1 The aim of all sentencing courts, without exception, is to arrive at an appropriate sentence that befits the crime committed by the offender, after having regard to all the relevant facts, circumstances and societal context surrounding the offence. It is this infinite permutation of relevant considerations that renders sentencing a fluid exercise, and which accords the sentencing judge a degree of flexibility and autonomy in arriving at the appropriate sentence. This appropriate sentence is not defined by a rigid formula or a set of unyielding rules. Instead, the court’s discretion is guided by broad general principles of sentencing.

2 In Singapore, the sentencing process is further aided by the recent proliferation of sentencing guidelines and frameworks in our courts’ jurisprudence that guide, rather than restrict, the sentencing court’s discretion. These judicial creations take a wide variety of forms and have been applied to a wide assortment of offences. Constructed well, they are tools that promote consistency and transparency in our criminal justice system (amongst other aims), while reducing uncertainty and arbitrariness. Constructed poorly, they may generate unintended gaps, discontinuities, ceilings and/or minimum sentences which may result in incoherence and uncertainty in the sentencing process.

3 When constructing frameworks and guidelines, the form that each framework or guideline takes is secondary. What matters is its substantive content, and whether it adheres to and abides by the broad general principles of sentencing. The present appeals present an opportunity to revisit some of these broad general principles of sentencing.

4 The present appeals concern one of Singapore’s largest private sector corruption cases to date. The appellants, Takaaki Masui (“Masui”) and Katsutoshi Ishibe (“Ishibe”), each faces 28 charges under s 6(a) read with s 29(a) of the Prevention of Corruption Act (Cap 241, 1993 Rev Ed) (“PCA”) for conspiring with one another to corruptly obtain bribes from one Koh Pee Chiang (“Koh”) as an inducement for doing acts in relation to their employers’ affairs. After a 15-day trial, the District Judge (“DJ”) convicted them on all charges. His decision can be found in Public Prosecutor v Katsutoshi Ishibe and another [2018] SGDC 239 (“Decision”). The DJ sentenced each appellant to 66 months’ imprisonment and a penalty of S$1,025,701 (in default to serve six months’ imprisonment) (Decision at [129], [132]).

5 The first charge for Masui is reproduced as follows:

6 The remaining 27 charges for Masui differ only in respect of: (a) the name of the appellants’ principal as it was renamed after a corporate merger; (b) the date on which the gratification was received; and (c) the amount of gratification received. Ishibe, as the co-conspirator faced the same 28 charges, except that the individual receiving the gratification from Koh was always Masui. For ease of reference, I refer to their respective charges as C1 to C28, with the understanding that each of these refers to one charge for Masui and/or one charge for Ishibe, as the case may be, ie, that C1 represents the first charge proceeded against Masui and/or Ishibe, as the case may be.

7 The appellants appealed against both their conviction and sentence, and their appeals were heard over the course of four non-consecutive days between 26 July 2019 and 21 August 2020.

8 On 26 July 2019, I allowed Ishibe’s criminal motion in Criminal Motion No 35 of 2019 to adduce further evidence in the form of the original charges against him dated 25 February 2015. ¹ I also allowed Masui’s criminal motion in Criminal Motion No 36 of 2019 to adduce further evidence in the form of two versions of an email from Masui to Koh dated 10 February 2004, an affidavit from a forensic consultant in respect of the email and the original charges against him dated 25 February 2015. ²

9 On 24 February 2020, after hearing the parties’ submissions and going through the evidence in substantial detail, I upheld the DJ’s conviction on all 28 charges. However, I amended the gratification quanta stated in the appellants’ C21 from S$102,115 to S$86,275, and C25 from S$137,340 to S$111,211. The total quantum of gratification received by the appellants is thus S$2,009,433.

10 Since the DJ’s decision on 19 September 2018, the law on sentencing for corruption offences has developed rapidly. Two new high court decisions on this general subject were published by the time the hearing of these appeals was completed: Hoo Sheau Peng J’s decision in PP v Tan Kok Ming Michael and other appeals [2019] 5 SLR 926 (“Michael Tan”) which involved the corruption of foreign public officials, and Sundaresh Menon CJ’s decision in Public Prosecutor v Wong Chee Meng and another appeal [2020] SGHC 144 (“Wong Chee Meng”) which laid down a sentencing framework for offences under s 6 read with s 7 of the PCA. I will deal with the impact of these decisions as and when they arise in the course of this judgment.

11 In line with these developments, the parties’ positions have also evolved during these appeal proceedings. To avoid confusion, I have taken the latest positions of the parties to be their final positions and will be referring to them throughout the course of this judgment, unless otherwise stated.

12 Apart from the exact quantum of the gratification received by the appellants, I am in agreement with the DJ’s findings of fact which can be found at [8]–[73] of the Decision. As this judgment focuses on the appeals against sentence, I shall only reproduce the salient facts as are necessary for an appreciation of the issues on sentence. I will also explain my decision to amend C21 and C25, and Ishibe’s new argument at the end of this section (see [32] onwards).

13 At the material time, the appellants worked as employees of Nissho Iwai Corporation (“Nissho Japan”), and following a merger between Nissho Japan and another company in April 2004, as employees of Sojitz Corporation (“Sojitz Japan”). ³ At various points in their careers with the Japan Company, the appellants were seconded to Singapore to work for the wholly owned Singapore subsidiary of Nissho Japan, namely, Nissho Iwai International (Singapore) Ltd (“Nissho Singapore”). Following the abovementioned merger in April 2004, Nissho Singapore was renamed Sojitz Asia Pte Ltd (“Sojitz Singapore”). ⁴ For ease of reference, Nissho Japan and Sojitz Japan (of which Nissho Japan became a part of after the merger) will be referred to collectively as the “Japan Company” if the temporal dimension is not important; and Nissho Singapore and Sojitz Singapore (the renamed entity after the merger) will be referred to as the “Singapore Company”. At all times, the appellants were agents of the Singapore Company and Japan Company.

14 The Japan Company is a trading company dealing in various commodities. ⁵ Two of the products traded overseas by the Japan Company through its overseas subsidiary, ie, the Singapore Company, are edible and industrial wheat flour, the manufacturer and supplier of which is Nippon Flour Mills Co Ltd. (“Nippon Flour Mills”). ⁶ Nippon Flour Mills would appoint its distribution agent for edible and industrial flour through the Singapore Company. ⁷

15 Koh was the sole proprietor of Chia Lee & Co (“Chia Lee”), a longstanding distributor of edible flour for the Singapore Company. From 1978 to 2002, Chia Lee was the sole distributor of only edible flour for Nippon Flour Mills (through Nissho Japan) in Singapore. ⁸

16 Both appellants held senior roles in the Japan Company and Singapore Company. Masui started work for Nissho Japan in 1987 and progressed up the ranks. In April 2002, he was seconded to Singapore and was subsequently entrusted with the role of General Manager of Nissho Singapore’s foodstuffs department in January 2004. He left the Singapore office in February 2005. After his return to Japan, he was promoted to General Manager of Sojitz Japan’s foodstuffs department from April 2005 to September 2007. ⁹ There, he was responsible for the flour business in Japan and oversaw the flour business globally. ¹⁰ Ishibe joined Nissho Japan in 1989. In October 2004, he was promoted to Manager of Sojitz Japan’s foodstuffs department. ¹¹ Ishibe was responsible for, inter alia, the sale of flour to the Singapore Company from the Japan Company ¹² and thus signed off on various flour packing lists. ¹³

17 In the course of their employment, the appellants were in charge of setting the selling price of the edible flour, informing Koh (who was trading as Chia Lee) of the market price and negotiating with him in relation to the edible flour business. ¹⁴

18 Prior to 2002, the industrial flour distributor for Nippon Flour Mills was a company called Sin Heng Chan. ¹⁵ When Sin Heng Chan faced severe financial difficulties, Nissho Singapore searched for an alternative industrial flour distributor. ¹⁶

19 Sometime in 2002, the appellants approached Koh and asked him for a “favour”. Specifically, they wanted Koh (and by extension, Chia Lee) to enter the industrial flour business to sell industrial flour.

20 The appellants devised a scheme called the “profit-sharing arrangement” which pertained solely to industrial flour. At that time, the expected profits from the industrial flour business was US$23 per metric ton of industrial flour. Koh would receive US$3 per metric ton while the remaining US$20 would be passed to Masui who would then split it equally with Ishibe. ¹⁷ The appellants agreed with each other that they would split the received moneys equally. ¹⁸ The appellants claimed that their role was to find customers for industrial flour and negotiate with them. Meanwhile, Koh would handle the administrative paperwork, ¹⁹ such as presenting bills of lading to the customers and collecting payments from them. ²⁰

21 Koh agreed and Chia Lee was subsequently appointed to replace Sin Heng Chan as the industrial flour distributor. ²¹ This was in spite of the fact that in 2002, Koh had no expertise in the industrial flour business, which operated in a markedly different fashion from the edible flour business. ²²

22 The profit-sharing arrangement began in 2002 and lasted till 2007. The last payment from Koh to the appellants was made on 26 November 2007. ²³ Pursuant to this scheme, the appellants accepted numerous payments from Koh. ²⁴ From February 2004 to 26 November 2007, there were 28 distinct payments which formed the basis of Masui and Ishibe’s 28 charges under s 6(a) read with s 29(a) of the PCA. The total quantum of gratification received was reduced from S$2,051,402 to S$2,009,433 (after I amended the amount of gratification received in respect of the two charges referred to earlier at [9]).

23 The profit-sharing arrangement was a corrupt scheme devised by the appellants to extract bribes (ie, gratifications) from Koh in return for them continuing to “support and protect” Chia Lee’s edible flour business. ²⁵

24 Although Koh remained in the profit-sharing arrangement for close to six years, this was not by choice. The appellants knew that Koh cherished Chia Lee’s role as the sole distributor of edible flour for Nippon Flour Mills in Singapore. Koh depended on it for his livelihood, but more than that, it represented the sum of his life’s work. In Koh’s words: ²⁶

25 The appellants used Chia Lee’s edible flour sole distributorship as both carrot and stick to ensure Koh’s cooperation in the profit-sharing arrangement. In essence, as long as Koh remained in the profit-sharing arrangement, the appellants would safeguard Chia Lee’s position as the sole distributor of edible flour in Singapore. ²⁷ While the appellants were careful to mask their intention by calling it a “favour”, Koh knew that if he did not comply with their demands, the appellants might introduce new competitors who would sell edible flour in the Singapore market, hence threatening Chia Lee’s dominant market position and negatively affecting its business. ²⁸ When Koh sought to withdraw from the arrangement, the appellants explicitly told him that if he did not wish to continue, “[they] may change other people to do that”, and that they “will not continue to support and protect [Koh] anymore” in respect of the edible flour business. ²⁹ These were threats that Koh took very seriously.

26 I list a few salient features of the “profit-sharing arrangement”.

(a) At trial, Koh’s consistent evidence was that the profit-sharing arrangement was not profitable for him. Even in 2002, his share of the profits (ie, at a fixed rate of US$3 per metric ton) barely covered the costs of doing the industrial flour business. ³⁰

(b) Over the years, the profits from the industrial flour business increased from US$23 per metric ton to US$40, US$50 and even US$60 per metric ton. ³¹ Ironically, even though Chia Lee was the company handling the business, Koh’s share of the profits remained constant (ie, at US$3 per metric ton). The benefit of any increases in profits accrued solely to the appellants. Worse still, Koh estimated that as the industrial flour business flourished, the tax liability on the industrial flour business which fell directly on him and Chia Lee, would increase correspondingly as well. ³²

(c) Koh was unable to extricate himself from the scheme, ³³ even when it became harmful to Chia Lee, and by extension, Koh who depended on it for his livelihood. ³⁴ When he voiced his concerns about the scheme, the appellants threatened him with the withdrawal of their support and protection for the edible flour business. ³⁵ As a result, Koh felt that he had no choice but to continue his payments to the appellants even when Chia Lee faced grave financial difficulties in 2005. ³⁶

27 By June 2005, it was clear to the appellants that Chia Lee was in parlous financial straits. ³⁷ On 15 June 2005, the appellants transferred US$240,000 to Chia Lee. ³⁸ The payment was meant to keep the struggling Chia Lee afloat so that their corrupt scheme could continue. ³⁹

28 Nonetheless, Koh continued to make payments to the appellants up till 26 November 2007. ⁴⁰ After November 2007, the global financial crisis negatively hit the industrial flour industry. ⁴¹ Koh had no more money to pay the appellants and the profit-sharing arrangement thus came to an end. ⁴²

29 The profit-sharing arrangement was discovered by Sojitz Japan around end 2009. ⁴³ Subsequently, on 26 February 2010, Sojitz Japan terminated the appellants’ employment. ⁴⁴

30 Chia Lee continued being the sole distributor of edible flour for Nippon Flour Mills until May 2015 when it ceased operations. ⁴⁵

31 In this section, I will explain my decision to amend C21 and C25 and deal with Ishibe’s new argument on appeal that he had only received US$50,000 from the profit-sharing arrangement.

32 I begin by laying out how the bribes were quantified.

33 The DJ found that the appellants had received payments totalling S$2,051,402 from Koh. This finding was based on photocopies of Koh’s contemporaneous handwritten notes which listed various amounts earned from the distribution of industrial flour and the sums paid to the appellants (Decision at [19]–[25]).

34 On appeal, the appellants dispute the exact sums received from Koh on the 28 occasions that gave rise to their respective 28 charges, arguing that the documentary records relied upon by the Prosecution to prove these sums were incomplete and unreliable. ⁴⁶

35 I should explain that the Prosecution framed the 28 charges against each appellant with reference to several photocopied documents and Koh’s evidence. ⁴⁷ The sequence of events as narrated by Koh is as follows: ⁴⁸

(a) Koh would make contemporaneous handwritten records of the amounts earned from the distribution of the industrial flour (“Handwritten Notes”). These notes included his calculations of the amounts owed to the appellants and the relevant USD/SGD exchange rates at the various points in time.

(b) Koh would then photocopy the original Handwritten Notes using the “photocopy function” on his facsimile machine which then printed them out on thermal paper (ie, exhibit P24). I refer to these as the “Thermal Paper Records”.

(c) Koh would then hand over the amounts owed to Masui along with the original copy of his Handwritten Notes.

(d) Subsequently further copies were made of these Thermal Paper Records (ie, exhibits P21, P22, P23 and P26). I refer to these as the “Photocopied Records”.

36 Counsel for Masui, Mr Nicolas Tang, embarked upon an extensive and meticulous analysis of the Photocopied Records and argued that they do not prove the amounts of gratification received by the appellants beyond a reasonable doubt. His most convincing reasons were as follows:

(a) The Photocopied Records were not accurate reproductions of the Thermal Paper Records. Pages were missing from the Photocopied Records, which also included additional markings. ⁴⁹

(b) The Photocopied Records could not be relied upon to prove that the appellants received an aggregate sum of S$2,051,402. Koh testified that he would refer to the invoices from the Singapore Company to Chia Lee to determine the buying price of industrial flour, and to the invoices issued by Chia Lee to the buyer to determine the selling price. While some of the Photocopied Records could be verified by referring to those invoices, not all the relevant invoices had been adduced during the trial. In respect of the Photocopied Records in exhibit P26, there were no invoices nor any calculations for the industrial flour transactions from 25 March 2006 to 26 November 2007. ⁵⁰

(c) The invoices from Chia Lee and the Singapore Company were unreliable. Koh testified that the invoices from the Singapore Company (ie, to sell Chia Lee the industrial flour) would always be generated earlier than the invoices issued by Chia Lee (ie, to the buyers who purchased the shipment of industrial flour from Chia Lee). ⁵¹ Koh agreed that the dates on the original handwritten notes would be the dates on which he received payment from Chia Lee’s buyers for the industrial flour. However, the Photocopied Records in P21, P22 and P23 sometimes predated the dates reflected on those invoices. ⁵²

(d) Four of the Photocopied Records were visually unclear. ⁵³

37 To corroborate the sums stated in the Photocopied Records, the Prosecution adduced OCBC bank deposit slips to show that Koh deposited sums into Masui’s OCBC bank account, and bank account statements from Chia Lee’s UOB account to show that Koh withdrew sums of cash from Chia Lee’s UOB account to pay Masui. In respect of this, Mr Tang pointed out that the OCBC deposit slips only covered two payments from Koh to Masui on 7 April 2006 and 26 November 2007. ⁵⁴

38 Mr Tang also argued that there were gaps in Koh’s memory as he was also unable to concretely identify which withdrawals from Chia Lee’s UOB bank accounts were for the purpose of paying Masui, as reflected in the Photocopied Records. Koh admitted that he would sometimes withdraw amounts in excess of the sums that had to be paid to Masui for use as petty cash and to pay his workers. He thus could not fully remember the purposes for each withdrawal made. ⁵⁵ It was also strange that Koh testified that the payments which were the subject of C12 to C15 were “collected by Mr Masui” in cash when Masui was not in Singapore during the relevant period. ⁵⁶

39 In the light of the arguments raised by Mr Tang, I directed the Prosecution to prepare a table of evidence (“Table of Evidence”) and allowed parties to file supplementary submissions in respect of this. Subsequently, the Prosecution filed two further table of profits (“the Tables of Profits”).

