1 The Workplace Safety and Health Act 2006 (2020 Rev Ed) (the “Act”) was enacted to improve workplace safety by effecting a cultural change for employers and other stakeholders to take proactive measures to prevent accidents.

2 In line with that intent, Part 4 of the Act sets out a list of duties and offences of persons at the workplace, spanning different stakeholders and catering for different mental elements (the “Part 4 offences”). Notwithstanding this diversity, the Part 4 offences are punishable under the omnibus provision of s 50 save where otherwise specified:

3 The present appeal relates only to s 12(1) read with s 20 of the Act, ie, the breach of an employer’s duty to take reasonably practicable measures to ensure the safety and health of its employees. The Prosecution, which is the appellant, invites this court to review and revise the relevant existing sentencing frameworks relating to bodies corporate, as set out in Public Prosecutor v GS Engineering & Construction Corp [2017] 3 SLR 682 (“GS Engineering”) and MW Group Pte Ltd v Public Prosecutor [2019] 3 SLR 1300 (“MW Group”). The Prosecution further proposes that the revised sentencing framework should be applicable to all Part 4 offences punishable under s 50(b) of the Act.

4 The underlying facts of this appeal concern a worker who was struck by the suspended jib of a tower crane being erected on a vessel at a shipyard. The jib had not been rigged according to the manufacturer’s configuration. Unfortunately, the worker passed away as a result of his injuries.

5 Consequently, the respondent, which was the employer of the deceased, was charged with, pleaded guilty to, and was convicted of a charge under s 12(1) read with s 20 of the Act:

6 Sections 12(1) and 20 of the Act provide as follows:

7 In his grounds of decision on sentence (the “GD”), the District Judge (the “DJ”) applied the sentencing frameworks set out in GS Engineering and MW Group in relation to offences under s 12(1) read with s 20 of the Act. Although he noted that there were divergences between the sentencing frameworks, he did not consider them to be inconsistent with one another (GD at [51]). He found that in negligently failing to implement safe work procedures and a lifting plan, the respondent’s culpability was moderate (GD at [60]–[73]). He also found that the potential harm inherent in the rigging of the jib was high as one other worker besides the deceased was exposed to such risk (GD at [74]–[75]). Based on the applicable sentencing ranges in the frameworks for such cases of a fine between $120,000 and $300,000 (applying MW Group) or between $150,000 and $300,000 (applying GS Engineering), the DJ fixed the starting sentence at a fine of $210,000, which he took to be the midpoint of the respective sentencing ranges (GD at [77]). After balancing the aggravating factor on account of the death of the deceased against several mitigating factors such as the respondent’s early plea of guilt and its otherwise unblemished safety record, the starting sentence was marginally uplifted by $10,000 to $220,000 (GD at [80]).

8 In arriving at his decision, the DJ noted that the frameworks in GS Engineering and MW Group were likely in need of reconsideration in light of the High Court’s decision in Mao Xuezhong v Public Prosecutor and another appeal [2020] 5 SLR 580 (“Mao Xuezhong”), which related to offences under what is now 15(4) of the Act and which took a different approach in its consideration of harm and culpability (GD at [57]). Nonetheless, the DJ considered himself bound by GS Engineering and MW Group, and applied the frameworks in those cases accordingly (GD at [58]).

9 In this appeal, the Prosecution seeks to substitute the GS Engineering and MW Group sentencing frameworks with one more akin to the framework in Mao Xuezhong. Notably, it submits that its proposed sentencing framework should be applicable not just to offences under s 12(1) of the Act, but to all Part 4 offences punishable under s 50(b) generally. Applying its proposed sentencing framework, the Prosecution argues that a heavier fine of $260,000–$280,000 is appropriate.

10 The respondent, meanwhile, submits that no departure from the GS Engineering and MW Group frameworks is warranted, despite candidly acknowledging that the principles of the Mao Xuezhong sentencing framework are applicable to offences under s 12(1) of the Act. In any event, it argues, should a new sentencing framework be formulated, the doctrine of prospective overruling should be applied, such that the new sentencing framework should not apply to the present case.

