for offences concerning financial and commercial crimes. In Huang Ying-Chun v Public Prosecutor [2019] 3 SLR 606, I had adapted the same framework, with suitable modifications, to offences of money-laundering under s 44(1)(a) of the Corruption, Drug Trafficking and Other Serious Crimes (Confiscation of Benefits) Act (Cap 65A, 2000 Rev Ed). In Wong Chee Meng, Menon CJ had also similarly adapted the Logachev framework for offences of corruption under s 6 read with s 7 of the Prevention of Corruption Act (Cap 241, 1993 Rev Ed).

First step: Identify the level of harm and the level of culpability

Having regard to the various offence-specific factors set out in the table at [39] above, the first step is for the sentencing court to identify the fact-specific level of harm caused and the offender’s level of culpability.

I note that the Appellants have each proposed alternative approaches to address the harm element. The First Appellant argues that the determinant of the harm caused should be the proportion of unreported income to the reported income, while the Second Appellant contends that the determinant of harm should be the amount (and proportion) of restitution.

The First Appellant’s suggested approach, appended at Annex A, looks primarily to the amount of unreported income divided by the reported income. This determines whether the offender falls into either of three specified “harm levels”, calibrated according to whether the quotient is up to five times, between five to ten times or above ten times the reported income. In my view, the First Appellant’s suggested “proportion of unreported income” approach is untenable as it would allow individuals with a higher overall income to avoid a higher sentence, as long as the amount of income tax they evade is below a certain proportion. For example, applying the First Appellant’s approach, an individual A who earns $1 million a year and under-reports his income as $900,000, would have an unreported to reported income ratio of one to 0.11, which would fall into the First Appellant’s definition of level 1 harm. On the other hand, an individual B who earns $22,000 a year and under-reports an income of $2,000, would have an unreported to reported income ratio of one to 10, which would fall into the First Appellant’s definition of level 2 or level 3 harm. In such a situation, individual A would be liable for significantly less punishment as the income he had under-reported was far smaller in proportion to the income he had earned, notwithstanding the fact that individual A would have “cheated” the State of five times the amount of tax that individual B did. In my view, such an approach unnecessarily conflates the harm caused to the State with the harm that could have been caused had the offender decided to under-report a greater proportion of his income.

The Second Appellant’s suggested approach, appended at Annex B, proposes that the presence or absence of restitution, or the extent to which restitution has been made, should be the main indicia of harm caused by tax evasion offences. Under this approach, the offender can be said to fall within three specified “harm levels”, with the lowest level involving cases where full restitution has been made. In my view, the Second Appellant’s approach would also not be appropriate for two reasons. First, there may be many reasons why an offender may be unable to make restitution or would choose not to make restitution. Second, if the harm caused is correlated with the amount of restitution made, an offender would be able to “escape” the full consequences of his act no matter how egregious his actions, by simply “reducing” the harm caused by paying back the tax evaded. This would effectively subvert the deterrent effect of s 96(1) of the ITA.

In any event, I note that on either the First or Second Appellant’s suggested alternative approaches, their respective cases would fall within the lowest level of the harm/culpability matrix. This actually mirrors the DJ’s approach, based on the sentencing matrix proposed by the Prosecution, which is set out at [71] below. As such, I do not propose to delve into the details of each of the appellants’ suggested alternative approaches. The real point in contention is where their respective cases ought to appropriately lie within the relevant sentencing range, and how their sentences ought to be calibrated.

In both the Prosecution and the amicus curiae’s submissions, it was submitted that the level of harm should be stratified into one of three levels of severity, based on the quantum of the tax evaded. Both the Appellants argue that the Prosecution’s stratification into the three bands of less than $75,000 (Level 1 harm), $75,000–$150,000 (Level 2 harm), and more than $150,000 (Level 3 harm), are arbitrary, but neither has made substantial submissions as to why it is arbitrary other than pointing to the fact that there appears to be no upper limit.

I do not find the stratification of harm submitted by the Prosecution and the amicus curiae to be arbitrary. The respective levels were derived with reference to the data available from the sentencing precedents. In particular, about 75% of the prosecuted cases involved a quantum of tax evaded of less than $75,000, with roughly 12% of the prosecuted cases encompassing the other two levels of harm proposed respectively. While I am mindful that sentencing is neither a science nor an administrative exercise which can be determined with mathematical certainty (see Dinesh Singh Bhatia s/o Amarjeet Singh v Public Prosecutor [2005] 3 SLR(R) 1 at [24]), it does not mean that the sentencing court should eschew a sensible rationalisation of the various sentencing considerations, where it will facilitate the formulation of a framework to provide clearer practical guidance. On the other hand, it would be inadvisable to attempt to derive a “mathematically perfect graph” to identify a precise point for the sentencing court to arrive at in each case. The latter approach was not favoured by the Court of Appeal in Mohd Akebal s/o Ghulam Jilani v Public Prosecutor and another appeal [2020] 1 SLR 266 at [20]. Similar views were expressed by the three-judge Panel of the High Court in Lee Shing Chan v Public Prosecutor and another appeal [2020] 4 SLR 1174 at [34].