(a) The Table of Evidence summarised the testimonies of Koh, Masui and Ishibe and listed the invoices and OCBC deposit slips which were available to corroborate the stated quantum of gratification in each charge, along with the relevant references to the Records of Appeal.

(b) The first Table of Profits concerned C1 to C18 and provided a breakdown per charge of the amount of gratification that could be corroborated by reference to Chia Lee’s invoices, the Singapore Company’s invoices, both types of invoices and any further additions or deductions that ought to be made to the various sums. This table also included an additional column listing sums for which there was no corroborating documentary evidence.

(c) The second Table of Profits concerned C19 to C28 and listed the amounts stated, the amounts withdrawn from Chia Lee’s bank account or deposited into Masui’s bank account as reflected in the bank statements and Koh’s explanation in respect of each charge.

40 On 24 February 2020, after reviewing the parties’ submissions and the available evidence in substantial detail, I found that the Prosecution had proved the amounts of gratification received by the appellants from Koh beyond a reasonable doubt, except for C21 and C25 which were respectively amended to reflect a lower amount of gratification received. ⁵⁷ The appellants did not object to the amendment of C21 and C25.

41 I agreed with the DJ that Koh gave a cogent and credible account of how the Handwritten Notes and the Photocopied Records came to be created, which was corroborated by objective evidence. ⁵⁸ Koh was able to explain the process of creating each Handwritten Note (see above at [35]). ⁵⁹ Crucially, Koh was also able to explain the contents within each Handwritten Note as reflected in the Photocopied Records. The Photocopied Records can be split into two groups – those with calculations in P21, P22 and P23 (which reflected the bribe quanta in C1 to C18) and those without calculations in P26 (which reflected the bribe quanta in C19 to C28).

42 First, Koh was able to explain the calculations within each of the Photocopied Records in P21, P22 and P23 in extensive detail. I use the first Photocopied Record (ie, the photocopy of the first Handwritten Note from which the bribe quantum in C1 was derived) as an illustration of this. There were four groups of transactions totaling S$71,773 (ie, the bribe quantum stated in C1). Koh explained that: ⁶⁰

43 The sum of S$71,773 was the bribe quantum reflected in C1. As could be seen from the bolded words, the calculation of this sum was in accordance with the profit-sharing arrangement. The same process was repeated for each of the other 17 charges. ⁶¹ During cross-examination, Masui agreed that the method of calculation in each Photocopied Record correctly reflected the profit-sharing arrangement between Koh and the appellants. ⁶²

44 For C1 to C18, Koh’s testimony and the Photocopied Records could be corroborated partially by a mix of invoices from Chia Lee and the Singapore Company. Using the Table of Evidence, the Prosecution pointed out that a total of 306 transactions made up the 18 payments to the appellants (ie, the first 18 charges). Of these 306 transactions, 83% of them were corroborated by at least one invoice (ie, 255 transactions) and 51% of them were corroborated by both Chia Lee and Singapore Company’s invoices (ie, 155 transactions). ⁶³ I agreed with the Prosecution that taken as a whole, there was a high degree of corroboration.

45 Second, in respect of the Photocopied Record in P26 (ie, corresponding to C19 to C28), Koh provided a cogent explanation for why there were no calculations but simply a list of sums passed to the Appellants: in June 2005, Chia Lee was facing grave financial difficulties and Koh simply paid the appellants whatever he could afford. ⁶⁴ As recognised by Mr Tang (see above at [37]), two of the sums listed in P26 were corroborated by the OCBC bank deposit slips which showed deposits into Masui’s bank account on 7 April 2006 and 26 November 2007 (exhibit P27). These two deposits of S$13,750 and S$82,900 were the same amounts reflected in the C19 and C28.

46 In respect of the Defence’s argument that there were no bank records from Chia Lee prior to 2005, Koh explained that the bank statements might have been thrown away or lost when he shifted his warehouse. ⁶⁵ For C11 to C28 (excluding C19), Chia Lee’s bank records were available and Koh was able to identify various withdrawals which corresponded to some of the payments made to Masui. ⁶⁶

47 While I noted the Defence’s argument that Koh was at times confused and unable to recall details about specific withdrawals in Chia Lee’s bank records, this was unsurprising given the voluminous number of transactions which made up 28 different payments to the appellants and the fact that more than 10 years had passed since the last payment and the first day of trial. Furthermore, as the Prosecution rightly pointed out, Koh did not rely solely on his memory when he identified various withdrawals in the bank statements, but also referred to the exchange rates stated in the bank statements which tallied with what he had recorded in the Handwritten Notes. ⁶⁷

48 Ishibe’s counsel, Mr Sunil Sudheesan, submitted that the charges were not made out because the Prosecution did not adduce a complete set of documents. A complete set would mean: (a) signed invoices from Chia Lee; (b) invoices from the Japan Company; (c) bank records showing the withdrawal of moneys by Koh corresponding to each charge in terms of date and amount; and (d) Photocopied Records or Handwritten Notes with legible writing inclusive of dates that corresponded to the withdrawals and/or invoices. ⁶⁸

49 I disagreed. The Prosecution must prove the quantum of gratification in each of the 28 charges beyond a reasonable doubt, but this did not mean that the Prosecution had to produce a perfect set of evidence to prove it beyond all doubt. It must be appreciated that the invoices, bank statements and deposit slips served a corroborative purpose. The key evidence relied upon by the Prosecution was the Photocopied Records and Koh’s explanation in respect of them (see above at [35]). ⁶⁹

50 It will be obvious from the foregoing paragraphs that the mix of objective records adduced by the Prosecution was capable of corroborating almost all the values stated in the Photocopied Records such that they collectively strengthened their reliability. ⁷⁰

51 Furthermore, as observed by the DJ, Koh was an honest witness, “readily admitting when he was unable to recall, rather than fudge.” (Decision at [23]). This gave his evidence (which was already corroborated) a ring of truth and reinforced his credibility. In contrast, Masui, the direct recipient of those moneys, was not a credible witness (Decision at [24]–[25]). He feigned ignorance about the amounts received from Koh and claimed that he did not keep records of the payments. He also asserted that he had never seen or received the Handwritten Notes from Koh. When questioned on how he knew that Koh would not short-change the appellants, Masui blithely claimed that he did not check. This seemed illogical in the light of (a) Masui’s concession that it was important to the appellants that Koh was keeping to his part of the agreement, and (b) the appellants’ joint defence at trial that they had taken joint responsibility for the risks of the industrial flour business and would use the payments to them to cover these risks, if needed. ⁷¹

52 On the whole, I accepted that the Prosecution had proved the values of the bribe quanta in the individual charges (except C21 and C25) beyond a reasonable doubt.

53 C21 stated that Masui received S$102,115 from Koh on or about 12 July 2006. ⁷² This sum was reflected as US$65,000 in the Photocopied Record in exhibit P26 (USD/SGD exchange rate of 1/1.571 per the Photocopied Record). ⁷³ The only corroborative evidence in respect of this was Chia Lee’s bank account statements from which Koh identified four withdrawals totalling S$101,734. ⁷⁴ Koh also stated that he had approximately S$400 on hand which he used to make up the sum of S$102,115. ⁷⁵ However, Chia Lee’s bank account statement stated that one of the four withdrawals was made on 17 July 2006, five days after the purported payment in C21. ⁷⁶ This withdrawal was of US$10,000 (ie, S$15,840). ⁷⁷

54 C25 stated that Koh passed S$137,340 to Masui on or about 15 February 2007. ⁷⁸ This sum was reflected as US$90,000 in the Photocopied Record in exhibit P26 (USD/SGD exchange rate of 1/1.576 as stated in the Photocopied Record). ⁷⁹ However, according to the Photocopied Record, the US$90,000 was transferred on 8 February 2007. ⁸⁰ Again, the only corroborative evidence was Chia Lee’s bank account statements from which Koh identified four withdrawals totalling $145,237 and stated that he only paid $137,340 and had either kept the remainder or deposited it in the SGD account for other uses. ⁸¹ Crucially, the bank statement shows that one of the four withdrawals was made on 13 February 2007, five days after the purported payment in the Photocopied Record. This withdrawal was of US$17,000 (ie, $26,129). ⁸²

55 Given that the two abovementioned withdrawals occurred after the date reflected in either the charge or the documentary evidence, I was unable to accept that they should be factored into the gratification sums in the charges. I thus deducted the two withdrawals from the respective gratification quantum in C21 and C25. This gave sums of $86,275 and $111,211 respectively.

56 The total gratification quantum was thus reduced from $2,051,402 to $2,009,433. A breakdown of the total gratification quantum is set out at [303] below.

57 During the trial, it was common ground between the appellants that Koh always passed the money directly to Masui who would then split it equally with Ishibe.

58 On appeal, Ishibe advanced a new position. He claimed that while there was an agreement to split the moneys equally, he had no knowledge of the true sum passed to Masui and had in fact only received US$50,000. ⁸³ As proof of this, Ishibe pointed to his long statements given to officers from the Corrupt Practices Investigation Bureau (“CPIB”) on 12 March 2012 and 29 July 2013, and Masui’s statement to CPIB officers on 12 March 2013 that Masui passed “$50,000 to Ishibe”. ⁸⁴ Ishibe added that there was no objective evidence of the exact sum that Ishibe had received from Masui. ⁸⁵

59 I did not accept this argument. During the trial below, the appellants had run their defences on the basis that they had split the moneys equally, as per their agreement. Interestingly, when the sum of US$50,000 was brought up by the Prosecution, this was in the context of exploring inconsistencies between Ishibe’s testimony in court and the version of events given in his CPIB statements. Ishibe’s response was telling and worth reproducing in full: ⁸⁶

60 A few minutes later, Ishibe said that the sum of US$50,000 was a net figure that he was left with after covering the losses from the profit-sharing arrangement, but added that “[i]f you ask me that this amount is [a] hundred per cent correct, I am not confident”. ⁸⁷

61 Furthermore, it was stated in their joint mitigation plea that “[i]t is also undisputed that Ishibe and Masui shared whatever money received from Koh equally with each other”. ⁸⁸

62 In Mohd Suief bin Ismail v Public Prosecutor [2016] 2 SLR 893, the Court of Appeal clarified that an accused person is, strictly speaking, not precluded from relying upon a defence that is raised for the first time on appeal. However, the appellate court will have regard only to the evidence which had been led at the trial itself to ascertain whether that defence was reasonably available on the evidence before the court at the trial (at [25]). Accordingly, the “defence” that Ishibe only received US$50,000 was not available to Ishibe on appeal as it was unsupported (and in fact, contradicted) by his own testimony and position at trial.

63 On the face of the evidence, I found no reason to disagree with the finding of the DJ that the appellants split the bribe moneys equally. I will return to this point later for my decision on the appropriate penalty order.

64 The DJ found it appropriate to adopt a sentencing framework for corruption offences under ss 5 and 6 of the PCA (Decision at [81]) and adopted a sentencing band approach modelled after the Court of Appeal’s decision in Ng Kean Meng Terence v Public Prosecutor [2017] 2 SLR 449 (“Terence Ng”) (Decision at [85]). This comprised four broad steps (Decision at [86] – [111]):

(a) Step 1: Identify the significant offence-specific factors, such as the triggering of the public service rationale, premeditation, and abuse of trust and confidence.

(b) Step 2: Classify the offence into one of the four sentencing bands based on the number of significant offence-specific factors present, and their severity, to derive the indicative starting point sentence within the relevant sentencing band. The sentencing bands set out by the DJ were as follows (Decision at [102]):

(c) Step 3: Adjust the indicative starting point sentence to account for offender-specific factors, such as a plea of guilt or criminal antecedents.

(d) Step 4: Adjust the sentence to take into account the totality principle where an offender faced multiple charges, to ensure that the global sentence was not crushing.

65 The DJ referred to the decision of Menon CJ in Public Prosecutor v Syed Mostofa Romel [2015] 3 SLR 1166 (“Romel”), which dealt with private sector corruption. By way of background, Menon CJ identified, for the purposes of sentencing, three categories of cases concerning three different ways by which private sector corruption could manifest (at [26]). In the first category, the receiving party confers a benefit on the paying party, without regard to whether the paying party ought properly to have received the benefit. In the second category, the receiving party forbears from doing what he was duty bound to do, and thus confers a benefit on the paying party. In the third category, the receiving party forbears from inflicting harm on the paying party, even though there may be no lawful basis for such harm to be inflicted. This analytical tool devised by Menon CJ is commonly known, and shall be referred to in this judgment, as the “Romel categories” (for further elaboration, see [A.10] below). Consequently, when I am referring to the Romel categories, I will use the terminology “Romel category 1/2/3”, as the case may be. The DJ found that the appellant’s conduct fell within Romel category 3 as their corrupt scheme interfered with, and deprived Koh of, his legitimate rights. The DJ identified five offence-specific factors (Decision at [112]–[113]): (a) high quantum of gratification; (b) sustained period of offending; (c) heightened culpability of Masui and Ishibe as the masterminds behind the corrupt transactions; (d) significant abuse of trust and authority; and (e) premeditation.

66 The DJ imposed the following starting point sentences (Decision at [121]):

(a) 12 months’ imprisonment for charges where the amount of gratification was less than $30,000;

(b) at least 12 months’ imprisonment for charges where the amount of gratification ranged from $30,000 to $50,000;

(c) at least 15 months’ imprisonment for charges where the amount of gratification was more than $50,000 to $100,000;

(d) at least 18 months’ imprisonment for charges where the amount of gratification was more than $100,000.

67 There were no offender-specific factors which warranted a downward adjustment of the individual starting point sentences (Decision at [122]). After considering the totality principle, the DJ ordered the sentences for four charges C2, C6, C25 and C27 to run consecutively for both appellants, resulting in a sentence of 66 months’ imprisonment (Decision at [127]–[128]). As the total gratification involved was $2,051,402, he also ordered Masui and Ishibe to each pay a penalty of $1,025,701 (in-default six months’ imprisonment) under s 13 of the PCA.