11 To assist us in this appeal, we appointed a young amicus curiae, Ms Loh Jia Wen Dynyse (the “amicus”). The amicus agreed with the Prosecution that a framework based on Mao Xuezhong was appropriate.

12 The following issues present themselves for consideration:

(a) Whether the sentencing frameworks in GS Engineering and MW Group in relation to s 12(1) of the Act should be replaced with a new sentencing framework;

(b) If a new sentencing framework is to be established, whether it should apply to all other Part 4 offences punishable under s 50(b);

(c) If a new sentencing framework is to be established, whether this court should apply the doctrine of prospective overruling; and

(d) Whether the sentence meted out to the respondent should be altered.

13 We begin by tracing the development of sentencing frameworks for s 12(1) of the Act in GS Engineering and MW Group, along with the parallel developments in relation to the present s 15(4) of the Act in Mao Xuezhong.

14 A sentencing framework for s 12(1) of the Act was first formulated in GS Engineering. This took the form of a two-stage framework: in the first stage, an indicative starting point sentence is derived from the potential harm caused by the offence and the culpability of the offender, using the following sentencing benchmarks (GS Engineering at [77(a)]):

15 In the second stage, the sentence is calibrated by taking into account the offender-specific aggravating and mitigating factors of the case. It is here that the actual harm caused by the offence is considered: serious actual harm would be considered an aggravating factor (GS Engineering at [77(d)] and [77(e)]).

16 Subsequently, the GS Engineering sentencing framework was further refined in MW Group in two ways. First, the judge in MW Group explicitly identified and endorsed the implicit feature of the GS Engineering framework that more weight should be accorded to the potential harm caused by the offence than to the culpability of the offender (MW Group at [35]–[37]). Second, notwithstanding his broad agreement with the GS Engineering framework (MW Group at [26]), the judge in MW Group departed from the specific sentencing benchmarks laid down in GS Engineering. He considered that the presence of gaps therein between the sentencing bands for each level of potential harm was problematic, both because it meant that the statutory sentencing range was not being fully utilised, and because the large “jumps” were disruptive to the principle of proportionality (MW Group at [31]–[32]). Hence, the court in MW Group refined a table of sentencing benchmarks of its own (MW Group at [50]):

17 In our view, the modifications proposed in MW Group did not represent a radical departure from GS Engineering. The underlying principles guiding the derivation of an appropriate starting point for a sentence remained the same, as did the considerations driving the evaluation of culpability, potential harm, and the relevant aggravating and mitigating factors (MW Group at [26]–[28] and [60]–[61]).

18 However, we note that the disparity in the sentencing benchmarks set out in GS Engineering and MW Group is liable to result in difficulties for lower courts seeking guidance from the High Court, leading to uncertainty and inconsistency in sentencing. This is antithetical to the objective of sentencing guidelines.

19 The next key development in the case law was Mao Xuezhong. In that case, the High Court was asked to consider the appropriate sentencing framework for offences under s 15(3A) of the Workplace Safety and Health Act (Cap 354A, 2009 Rev Ed) (now s 15(4) of the Act):

20 The prevailing precedent in respect of such offences at that point was Nurun Novi Saydur Rahman v Public Prosecutor and another appeal [2019] 3 SLR 413 (“Nurun Novi”), which set out a two-stage sentencing framework similar to those in GS Engineering and MW Group; in particular, the first stage of identifying a starting sentence was based on culpability and potential harm, with greater weight being placed on the latter.