In Logachev at [43]–[50], in relation to an offence of cheating under s 172A of the Casino Control Act (Cap 33A, 2007 Rev Ed) (“CCA”), Menon CJ declined to adopt the Prosecution’s proposed sentencing bands based on the amount cheated for the following reasons:

(a) There is the potential to divert attention away from the other relevant sentencing considerations that go towards the harm caused by the offence. Such an approach is suitable where the offence in question is clearly targeted at a particular mischief which is measurable according to a single (usually quantitative) metric that assumes primacy in the sentencing analysis. On the facts in Logachev, it was determined that s 172A of the CCA sought to target criminal activity in general and not just the amount cheated (see Logachev at [44]);

(b) There is the potential to divert attention away from the relevant sentencing considerations that go towards the offender’s culpability (see Logachev at [47]–[48]);

(c) Singling out the amount cheated is not an approach necessarily supported by precedent (see Logachev at [49]);

(d) Where there is no upper limit to the amount cheated, there is a danger that sentencing bands based solely on the amount cheated might be arbitrary. On the facts of Logachev, it was noted that the lack of evidence to justify the banding militated against the placement of sentencing bands, although the absence of an upper limit did not necessarily bar the creation of sentencing bands (see Logachev at [50]).

With regard to [66(d)] above, in both Edwin s/o Suse Nathen v Public Prosecutor [2013] 4 SLR 1139 (“Edwin”) and Yap Ah Lai v Public Prosecutor [2014] 3 SLR 180 (“Yap Ah Lai”) (referenced by Logachev at [50]) Menon CJ had set out sentencing frameworks based on a specific metric notwithstanding the fact that the statutes in question had not set an upper limit. Reading Logachev together with Yap Ah Lai, it appears that Menon CJ had justified his approach on the basis of Parliament’s intention to control the importation of the controlled substance in question (see Yap Ah Lai at [24]). On this note, Menon CJ also observed in Edwin at [22] in relation to the sentencing benchmark he had set out for offences of drink-driving under s 67(1)(b) of the Road Traffic Act (Cap 276, 2004 Rev Ed) that:

While there is no upper limit to the amount of tax evaded under s 96(1) of the ITA, it is clear that the harm caused is to the State and is measurable according to a single quantitative metric vis-à-vis the amount of income tax evaded. Furthermore, the purpose of s 96(1) of the ITA can be contrasted against the purpose of s 172A of the CCA, in that the former is targeted specifically at tax evasion (see [35]–[36] above) as opposed to the targeting of “criminal activity in general” for the latter (see Logachev at [46]).

As to the effect that the mandatory penalty should have on the determination of the sentence, I am of the view that it should have little influence on the determination of the sentence for two reasons. First, the court has no discretion to determine if the mandatory penalty should be imposed (see Chng Gim Huat at [105]). Second, it would be plainly illogical that the greater the tax avoided and consequently the greater the mandatory penalty, the less likely an offender would be subject to a more severe sentence or longer period of imprisonment.

Accordingly, I am in agreement with the Prosecution that the primary indicium of harm should be the amount of tax evaded as stratified into the three separate levels of harm, which may then be adjusted in terms of “severity” on account of the other non-exhaustive indicia of harm, such as the involvement of a syndicate or the involvement of a transnational element. That being said, I am cognisant of the fact that the average income of the population will change with time. The various stratified levels of harm set out in the sentencing framework are not cast in stone and may have to be revisited from time to time.

Second step: Identify the applicable indicative sentencing range

Having identified the level of harm caused by the offence and the level of the offender’s culpability, the next step is to identify the applicable indicative sentencing range. Bearing in mind the sentencing range set out in s 96(1) of the ITA, and the stratified levels of harm identified, I adopt the indicative sentencing ranges in the Prosecution’s sentencing matrix:

In this regard, I note that the Prosecution has argued that a fine should only be imposed in “exceptional circumstances” following Chng Gim Huat. The peculiarity of s 96(1) of the ITA is that the mandatory imposition of the treble penalty for tax evaded potentially outstrips the “utility” of the maximum $10,000 fine. Deterrence remains the foremost consideration in sentencing for s 96(1) ITA offences, and a custodial sentence is generally the norm, but it is certainly not the case that a custodial sentence must invariably be imposed.