68 The Prosecution puts forth a five-step sentencing framework modelled after the two-stage, five-step framework adopted in the case of Logachev Vladislav v Public Prosecutor [2018] 4 SLR 609 (“Logachev”) (as developed in Wong Chee Meng ([10] supra)) which applies to cases under ss 6(a) and (b) of the PCA (“Logachev framework”). Applying its proposed framework, it submits that the sentences and penalties imposed on the appellants are eminently justified on the facts. ⁸⁹

69 Masui argues against the imposition of a sentencing framework for corruption offences under s 5 or s 6 of the PCA. ⁹⁰ He submits that his sentence is manifestly excessive as the DJ had misapplied the sentencing band framework by double counting the offence-specific factors and failing to account for offender-specific factors. ⁹¹ An appropriate sentence would be one that is less than the 98-week imprisonment term imposed on the offender in Public Prosecutor v Leng Kah Poh [2014] 4 SLR 1264 (“Leng Kah Poh”) who had solicited and received bribes worth $2,341,508. ⁹² Masui avers that the penalty imposed by the DJ is erroneous and ought not to exceed $500,525.23 as certain sums ought to be deducted (see [323] below). ⁹³

70 Ishibe agrees with Masui. Having regard to the precedents, Ishibe argues that a fair sentence would be 12 months’ imprisonment (approximately 52 weeks). ⁹⁴ In respect of the penalty, Ishibe was originally aligned with Masui in seeking the deduction of various sums from the penalty. ⁹⁵ However, he subsequently argues that the penalty imposed on him should not exceed US$50,000 given that Masui had only passed him US$50,000. ⁹⁶

71 Interestingly, although Ishibe had initially proposed the adoption of a sentencing framework based on the Logachev framework which, save for the way in which the indicative starting sentences were calibrated, appeared to be identical to the Prosecution’s current framework, ⁹⁷ Ishibe’s counsel indicated orally during the final hearing on 21 August 2020 that Ishibe no longer takes the position that a sentencing framework is necessary at all for corruption offences. I note however that Ishibe’s counsel had previously undertaken a graphical analysis showing how sentences in corruption cases varied according to the value of the bribes in order to illustrate that the sentences imposed by the DJ are manifestly excessive when compared to similar cases.

72 Bearing the above in mind, the following issues fall for my determination in the appeals against sentence:

(a) First, is it appropriate to develop a sentencing framework for corruption offences under the PCA?

(b) Second, assuming that the first question is answered in the affirmative, what type of framework should be employed and why?

(c) Third, what should the appropriate framework be?

(d) Fourth, applying the framework, what is the appropriate sentence for each appellant?

(e) Fifth, what is the appropriate penalty under s 13 of the PCA?

(f) Sixth, whether this is a case suitable for prospective overruling?

73 Before turning to the specific issues raised, it is useful to make some general observations about corruption offences under ss 6(a) and 6(b) of the PCA which penalise corrupt transactions with agents.

74 The relevant legal provisions are as follows:

75 Section 6 of the PCA dates back to the original enactment of the PCA, ie, the old Prevention of Corruption Ordinance 1960 (No 39 of 1960). It predates the wider s 5 of the PCA which criminalises corruption generally and was introduced in 1960 as a means of “providing wider powers to combat corruption” (Song Meng Choon Andrew v Public Prosecutor [2015] 4 SLR 1090 (“Andrew Song”) at [34]). Sections 6(a) and 6(b) of the PCA have remained largely unchanged since 1960.

76 There is currently no case in Singapore that sets out a general sentencing framework for the simpliciter offences under ss 6(a) and 6(b) of the PCA. I should clarify that I use the term “sentencing framework” in the narrow sense, ie, a framework that explains how the legislatively mandated sentencing spectrum of imprisonment for a term not exceeding 5 years and/or a fine not exceeding $100,000 ought to be spread out across the myriad fact scenarios that come within the ambit of ss 6(a) and 6(b) of the PCA.

77 At present, our sentencing courts have been guided by the landmark decision of Romel ([65] supra) which laid down the Romel categories. This useful analytical tool has helped our sentencing courts to broadly assess the type and seriousness of corruption disclosed in a case of private sector corruption (see [65] above).

78 In the absence of a general sentencing framework, our courts have typically approached sentencing in corruption cases under both ss 5 and 6 of the PCA by having regard to past cases which have identified a number of categories and factors pertinent to the sentencing process (see Wong Chee Meng ([10] supra) at [50] and Michael Tan ([10] supra) at [105]).

79 From the case law on corruption offences generally (ie, under ss 5, 6 and 7 of the PCA), it can be observed that there are four broad categories of corruption:

(a) Category 1: Corruption in the public sector which involves government servants or officers of public bodies. A custodial sentence is the norm for such cases in the light of the strong public interest in stamping out corruption in the public sector (see Romel at [15], Wong Chee Meng at [60]).

(b) Category 2: Corruption in the private sector which engages the public service rationale. For clarity, this refers to the “public interest in preventing a loss of confidence in Singapore’s public administration”. This sentencing principle is presumed to apply in cases of public sector corruption but has been extended to cases where private agents handle public money, supply public services or are involved in government contracts. This category also includes private sector offences that concern regulatory or oversight roles such a marine surveying (see Ang Seng Thor [2011] 4 SLR 217 (“Ang Seng Thor”) at [33(c)] citing Lim Teck Chye [2004] 2 SLR(R) 525 (“Lim Teck Chye”) at [66]–[68], and Romel at [24]). In such cases, a custodial sentence is often the norm (Ang Seng Thor at [33(d)] and Romel at [24]).

(c) Category 3: Corruption in the private sector which does not engage the public service rationale, ie, private sector agents performing purely commercial functions. I refer to this category as “purely private sector corruption”. While there is no norm in favour of non-custodial sentences in private sector corruption cases, the general trend indicates that where private sector agents performing purely commercial functions are concerned, offences which register a lower level of overall criminal culpability may be dealt with through the imposition of fines. However, our courts have repeatedly stressed that whether or not the custody threshold is breached will depend greatly on the “specific nature of corruption” [emphasis in original] that presents itself on the facts (see Romel at [20], Michael Tan at [64]). Examples of cases falling within this category include corruption that results in the loss of confidence in strategic industries such as bunkering or maritime industry (see Ang Seng Thor at [34], and Heng Tze Yong v Public Prosecutor [2017] 5 SLR 576 (“Heng Tze Yong”) at [39]), the corruption of foreign public officials (see Michael Tan generally and at [72], [108]).

(d) Category 4: Corruption cases for which there are established sentencing guidelines. This is an open category that has been included to accommodate any present and future judgments that provide sentencing guidelines tailored to a specific type of fact scenario. At present, the only types of cases falling within this category are: (a) those relating to sports-betting and match-fixing (see Ding Si Yang v Public Prosecutor and another appeal [2015] 2 SLR 229 (“Ding Si Yang”); and (b) cases involving offenders prosecuted under s 6 read with s 7 of the PCA (see Wong Chee Meng).

For the avoidance of doubt, these categories are distinct from the “Romel categories” (see [65] above).

80 Focussing on Category 4, it is pertinent to note that while our courts do typically have regard to precedents when sentencing corrupt offenders, there are in fact only two existing sentencing frameworks for corruption offences.

81 In the recent case of Wong Chee Meng, Menon CJ laid down a sentencing framework which applies solely to aggravated offences under s 6 read with s 7 of the PCA (at [61]). This framework is modelled after the two-stage, five-step Logachev framework. In his judgment, Menon CJ took pains to caution that this framework cannot be “adapted for use with the basic offence under s 6 simply by making … a downward adjustment to the indicative sentencing ranges to account for the lower sentencing range prescribed by the statute”. This is because the public service rationale will be implicated in virtually all cases falling under the aggravated offence under ss 6 and 7 of the PCA, while the same cannot be said for the simpliciter offence under s 6 of the PCA (at [60]).

82 At first glance, the framework in Wong Chee Meng ([10] supra) appears to be a general framework as it applies to all offences under s 6 read with s 7 of the PCA. This may include offences under both Categories 1 and 2, if the offender is convicted under ss 5 or 6 read with s 7. It is also included in Category 4 because: (a) s 7 of the PCA is a punishment enhancement provision that increases the maximum imprisonment term under ss 5 or 6 of the PCA from 5 to 7 years; and (b) it only applies in the specific situation where the offence under ss 5 or 6 of the PCA takes place in relation to contracts with the Government or other public bodies (at [48]). The Wong Chee Meng framework is thus a narrow one when one considers the wide breadth of fact scenarios in which corruption offences may be committed.

83 In Ding Si Yang ([79(d)] supra), I formulated a narrow sentencing framework which applies only in the specific scenario of football match-fixers convicted under s 5 of the PCA.

84 It is not difficult to understand why our courts have, thus far, declined to devise a general sentencing framework for all simpliciter corruption offences under ss 5 and 6 of the PCA.

85 First, while there is a degree of overlap between ss 5 and 6 of the PCA, they are distinct offence creating provisions. Section 5 punishes corruption generally and is of a much wider ambit than s 6, the latter of which focuses on punishing agents who have allowed their loyalty to their principal to become suborned through the corrupt receipt of gratification. Different sentencing considerations may thus be relevant depending on which section the offender is charged under (see Wong Chee Meng at [59], citing Andrew Song ([75] supra) at [31]).

86 Second, focussing more narrowly on ss 6(a) and 6(b) of the PCA, these two provisions will still cover a wide variety of factual scenarios. While “[t]he fact that corruption occurs in a wide variety of circumstances does not, in and of itself, preclude the adoption of a sentencing framework”, it cannot be denied that “[t]he wide variety of acts caught by ss 5 and 6 of the PCA would make [the] crafting of a single sentencing framework applicable to all such offences an extremely challenging task” [emphasis in original] (Wong Chee Meng ([10] supra) at [56] and Michael Tan ([10] supra) at [104]).

87 I end this section with a brief observation that the Prosecution had originally proposed the adoption of a general sentencing framework under ss 5 and 6 of the PCA. In the light of Wong Chee Meng, it no longer maintains this position and instead argues in favour of a general sentencing framework under ss 6(a) and 6(b) of the PCA. It is to this question that I now turn.

88 In my judgment, the time has come for this court to lay down a sentencing framework for offences under ss 6(a) and 6(b) of the PCA. This framework will be limited to cases of purely private corruption, ie, Category 3, as defined above at [79(c)].

89 In coming to my decision, I have had regard to a long list of approximately 50 precedents involving around 160 individual charges compiled by the Prosecution and supplemented by the Defence. This list represents the majority of the available written decisions on corruption under ss 5 and 6 of the PCA in Singapore.

90 First, given the broad variety of ways in which corruption may manifest itself, a sentencing framework will provide guidance for sentencing courts as to the appropriate sentence in novel or unusual fact scenarios where there are no analogous precedents. Having reviewed the abovementioned list of precedents, I am of the view that the present case is indeed one where there a dearth of analogous precedents. The only other cases involving similarly high aggregate bribe quantum are Leng Kah Poh ([69] supra) and Public Prosecutor v Andrew Tee Fook Boon [2011] SGHC 192. However, these cases can be distinguished on the facts given the egregious conduct of the appellants in abusing their position to extract bribes by threatening Koh. The present case falls within Romel category 3 for which there are few available precedents, and no precedents involving such a high aggregate bribe quantum. Given the lack of analogous precedents, I am thus persuaded that a sentencing framework will help me to determine if the sentences imposed by the DJ are indeed manifestly excessive, and if so, to derive appropriate sentences for the appellants.

91 Second, a sentencing framework will be beneficial for achieving broad consistency in sentencing for purely private sector corruption cases. The word “consistency” here is used in two senses: (a) consistency in methodology; and (b) consistency in sentencing outcome. The former requires sentencing courts to apply a methodology that is broadly consistent when faced with a particular type of case. This needs no further elucidation as it has been amply explained in Wong Chee Meng ([10] supra) at [56]–[57].

92 The latter requires that a sentencing court arrives at broadly the same outcome for the sentences imposed irrespective of the methodology applied given the same equivalent set of facts. Consistency in sentencing outcome requires that all things being equal, sentencing courts faced with two very similar cases should arrive at broadly similar outcomes, irrespective of the methodology used. This is ideal. But the reality is that different methodologies are likely to give rise to different sentencing outcomes, thus the need for a sentencing framework. If the same methodology is applied (as in a sentencing framework prescribed for a particular type of offences), it is more likely for the same sentencing outcome to be reached, provided always that the methodology (ie, the sentencing framework) is well-defined and does not itself lead to different sentencing outcomes when applied to the same set of facts. If it does, then the methodology or the sentencing framework itself needs refinement.

93 Focusing on the latter, I agree with the Prosecution that there are inconsistencies in the case law. For example, in Public Prosecutor v Ng Sing Yuen [2007] SGDC 203 (upheld in Magistrate’s Appeal No 37 of 2007), a sentence of eight months’ imprisonment was imposed for a charge involving $100,000 with aggravating factors like a breach of trust and persistent and sustained offending. Meanwhile in Leng Kah Poh, a charge involving $86,000 with similar aggravating factors only attracted 14 weeks’ imprisonment (or approximately 3.23 months’ imprisonment at 4.33 weeks/month).

94 I acknowledge that full consistency is not possible given that: (a) no two cases are identical on the facts; and (b) a measure of discretion is accorded to sentencing judges in arriving at an appropriate sentence. That said, I believe that a prescribed sentencing framework will help sentencing courts to achieve a broadly consistent sentencing outcome. First, it will help the court to understand where a particular offender falls within the spectrum of the severity of offending. Second, it will enable the court, prosecutors and defence counsel to weed out precedent cases with sentencing outcomes that are wildly inconsistent with the general trend of similar cases. Third, it will cause sentencing courts to apply the same broad methodology, barring exceptional circumstances.

95 It is not appropriate at this juncture to lay down a general sentencing framework that deals with all categories of corruption for all offences under ss 6(a) and 6(b) of the PCA.

96 First, the development of a framework that pertains only to purely private sector corruption is in line with the nuanced context-specific approach that our sentencing courts have taken over the course of the past 80 years, ie, in developing lines of case law applicable to the various broad contexts in which corruption occurs (see Michael Tan ([10] supra) at [108]).

97 In this regard, I disagree with the Prosecution’s submission that it is “unclear” if a context-specific framework will be in line with the legislative intention behind ss 6(a) and 6(b) of the PCA given that Parliament has legislated for a single offence-creating provision with a single sentencing spectrum. ⁹⁸ No authority is cited for this proposition, which ignores the fact that the same context-specific approach has been used by our courts for a variety of offences. Under s 338(b) of the Penal Code (Cap 224, 2008 Rev Ed) (“Penal Code”) which deals with negligent offenders who cause grievous hurt, the High Court in Tang Ling Lee v Public Prosecutor [2018] 4 SLR 813 laid down a sentencing framework that applies only to cases involving traffic accidents.

98 The utility of the context-specific approach lies precisely in its ability to recognise that certain aggravating (or mitigating) factors may take on exceptional weight in specified contexts. The context-specific approach is best applied in cases where the offence-creating provision is worded so broadly as to encompass a wide variety of factual situations.

99 Second, the weight that our courts have placed on the public service rationale may well mean that any sentencing framework will have to be applied differently based on the absence or presence of such factors (see Wong Chee Meng ([10] supra) at [60] in respect of the public service rationale).

100 In any event, the present appeals concern a case of purely private sector corruption. It is not necessary for me to formulate a general sentencing framework under ss 6(a) and 6(b) to come to a determination on the present appeals against sentence. As such, the framework that I set out below concerns only cases involving purely private sector corruption under ss 6(a) and 6(b) of the PCA. I leave open the question of a general sentencing framework for corruption offences for a future court.

101 In the course of these appeals, various models of sentencing have been put forth by the parties. They are alternately described as “sentencing guidelines” and “sentencing frameworks”.

102 For conceptual clarity, where the expression “sentencing guideline” is used in this judgment, it will refer to a court’s guidance as to the presumptive sentences that should be imposed for the commission of an offence in defined factual scenarios (see Terence Ng ([64] supra) at [25]).

103 A sentencing framework as defined above at [76] is a type of sentencing guideline that explains how the entire statutory range of sentences for a particular offence ought to be spread across the myriad fact scenarios that fall under the offence-creating provision. Sentencing frameworks, as a form of sentencing guideline, can be contrasted with another form of sentencing guideline called sentencing “benchmarks”, in which the court identifies an archetypal case (or a series of archetypal cases) and the sentence which should be imposed in such a case. An example of a benchmark sentencing guideline can be found in Wong Hoi Len v Public Prosecutor [2009] 1 SLR(R) 115. There, the High Court found that the benchmark sentence for an uncontested charge of assaulting a public transport worker under s 323 of the Penal Code is 4 weeks’ imprisonment (see Terence Ng at [31]–[32]).