21 The High Court in Mao Xuezhong declined to endorse the Nurun Novi framework (Mao Xuezhong at [48]–[61]). While it adopted a similar two-stage framework of identifying an indicative starting point and calibrating thereafter based on offender-specific factors as well, the first stage of the framework departed from the approach in Nurun Novi (and therefore from GS Engineering and MW Group as well) in two notable ways:

(a) First, the first stage of the framework was to consider harm in the sense of both potential harm and actual harm (ie, the harm occasioned which bears a causal or contributory link created by the negligent act) caused, rather than considering only potential harm and relegating actual harm to the second stage; and

(b) Second, the first stage of the framework was to give equal consideration to both harm and culpability.

22 Notwithstanding that Mao Xuezhong was a case relating to what is now s 15(4) of the Act and not s 12(1), the parties and the amicus are in agreement that the principles underlying the approach in Mao Xuezhong – ie, the consideration of both potential harm and actual harm in the first stage of the sentencing framework, and the equal weight accorded to harm and culpability in that first stage – should apply to offences under s 12(1) read with s 20 of the Act as well.

23 We agree with this submission. The GS Engineering and MW Group frameworks place particular emphasis on potential harm, taking it to be the dominant consideration in determining an indicative starting sentence. This approach was said to be rooted in the legislative intent of the Act, as demonstrated in the material parliamentary debates (see GS Engineering at [50] and [65] and MW Group at [35]). However, as alluded to in Mao Xuezhong at [67], support for this approach is not apparent in the parliamentary debates. It is therefore useful to set out below the relevant portions of the speech by Dr Ng Eng Hen, the then-Minister for Manpower (Singapore Parliamentary Debates, Official Report (17 January 2006), vol 80 at cols 2206 and 2215):

24 As noted by Dr Ng, the Act was intended to address the “inadequacy” of the penalty regime of its predecessor, which was concerned solely with actual harm. To that end, the Act sought to deter poor safety management and effect a cultural change for employers and other stakeholders to take proactive measures to prevent accidents at the workplace, as was rightly pointed out in GS Engineering at [51]. However, the fact that Parliament consciously sought to expand the recognition of harm in the Act to include potential harm does not necessarily mean that potential harm was thereafter to be accorded precedence over other factors. We note that Dr Ng’s explanation of “all the relevant circumstances” listed culpability, potential harm and actual harm without giving primacy to any of them. The better understanding of the Act, in our opinion, is that it was not meant to displace the significance of actual harm, but was instead meant to give effect to a more nuanced notion of “harm”: not just “risks and accidents”, but the underlying “economic and social cost” as well.

25 In this light, we hold that the proper approach is to evaluate the level of harm as a whole, including both the potential harm and the actual harm. Indeed, when actual harm has occurred, as is often the case in matters prosecuted under s 12(1) of the Act, it is artificial to relegate it to a secondary concern, rather than treating it as one of the primary factors in determining the indicative starting sentence for an offender. As noted in Public Prosecutor v Hue An Li [2014] 4 SLR 661 (“Hue An Li”) at [68]–[74], in cases of criminal negligence, the extent of the harm which eventuates from the negligent act is a relevant sentencing factor. This proposition is equally applicable to an employer’s failure to take reasonably practicable measures in relation to the health and safety of its employees at work, which in our view is akin to criminal negligence.

26 Ultimately, we recognise that there may not be any significant difference between the approach to harm favoured by GS Engineering and MW Group and that which we now endorse, in terms of the eventual sentence. However, in our judgment, the holistic consideration of harm in the first stage of the framework better reflects both the intent underlying the Act and the “intuitive moral sense that outcomes do matter” (Hue An Li at [70]).

27 As for the relationship between harm and culpability, it was noted in Mao Xuezhong that both were equally important considerations in s 15(3A) offences, and that the debates concerning the Act did not favour either over the other (at [67]). We agree and find that the same equally applies to s 12(1) offences.