The possibility of a non-custodial sentence should be considered where appropriate, and to that extent, I agree with the Second Appellant’s submission. In my view, a fine can be imposed for offences where the deterrent effect of the fine would not be eclipsed by the imposition of the mandatory treble penalty, for example where the amount of tax evaded is in the lower range. Viewed thus, the holding in Chng Gim Huat at [107] that “a custodial sentence should normally be imposed in order to meet the needs of general deterrence” would not be inconsistent insofar as where the amount of tax evaded would result in a mandatory penalty that would outstrip the maximum fine, a custodial sentence should be imposed as a fine would have little or no deterrent effect. On the facts of Chng Gim Huat, the offender had evaded a total tax liability of $354,645.65 and was subject to a mandatory penalty of over $1 million, which would have rendered the deterrent effect of the maximum $10,000 fine nugatory.

Third step: Identify the appropriate starting point within the indicative sentencing range

Having identified the indicative sentencing range, the third step is to identify the appropriate starting point within that range. The court will have to re-examine the offence-specific factors again. As explained above at [70], in view of the fact that the primary indicium of harm is the amount of tax evaded, the court must be mindful to adjust the “severity” of the harm in relation to the other harm factors, as well as the culpability factors.

Fourth step: Make adjustments to the starting point to take into account offender-specific factors

The fourth step is for the court to make adjustments to the starting point identified in the preceding steps, based on the applicable offender-specific aggravating and mitigating factors. At this stage, it is possible for the adjustment to bring the sentence outside the indicative sentencing range, which are themselves not cast in stone, provided that clear and coherent reasons are given (see Logachev at [80]).

In this regard, the total amount of tax evaded must also be considered at this step. This is consistent with Gan Chai Bee Anne v Public Prosecutor [2019] 4 SLR 838 (“Anne”) at [39] and [75], where Menon CJ held that the total amount involved in the charges proceeded with and taken into consideration should be analysed as part of the totality of the offender’s conduct:

Fifth step: Make further adjustments to take into account the totality principle

The fifth step is for the sentencing court to make further adjustments to the sentence on account of the totality principle. In cases where an offender is convicted of multiple charges, the general rule is that consecutive sentences should be ordered for unrelated offences, subject to the totality principle, the one-transaction rule, and any statutory provisions that supersede the general rule (see Public Prosecutor v Raveen Balakrishnan [2018] 5 SLR 799 (“Raveen”) at [65]–[67]).

In relation to the totality principle, the court has to first examine whether the aggregate sentence is substantially above the sentences normally meted out for the most serious of the individual sentences committed, then proceed to consider whether the effect of the sentence on the offender is crushing and not in keeping with his past record and his future prospects (see Mohamed Shouffee bin Adam v Public Prosecutor [2014] 2 SLR 998 at [54] and [57]).

Application of the sentencing framework to the First Appellant

I now turn to consider the First Appellant’s case, and to determine if the sentencing framework, which had been similarly adopted by the DJ, had been appropriately applied.

Based on the amount of tax evaded, which was between $34,000 to $35,000 for each of the proceeded charges, I agree with the DJ that the harm caused was justifiably classified as “Level 1 harm”. As to the culpability of the First Appellant, I agree that there was a degree of planning and premeditation on his part when he conspired with Lim to ensure that TNTPL’s sales revenue did not exceed $1 million in order to avoid GST registration, and to ensure that TAS’s net profit was kept below $100,000, in addition to having failed to report holiday reimbursements. I also agree that there was no evidence of the First Appellant having used sophisticated means to evade his income tax. The substantial length of time the scheme was in place as well as the First Appellant’s role in the scheme had been accounted for.

However, I find it difficult to accept the DJ’s finding that the primary aim of the scheme was not for the First Appellant to evade income tax for two reasons. First, while the artificial transfer of sales from TNTPL to TAS was necessary to enable TNTPL to avoid GST registration, I do not see why TAS’s net profit figures had to be subject to an artificial cap of $100,000. The First Appellant could have just as easily ensured that the total amount of income he actually received was reported even if he had sought to avoid GST registration for TNTPL. Second, the First Appellant’s deliberate failure to report his holiday reimbursements is another indicator that he had attempted to artificially reduce the amount of income he had received for each of the taxable years involved.

Notwithstanding, I agree that at the first step of the sentencing framework, the First Appellant’s culpability was low, and that consequently the applicable sentencing range at the second step of the sentencing framework is up to six months’ imprisonment. At the third step of the sentencing framework, I agree that in light of the absence of other harm factors, a “starting point” sentence of ten weeks’ imprisonment was appropriate. In the First Appellant’s case, in view of the quantum of income tax evaded per charge which was in itself already over three times the maximum fine that can be imposed, a fine would manifestly not be an appropriate sentence.