104 The parties’ proposed models take the form of sentencing frameworks.

105 The Prosecution proposes a five-step sentencing framework modelled on the two-stage, five-step Logachev framework. This framework applies to determine the sentence (per charge) for an offender who claims trial to offence(s) under ss 6(a) and 6(b) of the PCA. As in Logachev ([68] supra), the framework can broadly be split into two stages.

106 At the first stage (“Stage 1”), the court has regard to the severity of the offence committed by having regard to all the offence-specific factors present on the facts of the case and arrives at an indicative sentence to reflect it. Offence-specific factors are those which relate to the manner and mode in which the offence was committed as well as the harm caused to the victim or to the wider society. The offence-specific factors can be further grouped into factors that go towards the harm caused by the offence and the offender’s culpability (see Terence Ng ([64] supra) at [39], Logachev at [34]–[36]). In this sense, the overall severity of the offence is determined by the interaction between factors that separately go towards (a) harm, and (b) culpability.

107 At the second stage (“Stage 2”), the court considers all the offender-specific factors to derive a sentence for each individual charge. These are aggravating and mitigating factors that are personal to the offender. They do not relate directly to the commission of the offence but rather, to the personal circumstances of the offender (see Terence Ng at [39], Logachev at [34]–[36]), eg, a plea of guilt. In that regard, the offence-specific factors and offender-specific factors are generally treated as being mutually exclusive in nature.

108 The five steps of the Prosecution’s sentencing framework are as follows: ⁹⁹

(a) Step 1: Identify the offence-specific factors that go towards harm and culpability. Culpability, as assessed by the manner, mode and extent of the offender’s involvement in the criminal act, is a measure of his relative blameworthiness. Harm is a measure of the injury caused to the society by the commission of the offence.

(b) Step 2: Identify the applicable indicative sentencing range having regard to the prescribed sentencing range in a sentencing matrix based on the two elements of “harm” and “culpability” (ie, the harm-culpability matrix). The available indicative sentencing ranges set out in the sentencing matrix fall within the spectrum of punishment prescribed by Parliament for ss 6(a) and 6(b) of the PCA.

(c) Step 3: Identify the appropriate indicative starting point for the sentence (“indicative sentence” or “indicative starting sentence” being used and referred to hereafter interchangeably) within the indicative sentencing range identified in Step 2, as determined by an examination of the offence-specific factors.

(d) Step 4: Make adjustments to the indicative starting sentence to account for offender-specific factors.

(e) Step 5: Make further adjustments to the individual sentences, if necessary, to take into account the totality principle.

109 Steps 1 to 3 fall under Stage 1, Step 4 falls under Stage 2 and Step 5 relates to the determination of the final sentence per charge after the court considers the number of sentences of imprisonment to be run consecutively or concurrently for all the charges and the overall imprisonment sentence to be imposed on the offender having regard to the totality principle.

110 There is no substantial dispute in relation to the relevance of the offence and offender-specific factors listed out by the Prosecution and I will not go through them here. Instead, I focus on the determination of the indicative starting sentence.

111 The Prosecution’s harm-culpability matrix is reproduced below (with a mere re-arrangement of the layout of the boxes to reflect increasing harm from the left to the right boxes; and increasing culpability from the bottom to the top boxes and with no change whatsoever to the content or indicative sentencing range specified in each of the boxes). The general manner of spreading out the sentences by the Prosecution is typical of many of such matrices found in precedent cases setting out similar harm-culpability matrices.

112 I previously mentioned that Ishibe had also initially proposed the adoption of a sentencing framework modelled on the Logachev framework. As Ishibe did not delve into the specifics of the structure of his framework in his submissions, I can only assume that Ishibe’s proposed harm-culpability matrix operates in a similar fashion to the Prosecution’s framework but only differs in respect of the sentencing range prescribed for each box in his proposed matrix. ¹⁰⁰

113 Ishibe’s harm-culpability matrix is reproduced as follows:

114 In addition to the five-step Logachev framework, Ishibe also undertook a graphical analysis to broadly show that the sentences imposed by the DJ are manifestly excessive when compared to other precedents involving high amounts of gratification. ¹⁰¹

115 Ishibe collated 13 corruption precedents under ss 5 and 6 of the PCA and plotted the total gratification received in those cases against the eventual imprisonment sentence passed on the offender. The graphs are designed to show the very broad correlation between the overall bribe quanta as a principal determinant of sentence (ie, the independent variable) and the indicative sentence (ie, the dependent variable). Ishibe was not trying to use his graphs to assist in determining an appropriate final sentence for himself based on the total bribe quantum involved in the present appeals.

116 I have not reproduced Ishibe’s graphs as they contain outdated cases (see, for example, the sentence of the offender in Public Prosecutor v Gursharan Kaur Sharon Rachel [2018] SGDC 217 which was recently enhanced on appeal in Michael Tan ([10] supra) at [161]–[162]), and cases involving public sector corruption. ¹⁰² Nonetheless, I bear in mind this approach as a possible sentencing framework for purely private sector corruption under ss 6(a) and 6(b) of the PCA.

117 The last sentencing framework is the sentencing band framework employed by the DJ. After Hoo J’s decision in Michael Tan, it is clear that this approach is unsuitable for corruption offences under the PCA (Michael Tan at [106]–[108]), and I say no more about it

118 There are thus two proposed sentencing frameworks for my consideration in these appeals, ie, the graphical analysis premised on bribe quantum and the harm-culpability matrix.

119 I begin with a general discussion on the structure, form and design of a sentencing guideline, focussing on the sentencing framework model and the principles that need to be adhered to in order to achieve as much consistency as possible in the sentencing outcomes when the same framework is being applied to the same fact situations.

120 In Singapore, the task of issuing sentencing guidelines falls on the judiciary, rather than an executive body specially constituted for this purpose. As was observed by the Court of Appeal in Terence Ng ([64] supra) at [26], there are many forms of sentencing guidelines such as the benchmark approach, the single starting points approach and the sentencing matrix approach.

121 Constructed well, sentencing guidelines are formidable tools that help future sentencing courts work towards collectively achieving key goals such as:

(a) Promoting consistency in sentencing while maintaining an appropriate level of flexibility and discretion for sentencing courts.

(b) Encouraging transparency in reasoning. This is especially when future courts apply a similar methodology which requires them to explain their reasoning processes at different stages of the sentencing analysis.

(c) Creating a coherent picture of sentencing for a particular offence. The sentencing guideline should respect the statutory context by taking into account the whole range of penalties prescribed by the relevant statute. It should also ensure rationality in the spread of sentences and avoid arbitrariness in the indicative starting sentences laid out.

(See Terence Ng at [23] and Benny Tan Zhi Peng, “Assessing the Effectiveness of Sentencing Guideline Judgments in Singapore issued Post-March 2013 and A Guide to Constructing Frameworks” (2018) 30 SAcLJ 1004 at para 20.)

122 In Singapore, sentencing guideline judgments are found in the decisions of the High Court or the Court of Appeal. In accordance with the doctrine of stare decisis, such judgments are binding on any lower courts. This means that they will not only influence the way that future offenders are to be treated by other judges, but also have the potential to radically reform future sentencing trends for a particular offence.

123 It is for this very reason that the decision to lay down a sentencing guideline is not one that is undertaken lightly by our courts. In all cases where a court chooses to do so, the eventual guideline is invariably a product of intense deliberation. At the minimum, the court will have to answer two questions in respect of the type and content of the sentencing guideline.

124 First, what type of guideline will be best suited for the particular offence or the particular factual matrix? Sentencing guidelines are judicial creations that take many forms. Each comes with its own advantages and disadvantages that may make it uniquely well-suited to a specific context, but completely disastrous in another (see generally Terence Ng ([64] supra) at [26]–[34]).

125 As the parties’ submissions focus on sentencing frameworks, I will not delve into the advantages and disadvantages of other types of guideline judgments such as the benchmark approach. This has been adequately explored by the Court of Appeal in Terence Ng at [31]–[32]. Instead, I focus directly on sentencing frameworks.

126 It will be apparent from my summary of the parties’ proposed sentencing frameworks that sentencing frameworks tend to focus on deriving a preliminary sentence based on the presence or absence of certain key sentencing parameters. The preliminary sentence is then adjusted after the sentencing court has regard to the other relevant factors present in the case. The key sentencing parameters are “independent variables” as they will independently have an impact on the indicative starting sentence for a particular offender. The preliminary sentence is the “indicative sentence” or the “indicative starting sentence” as referred to above. The preliminary sentence is also the “dependent variable” as it is determined by the interaction of the independent variables.

127 To forestall confusion later, I elaborate on the difference between a sentencing parameter and a sentencing factor. A sentencing factor is an aggravating or mitigating factor that exists in the particular factual matrix of the offender’s offence. A sentencing parameter exists because of the design of the sentencing framework in question – the creator of a sentencing framework designs the framework such that any individual applying the framework will have to have regard to one or more sets of considerations (ie, parameters) to arrive at the indicative starting sentence. In some cases, the value and extent of a sentencing parameter may be derived from a number of sentencing factors. For example, the level of harm caused by an offence may be determined by the level of physical and emotional harm suffered by the victim and the public disquiet caused by the offence. In other cases, the value and extent of a sentencing parameter may be derived from just one main sentencing factor. An example of this is in Poh Boon Kiat v Public Prosecutor [2014] 4 SLR 892 (“Poh Boon Kiat”). There, the principal factual elements of vice-related offences were held to be (a) the manner and extent of the offender’s role in the vice syndicate (ie, the chosen primary determinant of offender’s culpability); and (b) treatment of the prostitute (ie, the chosen primary determinant of the harm caused by the offence) at [75]–[76] (see also Terence Ng at [34]). The two sentencing parameters of harm and culpability are thus determined with reference to only one sentencing factor each.

128 Returning to the point at [124], a general sentencing framework may take the form of a single independent variable sentencing framework (“Single Variable Framework” or “Single Variable Sentencing Framework”) that focuses on the presence of one dominant sentencing parameter to determine an indicative sentence. This framework is well suited for offences like drug trafficking and cigarette smuggling for two reasons: (a) there is one principal determinant of sentence for all factual scenarios covered by the offence; and (b) this independent variable is measurable according to a single (usually quantitative) metric (see Terence Ng at [29]–[30]). Ishibe’s graphical analysis is essentially a Single Variable Framework, where the dominant sentencing parameter is taken (or first assumed) to be the amount of gratification received and used to derive an indicative starting sentence, which is then adjusted to take into account all other aggravating and mitigating factors to arrive at the final sentence.

129 However, a Single Variable Framework will be less suitable for offences in which there is more than one dominant or principal determinant of the indicative sentence, one example being the offence of causing hurt under s 323 of the Penal Code (see Low Song Chye v Public Prosecutor and another appeal [2019] 5 SLR 526). There, the two principal factual elements of “harm” and “culpability” separately play major roles in determining the indicative starting sentence. It is thus necessary to have regard to a two variable sentencing framework (“Double Variable Framework” or “Double Variable Sentencing Framework”). This may take the form of a sentencing matrix as defined in Terence Ng ([64] supra) at [33], which considers only two principal factual elements (ie, two sentencing parameters) to determine an indicative starting sentence. An example of this can be seen in Poh Boon Kiat in the context of vice-related offences (see above at [127]). More recently, Double Variable Frameworks have taken the form of a harm-culpability matrix which focuses on a wide variety of offence-specific factors which go towards the harm caused by the offence and the offender’s culpability to determine the severity of the offence (see Logachev ([68] supra) at [36]).

130 For clarity, I must add that Double Variable Frameworks do not invariably require a consideration of “harm” and “culpability” as the only two possible dominant independent variables influencing the determination of the indicative starting sentence. It may be any two principal factual elements (ie, parameters) that are key determinants of sentence. For example, in Tay Wee Kiat and another v Public Prosecutor and another appeal [2018] 4 SLR 1315, the three-judge High Court set out a sentencing framework for offences under s 323 read with s 73 of the Penal Code which uses physical harm and psychological harm as the two principal determinants of the indicative starting sentence (at [70]–[71]).

131 There are also offences where more than two key variables have a major impact on the indicative sentence. A sentencing framework that is capable of factoring more than two key independent variables to determine an indicative sentence is a multi-variable framework (“Multi-Variable Framework” or “Multi-Variable Sentencing Framework”). An example of this is the approach taken to the offence of unlawful stalking under s 7 of the Protection from Harassment Act (Cap 256A, 2015 Rev Ed) which requires consideration of, inter alia, the duration and frequency of stalking, the degree of intrusion into the victim’s life, the vulnerability of the victim, etc (see Lim Teck Kim v Public Prosecutor [2019] 5 SLR 279 (“Lim Teck Kim”) at [30]). For a Multi-Variable Framework, the design tends to take the form of a points-based scoring system because the framework has to handle multiple variables each with its own weight affecting the final outcome (see [140] below). I add that a points-based framework to determine the indicative starting sentence is only possible if the court is also able to determine the relative weights to be given to each of the various key sentencing parameters (ie, independent variables) and calibrate the sentencing framework to take this into account.

132 As the number of key sentencing parameters (ie, independent variables) in the sentencing framework increases, the indicative starting sentence will be more nuanced and fine-tuned to the facts. In other words, the magnitude of adjustment required from the indicative starting sentence to the final sentence is likely to be smaller for a Multi-Variable Framework than say, a Single or Double Variable Framework given that a larger number of key sentencing parameters have already been considered and accounted for when determining the indicative starting sentence in a Multi-Variable Framework. The attendant downside to this is that Multi-Variable Frameworks tend to be more complex than Single or Double Variable Frameworks when it comes to determining the indicative starting sentence. As such, there will always be a trade-off in terms of refinement (ie, the number of independent variables to be taken into account when determining the indicative starting sentence as the dependent variable) and the ease of understanding and application of the sentencing framework.

133 A framework that is “suitable” in respect of a particular offence is a framework that strikes a good balance between the need for a refined indicative starting sentence that takes into account the key sentencing parameter(s) and its ease of application.

134 The important point to appreciate is that in all cases where a judge decides to lay down a sentencing guideline or framework, a decision must be made as to the type of the framework because each type of guideline or framework has its own strengths and weaknesses. At the end of the day, the court’s task remains the same, namely, to determine the appropriate sentence for the particular offender before it. To do so, the sentencing court must take into account all the relevant facts and circumstances of the case. Sentencing frameworks and guidelines simply serve as tools to guide its discretion.

135 Before moving on to the second question of “content”, I pause to emphasise an important point in respect of sentencing frameworks that take the form of a graph. A Single Variable Framework does not have to take the form of a two-dimensional (“2D”) graph. It may well be a table, eg, in Vasentha d/o Joseph v Public Prosecutor [2015] 5 SLR 122 at [47], in the context of the indicative starting sentences for first-time offenders trafficking in diamorphine.

136 A graph is no more than a simple pictorial presentation of the relationship between the single independent variable on the one hand (which is normally depicted on the “x” or horizontal axis of the graph) and the dependent variable on the other (which in our present context is the indicative starting sentence, normally depicted on the “y” or vertical axis of the graph). The same information represented in the graph may also be alternatively presented in a table format. The format of presentation (be it a graph or a table) is a mere matter of form and not of substance. Any argument that a graph must necessarily be inferior to a table because it is too mathematical is inherently an illogical one that should be rejected. The principal driving force in the choice of the presentation format (be it a graph or a table) should be in the relative ease of understanding and the efficacy of its application. Both presentation formats will give the same indicative starting sentence when given the same independent variable as input data to feed into the framework (eg, the weight of the drug being trafficked).