28 We therefore hold that the appropriate sentencing framework, inclusive of benchmarks for indicative starting sentences (the “Framework”), for offences under s 12(1) read with s 20 of the Act is as follows:

(a) In the first stage of the Framework, the sentencing judge is to determine the level of harm and the level of culpability, in order to derive the indicative starting point according to the benchmarks set out below:

(b) In evaluating the level of harm, we agree with the factors set out in GS Engineering at [77(c)] and MW Group at [27] in assessing potential harm: namely, the seriousness of the harm risked; the likelihood of that harm arising; and the number of people likely to be exposed to the risk of the harm. We also adopt in relation to s 12(1) offences the guidance from Mao Xuezhong that “[w]here the harm was likely to be death or serious injury (such as paralysis or loss of a limb), the harm could be considered to be high even though it did not materialise. If death or serious injury did occur, the harm would be graded near the top end of the high range.” (at [64(a)(i)]).

(c) As for the assessment of culpability, we similarly agree with the factors set out in GS Engineering at [77(b)] and MW Group at [28]: namely, the number of breaches or failures; the nature of the breaches; the seriousness of breaches; whether the breaches were systemic or isolated; and whether the breaches were intentional, rash or negligent.

(d) In the second stage of the Framework, the starting sentence should be calibrated according to offender-specific aggravating and mitigating factors. Notably, the actual harm caused should no longer be considered an aggravating factor, as it will already have been accounted for at the first stage of the analysis. Otherwise, we see no reason to depart from the aggravating and mitigating factors which were considered in GS Engineering at [77(e)] and MW Group at [60]–[61].

29 In respect of the benchmarks set out above, we note that the parties and the amicus each submitted roughly similar sets of benchmarks, distinguished only by slight differences in the numerical bands for each harm-culpability combination. We have selected the benchmarks presented by the Prosecution, which in our view best reflect the appropriate indicative starting sentences for each harm-culpability combination. In particular, we note that there is an unusually wide range of outcomes which may result from breaches of the duty under s 12(1) of the Act, from scenarios where no harm to persons or property was occasioned at all to catastrophes such as the Nicoll Highway collapse, to cite an example given in GS Engineering (at [55] and [87]). At the same time, the spectre of disaster should not skew the sensibilities of the court: we reiterate that death or serious injury to even a single person should still be considered an instance of high harm. In light of these considerations, a wider range of indicative starting sentences is warranted for the myriad situations that might be encapsulated in the high-harm, high-culpability category of cases, with sentences at or closed to the prescribed maximum fine being reserved for the types of disasters that involve significant loss of life or great loss to the economy and severe inconvenience to the public (see Mao Xuezhong at [68]).

30 The facts of this appeal only require us to consider the appropriate sentencing guidelines in relation to s 12(1) of the Act. Nevertheless, the Prosecution makes a persuasive case for the application of the Framework to all other Part 4 offences punishable under s 50(b).

31 Part 4 of the Act sets out the duties in relation to different categories of stakeholders in the workplace. Unless otherwise specified, the contravention of these duties is an offence under s 20 of the Act, and is punishable under s 50(a) for offenders who are natural persons, and s 50(b) for offenders which are bodies corporate.

32 In respect of s 50(b), we note that the Part 4 duties which can apply to bodies corporate and for which contravention is punishable under s 50(b) – namely, ss 11, 12(1) and (2), 14(1) and (3), 14A(1), 16(1), 17(1) and 19(2) – are largely similarly formulated. Each of these, with the exception of s 19(2), mandates that the category of stakeholders specified therein is to take reasonably practicable measures to ensure the safety and health of other parties in the workplace. Section 19(2), while not utilising this same specific language, provides that occupiers of common areas are to comply with any provision of the Act with respect to certain portions of the common area; it therefore imposes on these occupiers of common areas a similar duty to take reasonably practicable measures to ensure the safety and health of other parties.

33 Naturally, the specific contents of the duty to take reasonably practicable measures will differ between each category of stakeholders. Nonetheless, as the Prosecution pointed out, the shared language employed in the formulation of these duties indicates a common conceptual standard to which each stakeholder is to be held: to proactively take reasonable measures to address the risks it can be expected to control. In our view, the breach of these duties involves a common mental state akin to negligence (as noted at [25] above). Further, the slate of potential outcomes which may result from a breach by any given stakeholder is broadly similar (see [29] above). However, we generally do not develop a framework for offences which are not squarely before us. That said, subject to further arguments when an appropriate case is before us, our indicative view is that the two-stage sentencing approach as well as the benchmarks as outlined above should be applicable to Part 4 duties for which contravention is punishable under s 50(b).