At the fourth step, I agree with the aggravating factor identified by the DJ, which is the six other charges under s 96(1) of the ITA being taken into consideration for the purposes of sentencing, and which amounted to a total tax evaded of $119,186.21.

With regard to the First Appellant’s contention that the DJ had failed to consider the mitigating value of the First Appellant’s early plea of guilt, cooperation with IRAS, and full restitution for all the charges concerned, I am of the view that this submission is without merit. The DJ had correctly identified and given careful consideration to these factors in calibrating the First Appellant’s sentence downwards substantially from ten weeks’ imprisonment to six weeks’ imprisonment on each charge. I agree with the Prosecution that the sentences in fact appear to be fairly lenient, rather than being manifestly excessive.

At the fifth step, I am of the view that the DJ had not breached the one-transaction rule in any way by ordering both sentences to run consecutively. In Raveen, Menon CJ had stated:

Application of the sentencing framework to the Second Appellant

Next, I turn to consider the Second Appellant’s case, and whether the sentencing framework had been appropriately applied by the DJ.

Based on the amount of tax evaded, which was $79,142.13 for the proceeded s 96(1) ITA charge, the harm caused would be justifiably classified as “Level 2 harm”. I note that the DJ had re-classified the harm caused into “Level 1 harm”, in view of the lack of other harm factors, but conceptually it would be more appropriate to make such an adjustment at the third step of the sentencing framework.

With regard to the culpability of the Second Appellant, I agree with the DJ that the Second Appellant’s scheme was unsophisticated, as it essentially comprised of her asking Lucy to reduce the actual sales figures of either FCC or YHFT to below the $1 million threshold in order to avoid GST registration. No other subterfuge was employed to disguise her scheme.

However, I find difficulty agreeing with the DJ’s finding that the primary aim of the scheme was not for the Second Appellant to personally benefit from income tax fraud but to benefit her businesses instead. While the act of evading income tax was inextricably linked to the lowering of FCC’s and YHFT’s sales income in order to avoid GST registration, the Second Appellant was not prevented from declaring the total income she received for that year of assessment. It cannot be said that whenever an offender commits an offence under s 96(1)(b) of the ITA in consequence of having committed another tax offence, his culpability is reduced just because his primary intention was not to evade paying income tax.

Notwithstanding, I agree that at the first step of the sentencing framework, the Second Appellant’s culpability was low. However, in light of my finding that the harm caused was pegged at “Level 2 harm”, the applicable indicative sentencing range lay between six to 12 months’ imprisonment. At the third step of the sentencing framework, in light of the absence of other harm factors and the fact that the amount of tax evaded (ie, $79,142.13) lay just between the two levels of harm, the DJ was not wrong to have exercised his discretion to adjust the indicative starting sentence substantially downwards to 16 weeks’ imprisonment, bearing in mind that the categories set out in the sentencing matrix should not be regarded as “rigid or impermeable” (see Edwin at [22] and my observations to a similar effect in Public Prosecutor v Tan Thian Earn [2016] 3 SLR 269 at [32]).

At the fourth step, I agree with the aggravating factor identified by the DJ, which is the additional charge under s 96(1) of the ITA being taken into consideration for the purposes of sentencing. The tax evaded amounted to $7,654.46. The DJ however also gave full consideration to the mitigating factors, namely her early plea of guilt and cooperation during the investigations and emphasised the fact that full restitution had been made.

With regard to the Second Appellant’s contention that the DJ had failed to consider the Second Appellant’s personal circumstances, I am of the view that this submission is unmeritorious. The DJ had given due consideration to the Second Appellant’s personal circumstances and found that they did not warrant a further discount on the sentence. I am in agreement with the DJ on this point. The case of Tan Kiam Peng v Public Prosecutor [2008] 1 SLR 1 submitted by the Second Appellant is clearly distinguishable as the Second Appellant’s personal circumstances did not create a dire and exceptional situation which compelled her to commit the offences.

Finally, at the fifth step, as only one charge was proceeded with in the case of the Second Appellant, there is no need to adjust the sentence of ten weeks’ imprisonment on the basis of the totality principle. Overall, I do not see the need to disturb the sentence, though I do agree with the Prosecution that the sentence would appear to have been slightly lenient.

Conclusion

For the foregoing reasons, I am of the view that the individual and aggregate sentences in respect of both appellants are not manifestly excessive. Accordingly, I dismiss both appeals. I would like to convey my appreciation to the amicus curiae for diligently preparing an objective, detailed and comprehensive analysis from which I derived considerable assistance.

Annex A

First Appellant’s suggested sentencing matrix

Annex B

Second Appellant’s suggested sentencing matrix