137 I reiterate that a graph is merely an alternative method of representing the contents of a framework in a more understandable pictorial fashion. Like the tabular form or the harm-culpability matrix, it does not constrain a judge’s sentencing discretion (at least, not more than a table or the harm-culpability matrix does), nor does it transform the sentencing exercise into a mathematical inquiry.

138 The same applies to a Double Variable Framework. This may take the form of a table, a harm-culpability matrix which lays out indicative sentencing ranges, or a three-dimensional (“3D”) model which lays out the indicative sentencing points. An example of such a table can be seen in the decision of Public Prosecutor v Lai Teck Guan [2018] 5 SLR 852 (“Lai Teck Guan”) at [42]–[43]. There, the two independent variables are the weight of diamorphine and the criminal history of the offender (ie, whether he is a first-time or second-time offender). The interaction of these two independent variables gives rise to the indicative imprisonment sentence to be applied to a particular offender (ie, the dependent variable).

139 A Multi-Variable Framework cannot be properly represented by a 3D graph because it is near impossible to fully render a sentencing framework comprising of three or more key independent variables affecting the indicative starting sentence (ie, the dependent variable) in a 3D world. Neither can it be readily represented in a tabular format.

140 For such Multi-Variable Frameworks, I believe that the only practicable and sensible methodology is to adopt a “scoring system” or a “points system” where each of the multiple key independent variables is assigned a range of values depending on its relative weight (vis-à-vis other key independent variables) in affecting the indicative starting sentence to be scored (as a dependent variable) on a score sheet. The overall score is then mapped on to a scale that gives the indicative starting sentence. This methodology is neither unique nor innovative. It has been commonly applied in many different fields including those in education, science, economics, psychology, psychiatry and medicine.

141 A simple illustration of this is in how examination papers are scored. The dependent variable is the final mark given to a particular student (ie, the outcome of the examination). This is determined by the individual’s score on a variety of examination questions (ie, each question being a single independent variable). Each question may be assigned a different maximum score depending on the importance and difficulty of the question (ie, constituting their relative weight in affecting the final marks). The examiner will assess the number of marks to be given for each answer bearing in mind the maximum number of marks that can be given for that question. The aggregate marks scored for all the answers to all the questions will determine the final mark (or grade, eg, A, B, C, D, E or F) of the student (ie, the dependent variable) for that examination paper.

142 This scoring methodology where multiple independent variables are involved in the assessment for a single final outcome (ie, the single dependent variable) is employed in a plethora of situations, eg, in determining the Intelligence Quotient of a person in psychological tests, in testing for the presence and degree of severity of a particular mental illness in psychiatric examination, and even in assessing the corporate governance practices of Singapore-listed companies in the “Singapore Governance and Transparency Index” (“SGTI”). Focussing on the SGTI, a company’s score is based on two components. First, the base score which comprises of five domains: (a) board responsibilities (35 points); (b) rights of shareholders (20 points); (c) engagement of stakeholders (10 points); (d) accountability and audit (10 points); and (e) disclosure and transparency (25 points). Adjustments are then made to the base score in the form of adding bonuses and/or subtracting penalties to arrive at the company’s overall SGTI score.

143 The point is simply this: a sentencing framework of whatever form (even one represented graphically) is a mere sentencing tool to assist a sentencing judge. It does not and cannot override judicial discretion. A sentencing judge must still come to an independent decision as to the appropriate sentence (after considering the severity of the offence, the circumstances of the offender and any other relevant factors). Having said this, a sentencing tool, when designed, must nevertheless remain inherently logical, have conceptual integrity, and provide consistent indicative sentencing outcomes when applied to the same set of facts. Otherwise, it loses its attractiveness.

144 After considering the type of sentencing guideline, the second question the court must answer is what are the contents of the sentencing guideline? In this section, I focus specifically on the design of a sentencing framework because I have previously found that a general sentencing framework will be appropriate in these appeals.

145 The contents of a sentencing framework extend beyond a list of the relevant aggravating and mitigating factors and include a consideration of what the appropriate indicative starting sentences might be in various categories of the offence.

146 When filling in the contents of a sentencing framework and determining the indicative starting sentences that apply for different categories of fact scenarios, the court must bear in mind several important general principles, which if followed will help attain the objectives of having a coherent and logical framework that has conceptual integrity, and hopefully also provide broadly consistent indicative sentencing outcomes when applied to the same set of facts.

147 The sentence must be proportionate to the crime. All things being equal, as the severity of the crime increases, the sentence meted out must also increase, up to the statutory limit prescribed by Parliament. In the context of the sentencing frameworks, this means that the indicative starting sentences or indicative sentencing ranges prescribed must increase proportionally to the severity of the offence (as determined by the quantum of gratification, or the harm and culpability, or even multiple other factors as the case may be). I refer to this as the “Proportionality principle”.

148 The indicative starting sentences prescribed by the sentencing framework must increase smoothly and continuously, in tandem with the increasing severity of the crime (the “Continuity principle”). The sentencing framework aims to spread out the entire range of possible sentences across the full spectrum of offending under the particular offence-creating provision (or under a particular category of offending under the provision, eg, purely private sector corruption). There must not be any unexplained gaps in the indicative starting sentences prescribed under the framework. This is for two main reasons.

149 First, unexplained gaps are inconsistent with the aim of the sentencing framework model, ie, to spread out the entire range of possible sentences across the full spectrum of criminal offending that falls within a particular offence creating provision. The presence of unexplained gaps arbitrarily restricts the sentencing court from selecting certain indicative sentences (falling between the lower and upper end of that gap) even though they may be warranted on the facts.

150 Second, unexplained gaps in the sentences prescribed by the sentencing framework are inconsistent with the abovementioned general principle that sentences ought to be proportional to the severity of criminal conduct. An offender who falls into a gap not covered by the prescribed sentencing range will not be able to receive a sentence that increases in proportion to the severity of his crime: the possible sentence available to him will be restricted to the sentences on both sides of the gap. I provide a pictorial illustration for this at Annex C. If the indicative starting sentence shown in the sentencing framework suddenly jumps upwards when the overall criminal culpability only increases slightly, it will cause the indicative starting sentence to jump from the lower end to the upper end of that gap upon crossing that gap. This is tantamount to having a sentencing framework with an inherent minimum indicative sentence starting at the upper end of that gap. I illustrate this point with reference to the offence of drug trafficking under s 5 of the Misuse of Drugs Act (Cap 185, 2008 Rev Ed) (“MDA”). Assuming that: (a) at 5g of heroin trafficked, the indicative starting sentence is 5 years’ imprisonment; and (b) at 5.01g of heroin trafficked, the indicative starting sentence jumps to 7 years’ imprisonment, there will then be an upward gap of 2 years’ imprisonment at the 5g mark. This amounts to having a judicially created minimum indicative starting sentence of 7 years’ imprisonment embedded within the sentencing framework for trafficking in just slightly more than 5g of heroin. Gaps therefore have the effect of judicially legislating a minimum indicative starting sentence at points where gaps in the indicative starting sentences occur.

151 There may be cases in which gaps in the sentencing range are necessary and/or unavoidable. One example of this occurs when Parliament prescribes a mandatory minimum sentence for a particular instance of offending under the relevant provision. For example, an offender who traffics in exactly 10g of diamorphine will face a mandatory minimum sentence of 20 years’ imprisonment and 15 strokes of the cane (see Second Schedule to the MDA), while an identical offender who traffics in 9.99g of diamorphine will face a maximum of 15 years’ imprisonment and 11 strokes of the cane (see Lai Teck Guan ([138] supra) at [42]). However, apart from such instances where they are unavoidable, unexplained gaps in the sentencing ranges must be avoided when filling in the contents of a sentencing framework.

152 When a court decides upon the contents of a sentencing framework, it should include the full range of sentences as prescribed by Parliament (including the different types of sentences available), unless there are very good reasons for not doing so. I refer to this as the “Completeness principle”.

153 A sentencing framework that fails to do so runs the risks of prescribing judicially created minimum sentences, imposing caps on the maximum punishment available under the section, creating gaps in the sentence range, and/or completely ignoring a particular type of punishment made available under the provision.

154 Under ss 6(a) and 6(b) of the PCA, the maximum punishment prescribed by Parliament is 5 years’ imprisonment and a fine not exceeding $100,000. As such, a sentencing framework for purely private sector corruption cases must be capable of covering the entire range of imprisonment sentences and fines, and not simply ignore the possibility of imposing an imprisonment term in conjunction with a fine. The sentencing framework will be flawed if it ignores the fact that the maximum imprisonment term is 5 years, and only provides for imprisonment terms up to 4 years. It will also be flawed if it completely ignores the possibility of a fine in conjunction with an imprisonment term for an offender.

155 The fourth principle is of particular relevance to sentencing frameworks as it directly addresses the relationship between the degree of specificity of the values for the independent variables chosen as “inputs” to the framework and the degree of specificity of the dependent variable being the indicative starting sentence that is the “output” or ultimate result of applying the framework using the independent variables as inputs to the framework. The principle is that where the values of independent variables have been assessed with a high degree of specificity from a given set of facts, there should only be one indicative starting sentence (ie, the dependent variable) as an output from applying the framework and not an indicative range of starting sentences as an output. Where, however, the independent variables have not been concretely assessed to be of a particular severity (because of factual uncertainty or because the court has not yet been able to determine the more precise weight and severity of the various factors involved in the assessment of each of the specific parameters – ie, independent variables), then the values for the independent variables as inputs into the framework will not be a pinpoint value but a range. Inputs of independent variables as range inputs into the framework must necessarily result in a range output from applying the framework (ie, a range of indicative starting sentences will inevitably be thrown up as a result of applying the framework by inserting range inputs for the independent variables to the framework).

156 Put simply, a point input (being a specific value for each independent variable input) for a point output (being a definitive indicative starting sentence as a dependent variable output from applying the framework), and a range input (being the possible range of values assessed for each independent variable) for a range output (being the indicative sentencing range thrown up as a dependent variable output from applying the framework). I refer to it as the Single Point principle.

157 It is helpful to begin by considering the relationship between the range of prescribed sentences for an offence under the law, the specific indicative starting sentence derived from a sentencing framework and the final sentence imposed by the court. The prescribed sentencing range in a written law reflects the legislatively created sentencing range set out in the relevant statute for the particular offence. It is called a sentencing “range” because it is made up of multiple discrete points (eg, an imprisonment term of 1 day, 5 weeks, 10 years, etc, 12 strokes of the cane or a fine of $5, $5,000 or $50,000). An offender can only be sentenced to one final sentence out of the entire prescribed sentencing range: he or she cannot be sentenced to a range of sentences (eg, between 18 to 24 months’ imprisonment). The final sentence must be a specific pinpoint sentence, say of 18 months’ imprisonment, no more and no less. In the same way, in applying a sentencing framework, the offender can only receive one indicative starting sentence out of the entire prescribed sentencing range given a specific assessed degree of severity or value for each of the particular set or combination of independent variables chosen for the design of the sentencing framework (ie, assuming that there is no uncertainty as to the magnitude, degree or value of each independent variable upon assessment by the court).

158 In a sentencing framework, the indicative starting sentence is normally determined by reference to one particular independent variable for a Single Variable Framework, a combination of two independent variables for a Double Variable Framework, or a combination of more than two independent variables for a Multi-Variable Framework. Examples of these include a combination of the two independent variables of harm and culpability (ie, the harm-culpability matrix) in a Double Variable Framework; and a single independent variable comprising the amount of money misappropriated by an offender under s 409 of the Penal Code per the Single Variable Framework adopted for sentencing in Public Prosecutor v Ewe Pang Kooi [2019] SGHC 166 at [20].

159 The Single Point principle thus requires that the framework should throw up only one indicative starting sentence (as a theoretical concept) after the court completes its assessment of the factual matrix in which the crime was committed and is able to arrive at a particular and definitive level of severity or value for all the independent variable(s) chosen for the particular framework. In such a circumstance, the framework ought not to throw up a range of indicative starting sentences as its output. While there may be certain combinations of values for each of the various independent variables (or of one independent variable in a Single Variable Framework) that may give rise to the same indicative starting sentence, each combination of input values can only give rise to one indicative starting sentence as an output value of the dependent variable. This is for consistency of outcomes for one same set of facts. If one combination of input values for each of the various independent variables can have two possible indicative starting sentences, or a range of possible indicative starting sentences as output values, there will be much room for inconsistency when applying the framework to the same set of facts.

160 Using the example of a Single Variable Framework, if an offender is charged for trafficking in exactly 5g of heroin (the equivalent of a pinpoint input value to the framework), and the indicative starting sentence for trafficking in exactly 5g of heroin derived from applying the sentencing framework can be in a range of 5 to 7 years’ imprisonment, then a judge in one court can chose a starting sentence of 7 years’ imprisonment, and another judge in another court for another case involving trafficking in exactly 5g of heroin can chose to start with a totally different sentence of 5 years’ imprisonment as the indicative starting point, although both cases are identical and involve exactly the same set of facts of trafficking in 5g of heroin. This sort of framework inherently magnifies inconsistency in sentencing. The same applies for Double or Multiple Variable Frameworks. In essence, for each specific set of combination or combinations of the independent variables all with definitive assessed pinpoint values (ie, singular point value for each of the independent variable(s) as inputs to the sentencing framework), there can only be one identifiable indicative starting sentence as a pinpoint output from applying the sentencing framework.

161 In other words, once a sentencing court considers the overall factual matrix, assesses all the material facts and their weightage relevant to the assessment of the degree of severity or value of each of the independent variables and eventually arrives at a definitive pinpoint value for each of these independent variables to be inserted as inputs into the sentencing framework as designed, then the resulting output from the sentencing framework will have to be a definitive pinpoint indicative starting sentence and not a range of possible indicative starting sentences. Where it is not possible to assess the values of the independent variables with precision as inputs to the framework (which is often the case in a practical application), then one has to input a range of values ie, a range input into the framework (eg a range of possible values of harm assessed somewhere in the region of the lower end of the “high segment” on the scale for the independent variable of “harm”), which will then throw up a range output (ie a range of possible indicative starting sentences) for the court to choose from. The court will then have to use its discretion to eventually pick one indicative starting sentence out of that range output to work with for the next stage of sentencing. This does not mean that it is conceptually correct to say that a pinpoint input can give rise to a range output, or that a range input can give rise to a pinpoint output. In a properly designed framework, a pinpoint input can give rise only to a pinpoint output for overall consistency, and a range input must necessarily give rise to a range output.

162 The Single Point principle is not inconsistent with the use of indicative starting sentencing ranges in the process of sentencing. Indeed, it will be observed that sentencing frameworks (in particular those using a harm-culpability matrix to determine the indicative sentence) often mention the need to locate an indicative starting sentencing range that applies to the accused. There is no inconsistency for two reasons. One, because this is a preliminary step that a court will ordinarily have to cross before finally arriving at a precise point for the indicative starting sentence. Two, if there is uncertainty such that the court cannot arrive at a specific value for each of the independent variables or such that the court can only arrive at a specific range in its assessment of the values or degree of severity of the independent variables, the court then has no choice but to input a range input for each of the independent variables into the framework and consequently obtain a range output in the form of a range of indicative starting sentence when applying the framework. The framework itself cannot be faulted if it gives an imprecise output when given an imprecise input. But it can be faulted if the framework is unable to give a precise output when given a precise input. In other words, a court must work with an indicative starting sentencing range as the output if it has not yet or is unable to come to a precise determination for its own input which the framework requires. Using the example of a harm-culpability matrix, if the court is undecided as to the specific level of harm caused by the offender or of his culpability and can only say that the harm is preliminarily assessed to be of a certain range and the culpability is preliminarily assessed to be of a certain range in terms of the degree of severity, the consequence of which must necessarily be that the sentencing framework will provide a possible range of indicative starting sentences as guidance to the court, which has thus far not been able to make up its mind fully as yet. Similarly, if the court is faced with factual uncertainty and is unable to pinpoint the combination of harm and culpability, the court would derive an indicative sentencing range from the framework (see also [166]–[169] below). But at some point, the court must inevitably come down to a pinpoint sentence in the exercise of its judicial discretion having regard to whatever guidance and information that may be provided by the sentencing framework.