34 Three additional observations are nonetheless in order. First, it should be noted that the harm risked by certain categories of stakeholders in breaching their duties may likely be more serious. For instance, s 16(1) of the Act requires persons who manufactures or supplies any machinery, equipment or hazardous substance for use at work to take various measures as far as is reasonably practicable. The involvement of such industrial tools makes it quite possible that a breach of s 16(1) may be more likely to result in greater harm than an employer’s breach of s 12(1). Nonetheless, this speaks to the possibilities of various outcomes, but not the overall range of potential outcomes. The holistic consideration of harm at the first stage of the Framework would accommodate and address such instances: the greater harm risked or caused may be accounted for by assigning a higher rating of harm to such breaches.

35 Second, the above analysis should in principle only apply to Part 4 offences punishable under s 50(b), ie, only where the offender is a body corporate. In this respect, we would note that the duties which we have listed at [‎32] above also apply to natural persons who fall into the specified categories of stakeholders. There are also two further Part 4 duties for which contravention is punishable under s 50, namely, s 13(1) and s 15(3):

Although s 2 of the Interpretation Act 1965 provides that “person” shall generally include “any company or association or body of persons, corporate or unincorporate”, it is plain that in the context of s 13(1) and s 15(3), “self-employed person” and “person at work” can only refer to natural persons.

36 Where the offender is a natural person, the sentencing options provided for under s 50 are different: a maximum fine of $200,000 and/or imprisonment for a term not exceeding two years, rather than the maximum fine of $500,000 applicable to bodies corporate. It follows that the benchmarks we have outlined at [28(a)] above cannot simply be transposed to natural persons punishable under s 50(a). Nor will any sort of formulaic conversion of the benchmarks be appropriate: as we noted in Mao Xuezhong at [55], a “treatment of fines and imprisonment as interchangeable and ‘convertible’ is difficult to justify in principle”, with the two forms of punishment being qualitatively different.

37 Nonetheless, we are of the view that the overall two-stage sentencing approach we have outlined above should remain applicable to natural persons punishable under s 50(a). Where the same duty to take reasonably practicable measures to ensure safety and health applies, the same considerations surrounding harm, culpability and the relevant aggravating and mitigating factors prevail, notwithstanding that the offender is a natural person. This should extend to s 13(1) as well, which is also framed with reference to reasonably practicable measures necessary to ensure safety and health. Similarly, while the mental element encapsulated in s 15(3) – wilfulness or recklessness – is different from the negligence that characterises the other offences punishable under s 50(a), we are of the provisional view that the two-stage sentencing approach should remain applicable. Preliminarily, we observe that an offender who is reckless or wilful will be considered to be of higher culpability, subject to other factors relevant to the assessment of culpability.

38 Finally, there are also Part 4 offences for which the applicable punishments are separately specified and distinct from those prescribed in s 50 (namely, ss 15(1), 15(2), 15(4) and 18 of the Act). As with those offences punishable under s 50(a), it follows that the benchmarks set out at [28(a)] above would not be applicable. Further, these duties are largely of a different character from those for which contravention is punishable under s 50. Different considerations may underlie these duties, which may render the two-stage sentencing approach we have outlined above inappropriate (notwithstanding that a similar approach was applied to s 15(4) in Mao Xuezhong). For instance, s 18 of the Act sets out a variety of duties owed by occupiers and employers, such as a duty to not dismiss or threaten to dismiss a whistleblower employee (s 18(2)(a)). In such a case, harm and culpability may not be the dominant sentencing considerations; even if they are, the factors by which they might be measured are likely to be quite different. However, given that no submissions were placed before us as to the appropriate approach to sentencing for breaches of Part 4 duties not punishable under s 50, we reserve consideration of this issue to a future case.