163 This does not mean that the framework in its design therefore should provide a range of indicative sentences for each precise point in the framework (ie, for each precise point along the scale for the single independent variable in a Single Variable Framework or for each combination in the array of different combinations of multiple independent variables of different precise values in a Double or Multi-Variable Framework). A framework designed to provide a range of starting sentences available to be chosen for each precise point or each precise combination is to be eschewed because it creates incoherency and inconsistent outcomes when applied. The logic of doing so is also difficult to comprehend. I see no good justification for such a design.

164 In other words, if at the first stage the court is able to determine, based on the factual matrix of the case, a particular precise level of harm and a particular precise level of culpability, then the sentencing framework must provide one indicative starting sentence only. However, if the court is unable to do so (or chooses to do so at a later stage) and can only assess the level of harm as possibly falling into a certain range on the scale of harm, and the level of culpability as falling into a certain range on the scale of culpability, then the same sentencing framework (designed with the Single Point principle in mind) will inevitably furnish an indicative starting sentence as possibly falling within a certain range as a guide to the court when exercising its judicial discretion before finally determining a pinpoint sentence for the offender. I reiterate that no offender can be sentenced to between 2 to 3 years’ imprisonment, 5 to 7 strokes of a cane and a fine of $2,000 to $3,000.

165 I illustrate the above point by again using the quintessential example of a Single Variable Framework: the sentencing framework for first-time drug traffickers. If the framework states that for trafficking in precisely 5g of heroin, the indicative starting sentence is precisely 5 years’ imprisonment, and for trafficking in precisely 6g of heroin, the indicative starting sentence is precisely 6 years’ imprisonment, and if the court is faced with a case where the court is unable to come to a landing at the first stage of sentencing as to the precise level of heroin the offender has trafficked in and the court only able to estimate that the amount of heroin trafficked to be between 5.4g and 5.8g of heroin, then the same sentencing framework (designed with the Single Point principle in mind) will necessarily throw up an indicative starting sentence to be in a range between 5.4 years and 5.8 years, and the court is left to exercise its judicial discretion to decide what the eventual pinpoint starting sentence should be in the light of the circumstances of the case. However, if at the first stage of sentencing the court is in fact able to come to a landing on the precise value of that independent variable, eg, 5.6g of heroin being trafficked, then the sentencing framework will throw up an indicative starting sentence as precisely 5.6 years’ imprisonment as a guide to assist the sentencing judge to exercise his judicial discretion. In such a case, the sentencing framework will not throw up a range of indicative starting sentences say of between 5.4 years to 5.8 years when the court is already able to decide that the precise amount of heroin trafficked is 5.6g.

166 The Single Point principle similarly must apply to a Double Variable Framework. I illustrate this with reference to a harm-culpability matrix. If the court is able to come to a more precise landing as to the specific level of harm and the specific level of culpability, the framework will throw up a definitive starting indicative sentence and not a range of possible indicative sentences. However, if the court is faced with uncertainty and is unable to come to a more precise landing as the level of harm and culpability such that it can only ascertain a range of possible levels of harm and a range of possible levels of culpability given a certain factual matrix, then the sentencing framework will invariably throw up a range of possible indicative starting sentences to assist the court (since the data input into the framework is range data and not a specific data point).

167 The court has two options.

168 The first option is to bite the bullet and resolve the uncertainty at the first stage. In other words, after first identifying a range of indicative sentences that can apply to the specific offender, the court can then use this range as a guide for its exercise of judicial discretion to determine a pinpoint indicative starting sentence at stage one which can then be adjusted at subsequent stages of sentencing to arrive at the final sentence. Using the Prosecution’s framework as an example, a sentencing court can first recognise that a case falls within the moderate harm, medium culpability range but be unsure as to where exactly the case falls. The court will then arrive at an indicative sentencing range of 1 to 2 years’ imprisonment. Next, the court can recognise that the offender falls within the lower end of moderate harm and moderate culpability and thus narrow the indicative range down to 1 to 1.5 years’ imprisonment even though it still has not completely resolved the uncertainty. Eventually, the court must resolve the uncertainty in respect of the offence-specific factors and arrive at a particular pinpoint indicative starting sentence, say of 1 year and 3 months.

169 Alternatively, the court can defer matters to the second or even later stages and continue working with the range of indicative starting sentences thrown up by the framework (because of the continuing uncertainty as to the level of harm and the level of culpability at the first stage percolating to the second and later stages of sentencing). Thereafter, the court can adjust that range of indicative starting sentences at the second and later stages (or steps) of sentencing and end up with a range of possible final sentences at the end of the sentencing process. In the end, the court must still determine as best as it can, a pinpoint final sentence to sentence the offender with guidance from that range of possible final sentences. Returning to the same example, assuming that try as it might, the court cannot resolve the uncertainty at Stage 1 of the Prosecution’s framework with reference to the offence-specific factors. The indicative range remains at 1 to 2 years’ imprisonment. The court will then have to work with adjusting this range of 1 to 2 years’ imprisonment by having regard to the offender-specific factors. It may arrive at a range of 1 year and 3 months to 2 years and 3 months after factoring in the offender-specific factors, which it must eventually narrow to one final sentence to conclude the sentencing process.

170 Whichever method used is the court’s preference. But one point is abundantly clear: the court is simply not permitted to sentence an offender to a range of imprisonment terms simply because the court is unable to determine more precisely what the specific level of harm and the specific level of culpability is at the very first stage of sentencing. It would generate an absurd result – convicts sent to Changi Prison for 12 to 18 months, or 6 to 10 strokes of the cane.

171 I have discussed the Single Point principle at length because it is often misunderstood, especially in the context of sentencing frameworks that incorporate a harm-culpability matrix. At its heart, the principle simply means every single combination of a specified level of harm and a specified level of culpability based on a given set of facts gives rise to one indicative starting sentence which is reflective of the severity of the offence. As explained earlier, where the harm can only be ascertained to be within a particular range and/or the culpability can only be ascertained to be within a particular range, the sentencing framework will necessarily throw up for consideration a range of indicative starting sentences. Where both the harm and culpability can be ascertained to a more precise level, then the indicative starting sentence will be logically and correspondingly narrowed down to a more precise level in the sentencing framework. In short, a “point input” into the framework gives rise to a “point output”. A “range input” into the framework gives rise to a “range output”. In other words, a point output for a point input. A range output for a range input. Never a point output for a range input. And never a range output for a point input. Allow me to illustrate this with an example of what may go wrong when a pinpoint input into the sentencing framework gives rise to a range output for the indicative starting sentence. Let me assume that a pinpoint input of trafficking in 5g of heroin throws up a possible output range of indicative starting sentence of between 5 and 7 years’ of imprisonment; and a pinpoint input of trafficking in 6g throws up a possible output range of an indicative starting sentence of between 6 and 8 years’ of imprisonment when the sentencing framework is applied. An offender trafficking in 5g of heroin may be given an indicative starting sentence of 7 years’ imprisonment (ie, at the top end of the possible range). An offender trafficking in a higher amount of 6g of heroin may be given an indicative starting sentence of 6 years’ imprisonment (ie, at the lowest end of the possible range). The illogical and unprincipled result is that an offender trafficking in a smaller amount of heroin may be given a higher indicative starting sentence than one trafficking in a larger amount of heroin. Hence, a framework should not be designed such that a pinpoint input can give rise to a range output (ie, never a range output for a pinpoint input).

172 In any case, I re-emphasise that every offender must eventually be sentenced to one final precise pinpoint sentence at the end of the sentencing process even if a court, in the exercise of its sentencing discretion, is not able to come to a more precise determination as to the weight of all the factors relevant to determine more precisely where the key independent variables lie on their respective scales from the “very low level” to “very high level” as set out in the first stage of the sentencing framework.

173 I provide pictorial representations of Single and Double Variable Sentencing Frameworks that comply with the Proportionality and Continuity principles.

174 One way of representing the spread of sentences within a Single Variable Framework is a 2D graph – there is only one independent variable that independently determines the indicative starting sentence.

175 It can be seen from the curve that: (a) the increase in indicative sentence is smooth and continuous (ie, there must not be unexplained gaps or discontinuities); and (b) the increase slows and tapers off to a constant value which represents the statutorily prescribed maximum sentence for the most egregious forms of offending. For every precise point along the scale of the independent variable, there is only one corresponding indicative starting sentence along the scale for the indicative starting sentence, which is defined by the bold continuous line showing the relationship/correlation between the independent variable and the dependent variable. A precise point output is labelled “Point Output” on the vertical axis of the graph. It occurs if there is a “Point Input” as labelled on the horizontal axis of the graph. However, if there is range of uncertainty for the “input” for the independent variable, the input will be a “Range Input” as labelled on the horizontal axis, and the same graph will provide a “Range Output” for the dependent variable as labelled on the vertical axis which represents the range of possible values for the indicative starting sentence.

176 I should explain that the independent variable cannot start from zero. This is because when the independent variable is zero (or infinitesimally close to zero), no offence may be committed in the first place, and, as such there is no indicative sentence. An example of this is when a person is alleged to have trafficked in in 0g of diamorphine.

177 In designing a Single Variable Framework, the manner in which the sentences are spread out determines the gradient of the line graph. Taking the offence of smuggling uncustomed tobacco products as an example, it is well within the discretion of the court when designing a Single Variable Framework to decide that the bulk of the indicative sentencing range must be reserved for more severe forms of the offence (ie, larger quantities of cigarettes). This would mean that when the cigarette quantities are low, the indicative starting sentence will increase slowly (ie, with a gentler gradient), but when cigarette quantities are high, the indicative sentence will increase much faster (ie, with a steeper gradient). While the gradient of the line graph can change, it should generally be positive because the indicative starting sentence must increase in proportion to the severity of the crime, at least until the statutorily mandated maximum sentence. At that point, gradient is zero because the sentence can no longer increase further. What is clear is that the gradient cannot and must not be negative, because that will mean that as the quantity of cigarettes increases along the horizontal axis, the indicative starting sentence deceases along the vertical axis of the graph. This is wholly illogical and inconsistent with the Proportionality principle, and the design of the sentencing framework is therefore inherently flawed.

178 The visual representation of a Double Variable Framework is more complex. One way of representing the spread of indicative sentences within a Double Variable Framework is a 3D model or graph because three axes are needed: two “x” and “y” horizontal axes for the two independent variables and a third “z” vertical axis for the singular dependent variable (ie, the indicative starting sentence). The values ascribed along the “x” and “y” axes to represent each of the two independent variables determine the value of the dependent variable. In a 3D model that adheres to the Proportionality and Continuity principle, instead of a line generally trending upwards in a line graph pictorially representing a Single Variable Framework, there will be a surface generally trending upwards (never downwards) in a 3D model or 3D graph pictorially representing a Double Variable Framework (see Figure 4 below at [179].)

179 Below (Figure 4) is a 3D model to help visualise how the upwards sloping surface represents the indicative starting sentences in a harm-culpability matrix that complies with the Proportionality and Continuity Principles. The independent variables on the horizontal “x” and “y” axes are “harm” and “culpability” respectively on an increasing scale of severity extending outwards in the direction as shown, and the dependent variable will be the “indicative starting sentence” on the vertical “z” axis increasing in severity vertically upwards in the direction as shown. The positive gradient of the upwards sloping surface ensures that no embedded illogicality arises by way of a decreasing starting indicative sentence as the gravity of the harm and/or culpability increases in the Double Variable Framework.

180 Each of the yellow dots represents a single unique combination of harm and culpability each with its own specified level of severity, and which reflects the factual matrix of a particular offender’s offence.

181 Four observations may be made from the 3D model above:

(a) First, the indicative starting sentences increase smoothly and continuously as each of the two independent variables (eg, harm and/or culpability) increases in magnitude. This can be seen by the constantly upward sloping 3D surface. The direction of the three arrows on the model indicates increasing magnitude.

(b) Second, the indicative starting sentences increase until they reach a maximum value as represented by the flat expanding top of the 3D surface. As with the 2D graph, this represents the maximum sentence prescribed by Parliament. The flat top is expanding to represent the fact that when either one or both independent variables tend to infinity (eg, infinite harm and/or infinite culpability), the indicative starting sentence remains capped at the statutorily prescribed maximum. In such a situation, even if the one independent variable remains at a low value (eg, low culpability) but the other independent variable is at a very, very high level (eg, a very extreme level of harm approaching infinite harm), the indicative starting sentence can theoretically approach and may even reach the statutory maximum (as represented by the overhanging surface when culpability is near zero but harm tends towards infinity).

(c) Third, for illustration purposes, the 3D surface is divided up into different colours. This is meant to show that the indicative starting sentence is changing smoothly and continuously all along the 3D surface in any direction. Each part of the 3D surface can also be divided into a series of different lines, with each line joining all the points of the same vertical height measured along the vertical “z” axis (ie, akin to the contour lines of equal elevation in a topographic map) and each of these lines of equal elevation on the 3D surface will represent a different indicative starting sentence. The higher the elevation of the line, the greater the indicative starting sentence on the 3D model represented by that line. For illustration purposes, I have drawn out just two of the numerous lines of different indicative starting sentences (ie, the two red lines joining the yellow dots together). I term these “Elevation Lines of Equal Sentences” because the indicative starting sentence is the same along each and every point on each of those red lines. Each part of the red line is at same vertical height. In other words, while each combination of the two independent variables must yield a single indicative sentence, certain combinations may yield the same indicative starting sentence because they result in offences of the same overall severity. For instance, one point on the 3D surface representing a certain harm-culpability combination can have a higher harm but a lower culpability than another point on the 3D surface for another harm-culpability combination. As a result, there may be no change in the overall criminal severity between these two points because the effects of one independent variable increasing in severity is cancelled out by the effects of the other independent variable decreasing in severity. Accordingly, all the points on the 3D surface with the same indicative starting sentence may be joined up together to form one continuous contour line representing the same elevation, so to speak. This is represented by the numerous yellow dots (ie, different combinations of harm and culpability resulting in offences of the same severity and hence the same indicative sentences) joined together by a single red line. I elaborate on this below at [185] onwards.

182 I have previously explained how the steepness of the gradient of the 2D line graph will show how fast the indicative starting sentence increases with the severity of the offence (as determined by the single key independent variable chosen for a Single Variable Framework, eg, the weight of the drugs or the quantity of contraband cigarettes). The same applies here in respect of the gradient of the 3D surface for a Double Variable Framework. The steeper the slope of the 3D surface, the faster the indicative starting sentence increases for every corresponding increase in the severity of the offence (as determined by the combination of two key independent variables of harm and culpability).

183 To aid the reader in visualising the Double Variable Framework, I provide a picture of a physical mock-up of Figure 4. The flat expanding top representing the maximum sentence prescribed by Parliament (see [181(b)] above) has been omitted so as to focus the reader’s attention on the constantly upward sloping 3D surface representing the smooth increase of the indicative starting sentences as the severity of the two independent variables (ie, harm and culpability) increases (see above at [181(a)]).

184 I should explain that the above two diagrams and the image are merely pictorial representations to help in the conceptual understanding of how the spread of sentences in a Single or Double Variable Framework will generally look like. They are only conceptual visualisation tools to explain the operation of the Proportionality, Continuity, Completeness and Single Point principles, which must be complied with when designing sentencing frameworks. I have described them as conceptual visualisation tools because they have not been filled with any content or specific information such as the nature of the offence, the indicative sentencing range and the maximum sentence.

185 I previously alluded to the notion of Elevation Lines of Equal Sentences which exist in a Double Variable Framework.

186 Very briefly, the term “Equal Sentences” refers to the indicative starting sentences being the same. The term “Elevation Lines” refers to how they are represented on a 3D surface like those seen above in Figure 4. The magnitude of the indicative starting sentence is represented as the elevation or height on the vertical axis or the “z” axis. As such, when the 3D surface tilts upwards, the indicative starting sentence on each point on that 3D surface will increase as the elevation height of that point on the 3D surface increases. The vertical axis is made up of a range of values that represent the entire sentencing range for the offence – eg, in the context of ss 6(a) and 6(b) of the PCA, the “z” axis values will reflect the indicative imprisonment range from zero to a maximum of 5 years.

187 As mentioned previously, different combinations of the two independent variables (ie, combinations of different levels of the two parameters of harm and culpability) will give rise to offences of varying severities (and hence different indicative starting sentences). Some combinations of the two independent variables may give rise to offences of the same severity (and hence same indicative starting sentence). When these combinations of independent variables are portrayed on a 3D model, they will all be points on the same level or at the same height on the vertical “z” axis, therefore showing that they have same indicative starting sentences. Visually, this will look like a line of points of equal height on the vertical “z” axis much alike the contour lines of equal elevation on a topographic map, hence the term “Elevation Lines of Equal [indicative starting] Sentences”.

188 The concept of Elevation Lines of Equal Sentences is simply the idea that different combinations of the two independent variables in a Double Variable Framework can give rise to the same indicative starting sentence because they result in offences of the same overall level of criminal severity.

189 Corruption offences take place under a wide variety of factual situations and circumstances, even in the purely private sector context. Our courts have also recognised a number of key factors (eg, quanta of bribes, the involvement of a strategic industry, corruption of foreign public officials, the presence of threats and coercion) which have a significant impact on the severity of the offence, and hence the indicative starting sentence.

190 In this light, I agree with the Prosecution that Ishibe’s graphical approach (which is in essence a Single Variable Framework) premised on the use of the bribe quantum as the single independent variable to derive an indicative starting sentence may not give a good correlation of the seriousness of the offence with the indicative starting sentence because regard ought to be given to other more weighty sentencing parameters that affect the seriousness of the offence. While the quantum of gratification may be set as a sentencing parameter in a framework, and is an important sentencing factor in its own right, it may not always be the predominant consideration in the sentencing analysis for corruption offences; depending on the precise factual matrix before the sentencing court, other factors like coercion and threats (see eg, Romel ([65] supra)) may take centre stage. ¹⁰³ That said, I find Ishibe’s use of bribe quanta as a quantitative approach to sentencing under s 6(a) of the PCA to be an interesting idea.

191 I am also of the view that the use of a traditional sentencing matrix which focuses on only two key principal factual elements premised on two factors (as elucidated in Terence Ng ([64] supra) at [33]) is similarly incapable of fully encapsulating all the key sentencing factors that have an impact on the severity of the offence (this being distinct from the harm-culpability matrix in Logachev ([68] supra)).

192 Given the fact-specific nature of corruption offences under ss 6(a) and 6(b) of the PCA, I was originally inclined to craft a Multi-Variable Framework akin to the points-based system I had devised in Lim Teck Kim ([131] supra). To recap (see also [140]–[142] above), the points-based system identifies a number of key offence-specific factors that have an impact on the severity of the offence and ascribes to each of them a range of points. These points represent the intensity of that particular factor. The total number of points across all the key offence-specific factors are then tallied to obtain a rough approximation of the severity of the offence. This then corresponds to a particular indicative starting sentence. The points-based system is uniquely suited to offences that manifest in a large variety of ways. It allows the sentencing court to analyse a large number of key sentencing factors and weigh them individually. It results in an indicative starting sentence that is highly attuned to the facts of the case.

193 However, it is very difficult to devise a points-based system for offences under ss 6(a) and 6(b) of the PCA. Such a sentencing framework requires a court to determine the precise weight that ought to be placed on each key sentencing factor in the sentencing analysis relative to each of the other factors (as distinct from their individual magnitudes or intensities). The sentencing factors will then go towards determining the extent of the various sentencing parameters used to determine the severity of the offence (and hence the indicative starting sentence) in the framework. For example, assuming that a hypothetical offender has received a massive bribe of $10m from a private sector agent. Another hypothetical offender is a foreign official who receives a bribe of $10 in return for divulging highly confidential state secrets. Even assuming that both offenders are identical in all other respects, what weight is to be given to the sentencing factor of the quantum gratification as opposed to the involvement of foreign officials? No such relationship is forthcoming in the case law. Therefore, without an extensive database of very varied precedents to enable the relative weights of a variety of factors to be distilled from the “big data” so to speak, and to examine their relative influence on the seriousness of the offence and ultimately the indicative starting sentence to be determined, it is very challenging to design a points-based system for the wide factual spectrum of offences falling under ss 6(a) and 6(b) of the PCA.

194 In my judgment, an appropriate compromise solution is to adopt the Prosecution’s five-step framework (operating in two stages) which I am broadly in agreement with.

195 I term this a compromise solution because the Prosecution’s framework utilises a harm-culpability matrix to determine the indicative sentence. However, unlike a sentencing matrix which only has two principal factual elements, the Prosecution’s harm-culpability matrix is anchored on two principal agglomerated factual elements of harm and culpability (ie, two sentencing parameters), deriving each of these from the offence-specific factors found in each case. In other words, the Prosecution’s matrix allows the sentencing court to consider multiple offence-specific factors that go towards determining the extent of the harm and the culpability when determining the indicative starting sentence. Unlike the points-based system, the Prosecution’s sentencing framework does not set out the exact weight to be accorded to each offence-specific factor in determining the indicative sentence, but rather leaves it to the court’s discretion in each case.

196 As such, I agree with the Prosecution that its framework is capable of encapsulating the wide diversity of corrupt acts and varied factual situations in which purely private sector corruption may present itself. ¹⁰⁴ The consideration of both the weight and quality of factors that go towards harm and culpability at Steps 1 to 3 allows for a “general holistic assessment of the seriousness of the offence by reference to all the offence-specific factors” [emphasis in original] (Ye Lin Myint v Public Prosecutor [2019] 5 SLR 1005 (“Ye Lin Myint”) at [46]), as opposed to a consideration of one parameter like bribe-quantum, or two principal parameters premised on only two sentencing factors when determining the indicative starting sentence. This allows the indicative starting sentence to be highly attuned to the facts of the case.

197 In my view, the two-stage approach (ie, first considering offence-specific factors before turning to offender-specific factors, see [106]–[107] above) together with the Prosecution’s five-step framework is both intuitive and conceptually neat.

198 That being said, I do not agree with the structure and content of the Prosecution’s proposed table of indicative sentencing ranges, ie, the harm-culpability matrix. I am of the view that the Prosecution’s harm-culpability matrix has a number of inherent flaws which must be rectified.

199 Before explaining my views on this matter, I first make some conceptual clarifications about the Prosecution’s harm-culpability matrix as this will provide some context for the later discussion. In particular, these conceptual clarifications go some way to explain some of the flaws in the Prosecution’s harm-culpability matrix, and set the stage for the modified approach that I take.

200 For completeness, I add that Ishibe has also proposed a sentencing framework incorporating a harm-culpability matrix. However, I focus my attention on the Prosecution’s framework given his counsel’s confirmation that Ishibe is no longer proposing a sentencing framework. For the avoidance of doubt, the following analysis also applies to Ishibe’s framework.

201 To reiterate, the harm-culpability matrix in the Prosecution’s framework takes the form of a nine-grid box. The indicative sentencing range of zero to 5 years’ imprisonment and/or fine under ss 6(a) and 6(b) of the PCA is split up into nine indicative sentencing ranges situated within each box. To determine the indicative sentence, one must consider the various offence-specific factors that go towards harm and culpability. Harm and culpability are thus the two principal factual elements (ie, independent variables) derived from these factors and used to determine the indicative starting sentence (ie, dependent variable). I make some observations about this approach.

202 First, it is of utmost importance to understand the concept that harm and culpability are continuous and independent variables. They are independent variables in that they are parameters that operate separately and independently (of each other) to affect the indicative starting sentence. They are continuous in that they do not occur only in three discrete blocks of low, medium and high or slight, moderate and severe. Rather, they exist on a continuous spectrum that theoretically begins from zero and extends to infinity. This is because the offences under ss 6(a) and 6(b) of the PCA may be committed in a vast variety of ways, and correspondingly, be of vastly different severities. These severities cannot be forced into only three discrete categories of low, medium and high; or even nine further sub-divided and discrete categories of (a) low, medium and high in the low category; (b) low, medium and high in the medium category; and (c) low, medium and high in the high category, so on and so forth. This is why I describe them as being continuous independent variables that range in severity from very low to the very high in a continuous fashion, with no fixed categories or boxes as such. It is analogous to a line of continuously increasing severity, which comprises an infinite number of points of increasing severity adjacent to one another and packed very closely together to make up the said line.

203 This is a point that is often overlooked. There are a number of implications. One of them is that a sentencing court must recognise that even if two offenders fall within the same “box”, the offences committed may still be of varying severities within that “box”. For example, one offender may be in the lowest possible range within the medium-moderate box, while the other offender may be in the highest possible range within that the same box. They thus ought to be treated differently and accorded different indicative sentences. To illustrate this, consider the following box (Figure 5) for moderate harm and medium culpability in the Prosecution’s harm-culpability matrix:

The green dots represent two different offenders. The offender at the lower end of the moderate harm and medium culpability box (the lower left corner) should receive an indicative imprisonment of one year while the other offender (at the upper right corner) should receive two years according to range of indicative starting sentences provided for all offences falling within that box. Another implication of the fact that harm and culpability are continuous variables is that there is no leap in harm or culpability when going from, say, the lowest category to the middle one. Because harm and culpability are continuous variables, the increase in harm or culpability across the boundaries of these boxes is as smooth as the increase in harm or culpability within these boxes. This ties into my next observation.

204 Second, terms such as “low”, “medium”, or “high” are merely labels used to broadly refer to the level of harm and culpability present on the facts of a particular case. This is not a novel point and has been recently recognised by our courts (see for example, Wong Chee Meng ([10] supra) at [83], Ye Lin Myint ([196] supra) at [63]).

205 In a similar vein, while it may often be necessary to sub-divide the various degrees of harm and culpability into simple categories of low, medium and high when crafting a matrix (ie, the nine boxes representing different combinations of harm and culpability with different severities in the framework), this should be understood as a method of marking out generally how harm and culpability shift continuously and smoothly across the matrix in any direction (not merely just vertically and horizontally) rather than as an attempt to carve out nine discrete, non-overlapping blocks. There is a continuous increase in harm and culpability across these blocks, not jumps when the boundaries of these categories are crossed. In essence, the notion that there can only be nine possible combinations of harm and culpability for the purpose of obtaining an indicative starting sentence is wrong because there is in fact a whole canvas of continuous and adjacent points depicting all possible combinations of continuous harm and continuous culpability in each of the nine boxes in the whole harm-culpability matrix.

206 Third, it bears repeating that the purpose of Stage 1 of the Prosecution’s framework (see [106] above), comprising Steps 1 to 3 of the framework, is to derive a single indicative starting sentence for the offender based on one particular combination of a specific level of harm and culpability resulting from an assessment of all the relevant facts of the case. A sentencing court has the unfettered discretion to consider and weigh all the offence-specific factors going separately to harm and culpability that are present on the facts of the case. However, once it has completed the weighing exercise and reached a landing on a particular level of harm and culpability (if the court is able to do so), the Double Variable Framework of harm and culpability must provide one indicative starting sentence that is commensurate with the overall severity of the offender’s offence. This is the Single Point principle.

207 The Single Point principle does not, in any way, detract from the discretion of a sentencing court. To illustrate this, I use the example of two courts who separately try a hypothetical Offender A for an offence under ss 6(a) of the PCA using the Prosecution’s framework. The two courts are wholly entitled to come to different conclusions as to the severity of the offence committed by Offender A, after assessing the facts of the case and hearing the submissions of the parties, in the exercise of their discretion. However, if they make the exact same assessment of the severity of the offence (eg, by identifying the same level of harm and the same level of culpability with reference to the same sentencing framework), Offender A ought to receive the same indicative starting sentence because the indicative starting sentence is determined solely by reference to harm and culpability and nothing else at this first stage of the sentencing process. In other words, the discretion of the court lies in the assessment of how the offence-specific factors contribute to the severity of the offence through the independent variables of harm and culpability, but once a particular severity is arrived at, the court has to arrive at a particular indicative starting sentence by applying the framework.

208 Having set out some of the conceptual clarifications as groundwork, I turn now to explain the issues I have found with the Prosecution’s harm-culpability matrix. I reproduce the Prosecution’s harm-culpability matrix for ease of reference:

209 First, the Prosecution’s framework is incomplete as it is unable to account for situations involving harm or culpability that go beyond “high” or “severe”. As mentioned previously, harm and culpability are continuous independent variables that exist on a continuum given the vast number of possible fact scenarios that fall under ss 6(a) and 6(b) of the PCA. There may well be an abnormal case in which a private sector agent accepts a small bribe to do a seemingly innocuous act but ends up causing an industrial accident at a power plant that causes a massive explosion. At best, the Prosecution’s framework simply does not apply in such a situation because it is meant to cater to the vast majority of situations, not all situations falling under ss 6(a) and 6(b) of the PCA.

210 If that much is clear, no real problem exists. The problem, however, arises if a sentencing court assumes that it should still apply the framework – in such a case, the offender would fall in the low culpability and severe harm box where the highest prescribed indicative sentence is 2 years’ imprisonment. This is less than half of the statutory maximum of 5 years’ imprisonment. It should also be borne in mind that if there are no offender-specific factors that warrant an upward adjustment of the sentence at Step 4 of the Prosecution’s framework, the sentence will remain at 2 years’ imprisonment. This means that in cases where the severity of the harm tends to infinity while culpability remains low or moderate (and vice versa), the Prosecution’s sentencing framework (if applied in this manner) will artificially restrict the sentencing court’s discretion and guide it to impose an indicative starting sentence of a maximum of 2 years (see the range of sentence of 1 to 2 years imprisonment found in both the top left box and in the bottom right box), which is unlikely to be commensurate with the severity of the offence.

211 However, the first problem is not serious and is largely a problem of how the Prosecution’s matrix is used. It can be overcome by: (a) accepting that the Prosecution’s framework simply does not apply in cases that present extreme facts; or (b) expressly providing that when harm or culpability tend towards infinity, the indicative starting sentence would correspondingly tend towards the maximum sentence prescribed under s 6 of the PCA. That being said, although this is an issue of application, the design of the harm-culpability matrix can prevent this kind of misunderstanding. I bear this in mind in coming to the matrix which I propose below in expressly providing for potentially infinite harm and culpability (see [259], [264] and [275] below).

212 Second, the Prosecution’s harm-culpability matrix fails to utilise the entire sentencing spectrum prescribed by Parliament. Section 6 of the PCA states that an offender:

213 While the Prosecution’s harm-culpability matrix recognises the possibility of fine or imprisonment, it does not factor in the possibility that the appropriate indicative starting sentence can be a combination of both types of sentences. This is contrary to the Completeness principle (see [154] above) and creates gaps in the indicative sentencing as an entire type of indicative sentence (ie, a fine in addition to imprisonment) is almost completely missing without good reason.

214 This is most obvious at the high culpability, severe harm box (ie, the top right box). It does not seem right that if a sentencing court decides that the offence before it is so severe that it is one of the worst of its kind, it remains limited to an indicative sentence of 5 years’ imprisonment even though Parliament has decreed that a fine not exceeding $100,000 may also be levied in addition to an imprisonment term.

215 I reiterate that under the Prosecution’s framework, all the offence-specific factors are to be accounted for fully at the Steps 1 to 3 and nowhere else, especially not at Step 4 which is meant for offender-specific factors only. Even if the court subsequently imposes an additional fine of $100,000 after considering the offender-specific factors at Step 4, this is, in principle, different from a determination that the maximum sentence may well be warranted after a consideration of all the offence-specific factors.