39 In short, we are provisionally of the view that the two-stage sentencing approach we have outlined above should in principle apply to all Part 4 offences punishable under s 50 of the Act. In addition, where the offender is a body corporate, the benchmarks we have set out at [28(a)] above should apply. Appropriate benchmarks for natural persons punishable under s 50(a) and for offences for which punishments are otherwise specified in the Act may be considered and developed in future cases.

40 Having decided that the Framework should apply to Part 4 offences punishable under s 50(b) of the Act, we turn to consider whether the doctrine of prospective overruling is applicable, as the respondent contends.

41 Judicial pronouncements are retroactive by default, and the doctrine of prospective overruling may only be exceptionally invoked where it is necessary to avoid serious and demonstrable injustice to the parties or the administration of justice (Adri Anton Kalangie v Public Prosecutor [2018] 2 SLR 557 at [39]–[40] and [43]). In determining whether there is such injustice, the court considers (a) the extent to which the law or legal principle concerned is entrenched; (b) the extent of the change to the law; (c) the extent to which the change to the law is foreseeable; and (d) the extent of reliance on the law or legal principle concerned (Hue An Li at [124]).

42 We decline to apply the doctrine of prospective overruling on the facts of the present appeal. The respondent has not demonstrated a case of injustice. GS Engineering and MW Group are both relatively recent decisions, and the mere application of the sentencing frameworks therein in the courts below does not render them entrenched. The Framework we have adopted is largely based on principles similar to those set out in Mao Xuezhong, and the departure from GS Engineering and MW Group is more akin to a recalibration than a seismic change. This recalibration would also have been reasonably foreseeable: Mao Xuezhong was decided in May 2020 and the respondent pleaded guilty in January 2021. Pertinently, the analysis undertaken in Mao Xuezhong referred to both GS Engineering and the principles underlying the Act as a whole (see Mao Xuezhong at [63]–[68]). Finally, there is no apparent element of reliance here: there is no suggestion by the respondent that it had somehow relied on the GS Engineering and MW Group frameworks in improperly rigging the jib or in pleading guilty.

43 We now apply the Framework to the facts of this appeal.

44 The parties do not dispute the DJ’s finding that the respondent was of moderate culpability in failing to ensure the safety of the deceased employee, and that the potential harm caused by this failure was high. The respondent’s offence therefore falls into the moderate culpability and high harm category in the sentencing benchmarks set out at [28(a)] above. The indicative sentence would be a fine between $225,000 and $300,000.

45 We see no reason to depart from these findings, and add only that per [28(b)] above, the actual occurrence of death in the present case means that harm should be positioned near the top end of the high range. The indicative sentence should be gauged accordingly, ie, closer to $300,000.

46 Turning to the offender-specific factors in this case, we note that the only aggravating factor found by the DJ was that the offence had caused the death of the employee. However, we have already taken this into account in the consideration of harm in the first stage. We also see no reason to depart from the mitigating factors identified by the DJ, namely, that the respondent had pleaded guilty at an early stage, that it had cooperated with the investigations, that it had an otherwise unblemished safety record and that it had put in place post-accident rectification works (GD at [79]).

47 Having considered that the harm should be classified near the top end of the high range, we hold that a sentence of a fine of $250,000 is appropriate after taking into account the relevant mitigating factors.

48 The respondent argues that the fine imposed by the DJ of $220,000 is not manifestly inadequate, and so should not be enhanced. However, the fact that the approach taken below to derive the sentence has now been revised suffices for a re-evaluation of the sentence in principle to justify an enhancement of the sentence.

49 Accordingly, we allow the Prosecution’s appeal against sentence, and substitute the fine of $220,000 imposed by the DJ with a sentence of a fine of $250,000.

50 We thank the parties and the amicus for their helpful submissions in this matter.