216 That said, I recognise that in the absence of a concrete elucidation of the correlation between fines and imprisonment (eg, that a fine of $X is equivalent in its deterrent and punitive effect on an offender to an imprisonment term of Y months), it is incredibly difficult for a sentencing framework to be able to pinpoint an appropriate indicative starting sentence involving a custodial term and/or a fine. A simple solution in the interim is to state explicitly that the indicative sentencing ranges in all eight of the nine boxes of the harm-culpability matrix includes a possible fine from $0 up to $100,000 in addition to the imprisonment term in those boxes, as I do so in the matrices that I arrive at in this case (see [264(f)] and [276] below).

217 Third, there is ambiguity in the Prosecution’s harm-culpability matrix as it is commonly applied or interpreted, ie, without due regard to the conceptual clarifications that I have made at [201]–[207] above. This ambiguity presents itself at four intersection points of the boxes (highlighted in red and blue) within the nine-grid harm-culpability matrix as follows:

218 I illustrate the ambiguity by using the intersection point coloured in blue. Given that harm and culpability exist on a continuum, it is not inconceivable that a particular combination of factors going to harm and culpability may cause an offender to fall squarely on the intersection points between the four boxes. I refer to this hypothetical offender as Offender B. There is ambiguity in the indicative starting sentence because three possible options are open for Offender B, depending on where Offender B is taken to land in the four boxes surrounding the intersection point:

(a) If he is taken to be part of the slight harm, low culpability box: a fine, presumably of $100,000, given that the severity of Offender B’s crime is the highest possible combination of slight harm and low culpability in the left bottom box.

(b) If he is taken to be in the moderate harm, medium culpability box: an imprisonment term of a year, given that the severity of Offender B’s crime is the lowest possible combination of moderate harm and medium culpability in the box in the middle of the matrix.

(c) If he is taken to be in the (i) moderate harm, low culpability box; or (ii) slight harm, medium culpability box: an imprisonment term that must be significantly less than a year. If the former, Offender B would be taken to have the harm caused that is the lowest possible in the “moderate” range but culpability is the highest possible in the “low” range. The 1-year indicative imprisonment term is reserved for offenders who present with the highest possible harm and culpability in the moderate harm, low culpability box, which is the bottom middle box in the matrix. For similar reasons, if Offender B is taken to fall within the slight harm, medium culpability box, his indicative imprisonment sentence must be significantly less than a year because the harm caused is the highest possible in the “slight” range, but culpability is the lowest possible in the “medium” range. The 1-year indicative imprisonment term is reserved for offenders who present with the highest possible harm and culpability in the slight harm, medium culpability box, which is the middle left box in the matrix.

219 It will be obvious that these options are mutually exclusive. Options (a) and (b) in particular, are quite far apart. This is a problem because it means that the three different courts may assess the same Offender B and arrive at the exact same decision in terms of the severity of his offence but may give him three drastically different indicative starting sentences. This cannot be explained away on the basis of the sentencing courts’ discretion.

220 The same ambiguity problem arises at each of the three other intersection points coloured in red in Figure 7 above.

221 This ambiguity is a fundamental flaw in the Prosecution’s sentencing framework. It arises because the sentencing framework is designed in such a way that presents these points of ambiguity as intersections between multiple possible combinations of different levels of harm and culpability. It appears to suggest that each of these points potentially represent multiple degrees of severity, and therefore, give rise to multiple possible sentences, which offends the Single Point principle – when in reality, they refer to the exact same point on the matrix with the same ascertained combination of factors going to the assessment of the degree of severity (ie, the values) for harm and culpability for that point. Further, the relative relationship between each of these intersections and the categories surrounding them is not made clear in such a sentencing framework. As commonly applied and interpreted, this gives more weight to the labels of slight, moderate, and severe harm; and low, medium, or high culpability, rather than the specific and determinate combinations of the values assessed for the two independent variables of harm and culpability. Hence, any given point appears to bear different significance depending on whether it is compared to the labels to the top and bottom, or left and right. It bears reminding that not all offences under ss 6(a) and 6(b) of the PCA necessarily fall neatly within one box. As mentioned previously, the labels of slight, moderate and severe for harm; and low, medium, high for culpability are merely labels used to broadly categorise the various levels of harm and culpability. Similarly, the nine boxes in the harm-culpability are also broad categorisations of the infinite number of possible combinations of harm and culpability for the offence. A sentencing framework should be designed to avoid these ambiguities and to reflect these conceptual clarifications.

222 Fourth, there are unexplained gaps in the indicative sentencing range in the Prosecution’s harm-culpability matrix. I will refer to these unexplained gaps as “cliffs” or “discontinuities”.

223 At first glance, the Prosecution’s harm-culpability framework appears to partially adhere to the Continuity principle by spreading out the entire range of possible imprisonment sentences across the full spectrum of offending under ss 6(a) and 6(b) of the PCA. All nine boxes in the matrix (excluding the first slight harm and low culpability box) appear to have continuous indicative sentencing ranges when one moves horizontally or vertically across the boxes.

224 I should explain that when I say “move across the boxes”, I mean that the court first considers a range of different offences that fall broadly within one box, and then moves on to consider more offences that fall broadly within another adjacent box. By way of illustration, as one moves horizontally from the “slight harm, medium culpability” box to the next box of “moderate harm, medium culpability” and finally to the third box of “severe harm, medium culpability”, the indicative imprisonment range increases smoothly from up to 1 year, 1 to 2 years, and finally, 2 to 3 years.

225 However, one is not restricted to moving horizontally or vertically across the harm-culpability matrix, one may also move diagonally.

226 The first type of cliff arises when one moves diagonally within the harm-culpability matrix, ie, where both harm and culpability are increasing or decreasing at the same time. I refer to these cliffs as “diagonal cliffs” and they occur at the areas marked out with red arrows in Figure 8 below.

227 To illustrate the diagonal cliffs, I use hypothetical Offenders C and D. Their positions in the matrix are denoted by the blue dots marked with “C” and “D” in Figure 8. Offender C’s harm and culpability are assessed to be the maximum possible such that “C” can fall within the slight harm, medium culpability box. Offender D’s harm and culpability are assessed to be minimum possible for “D” to fall within the moderate harm, high culpability box. The levels of both harm and culpability have increased very slightly as one moves from considering Offenders C to D, ie, a diagonal shift upwards. Theoretically, given that harm and culpability are continuous independent variables in the Prosecution’s harm-culpability matrix, Offender D should receive a slightly higher indicative sentence than Offender C.

228 Referring to the harm-culpability matrix above, this is however not the case. Offender C’s indicative sentence is 1 year’s imprisonment as “C” is the most egregious offender falling within the box, whereas Offender D’s indicative sentence is 2 year’s imprisonment as “D” is the least egregious offender within the box. There is a large diagonal cliff because the indicative starting imprisonment term doubles without any apparent explanation even though the severity of the criminal conduct has only increased very slightly from Offender C to D, both of whom represent a single point of sentence on the continuous scales of harm and culpability.

229 As the indicative starting sentences do not increase smoothly and continuously, in proportion to the severity of the criminal conduct, the Prosecution’s harm-culpability matrix does not comply with the principles of Proportionality and Continuity. Such a gap cannot be remedied by having regard to Step 4 of the framework. The gap is in the indicative starting sentence which is determined solely by reference to the offence-specific factors. Step 4 is concerned with finalising the indicative starting sentence by having regard to the offender-specific factors. The considerations are entirely different.

230 Furthermore, the gap cannot be explained by reference to the fact that both harm and culpability have increased in tandem, in contrast to a situation where only harm or culpability increases. The Prosecution’s framework accords equal weight to harm and culpability, technically speaking an increase in harm by two units (culpability remaining constant), should be treated in the same way as an increase in culpability and harm by one unit each. Further, this gap also cannot be explained away simply by observing that there is an increase in both the categories of harm and culpability (from slight to moderate, and medium to high respectively). As observed earlier (see [203]–[205] above), these categories are just labels, and there should be no significance attached to the crossing of these thresholds. Hence, the focus should only be on the relative increases of harm and culpability between Offenders C and D, rather than the fact that Offender D is in a higher category for harm and a higher category for culpability.

231 The Prosecution’s harm-culpability matrix also has linear discontinuities. These occur along the lines that demarcate the nine boxes. In Figure 9 below, vertical linear discontinuities are represented by purple lines, and horizontal linear discontinuities are represented by blue lines. The indicative starting sentences of the most and least egregious offenders falling within each box (apart from the low culpability and slight harm box) are denoted respectively by the red and green dots. The hypothetical Offenders E, F and G are used to illustrate the two types of linear discontinuities.

232 I first illustrate a vertical linear discontinuity, each of which is denoted by a purple line in the matrix shown above.

233 Offender E is the most egregious offender who may still be classified as falling within the slight harm, high culpability box. Offender F is the offender with the lowest degree of harm and culpability that allows him to fall within the moderate harm, high culpability box. The levels of harm assessed for Offender E and F are about the same but the level of culpability for Offender E is assessed to be far higher than that for Offender F. Yet, the indicative starting sentences for both Offenders E and F are the same at about 2 years’ imprisonment. This does not make any logical sense. The result is an embedded vertical linear discontinuity in the harm-culpability matrix because even a significant relative drop in the level of culpability from Offenders E to F (with the level of harm remaining somewhat constant) does not result in any change in the indicative starting sentence.

234 I turn now to horizontal linear discontinuities, which are denoted by the blue lines in Figure 9 above. Offender G is the most egregious offender that falls within the moderate harm, medium culpability box, and Offender F is the least egregious offender that falls within the moderate harm, high culpability box. There is a minute difference in the extent of culpability between Offenders G and F, but Offender G has caused a significantly higher level of harm when compared with Offender F. Yet there is no difference in the indicative starting sentence of 2 years’ imprisonment for both Offenders F and G. This illogical outcome results in a horizontal linear discontinuity in the harm-culpability matrix.

235 It is clear that at various points in the Prosecution’s framework, minute changes in harm and/or culpability cause discontinuities and disproportionate jumps and drops in the form of cliffs in the indicative starting sentences. I provide a pictorial representation of the various gaps in the indicative starting sentences within the Prosecution’s harm-culpability matrix at Annex C below.

236 In my judgment, a modified two-stage, five-step sentencing framework will resolve the problems faced by the Prosecution’s sentencing framework, and adhere to the Proportionality, Continuity, Completeness and Single Point principles (“the Modified Framework”).

237 The two-stages reflect the court’s consideration of offence-specific factors and offender-specific factors (see above at [106]–[107]). The five steps are also the same as that laid out by the Prosecution (above at [108]). The main difference between the Modified Framework and the Prosecution’s sentencing framework lies in the use and content of the harm-culpability matrix when assessing the severity of the offence at Steps 2 and 3 and the excluded consideration of the public service rationale.

238 In Step 1, the court identifies and assesses the relevant offence-specific factors present on the facts of the case.

239 The following, non-exhaustive offence-specific factors apply to purely private sector corruption cases under ss 6(a) and 6(b) of the PCA. ¹⁰⁵

For clarity, the offence-specific factors such as “actual loss to the principal” or “level of sophistication” are sentencing factors. The sentencing parameters for the Modified Framework are “harm” and “culpability”. The sentencing parameters must always be considered when applying the Modified Framework, while each of the sentencing factors may or may not be present on the facts (see also [127] above).

240 Many of the above factors are trite and have been routinely identified in the context of corruption offences. The table of offence-specific factors is almost identical to Menon CJ’s table in Wong Chee Meng ([10] supra) at [62], except for the addition of three items (highlighted in italics in the table above): (a) “harm caused to the giver of gratification”; (b) “the presence of threats, pressure or coercion”; and (c) the relevance of “amount or value of gratification given or received” to both “harm” and “culpability”. I agree with Menon CJ’s observations in respect of the various offence-specific factors and will only provide explanations in respect of those three additional items.

241 The factor of “harm caused to the giver of gratification” applies in cases where the corrupt acts caused or have the potential to cause significant detriment to the giver of the gratification. This may occur when the legitimate business of the giver was in fact interfered with because of the receiver’s conduct (ie, in a Romel category 3 situation, see below at [A.10]).

242 In Public Prosecutor v Tai Ah Poh Magistrate’s Appeal No 9046 of 2014 (26 March 2015) (as elucidated in Romel ([65] supra) at [52]–[58]), the receiver extracted a commitment from the giver to pay him $600 a month in exchange for the giver saving $450 to $500 a month by not having to employ a second cleaner for the nightshift. This was clearly a bad deal, but the giver acceded to it because the receiver had threatened to prevent the giver from obtaining subsequent cleaning contracts. The harm caused to the giver was not merely financial, but also in the impairment of his right to fairly access commercial opportunities. In other words, the giver was faced with the prospect of unfair prejudice where the renewal of his cleaning contract was concerned, unless he paid a bribe.

243 The presence of threats, coercion or pressure is a highly aggravating factor that is intended to capture a narrow stratum of cases that fall under Romel category 3. Courts should be sensitive to the various ways in which offenders may mask their threats, eg, by calling them “favours”. The presence of this offence-specific factor will heighten the culpability of the receiving party (see Romel at [29]). However, courts should be careful not to double count this factor insofar as it has already been considered in relation to an offender’s abuse of position.

244 There are presently conflicting High Court authorities in relation to the conceptual relevance of the amount or value of gratification given or received by an offender. In Ang Seng Thor ([79(b)79(c)] supra) at [46], V K Rajah JA pointed out that the size of the bribes in a corrupt gratification offence is not only linked to the culpability of the offender, but also to the harm caused by the offence. In Wong Chee Meng ([10] supra) at [71], Menon CJ acknowledged that “the quantum [of the gratification given or received] may to some extent also serve as a barometer of the degree of harm caused”, but held that “it would be preferable, in the interests of conceptual clarity, to regard the quantum of gratification as a factor going only towards culpability as far as the sentencing framework set out in [Wong Chee Meng] is concerned”. Menon CJ was of the view that “the degree of harm is sufficiently taken into account, for purposes of the sentencing framework here, by the [other] offence-specific factors going towards harm”.

245 In my view, the quantum or value of gratification received or given is a factor going towards both “harm” and “culpability” for the purposes of the Modified Framework.

246 The relevance of the amount or value of the gratification under the Modified Framework does not inhere in a simple numerical figure; it is not relevant in and of itself. The design of the Modified Framework is such that this factor is relevant only insofar as it contributes to the level of harm caused by the offender’s conduct and/or demonstrates the offender’s culpability. These two inquiries – into the contribution of the level of harm and the demonstration of the offender’s culpability – are separate and distinct inquiries. As Rajah JA aptly pointed out, the amount or value of the gratification is relevant to harm because the higher the bribe, the greater the corrupt influence exerted on the receiver. This presumptively leads to a greater subversion of the public interest because larger bribes are generally paid to lead receivers into graver transgressions. The focus is on the gravity of the particular “transgression” and as such, this factor must be “assessed along with the importance of the transaction sought to be influenced” in order to see its relevance to harm (Ang Seng Thor at [46]). In certain cases, the giver of the gratification may suffer harm because of the magnitude of the “gift” extracted from him by the offender, eg, because he has to divert much needed funds from elsewhere to fund the bribe demanded of him.

247 On the flip side of the coin, the amount or value of the gratification reflects an offender’s culpability. For the receiver, the size of the bribe is a good proxy for measuring his greed or personal gain from the offence. For the giver, it is a reflection of the level of influence or advantage that he wishes to secure through the bribe (Ang Seng Thor at [47]).

248 By way of example, let us assume that the hypothetical offender H works in Company A which recently granted two identical building projects worth $1,000,000 each to Companies B and C. Offender H subsequently choses to solicit $100,000 in bribes from the owners of Company B and C respectively by threatening to sabotage their projects with Company A if the bribes demanded are not paid. Unbeknownst to Offender H, Company C is facing financial difficulties and desperately needs the building project to avoid liquidation. Company C pays the bribe, but this further jeopardises its financial health and eventually causes it to enter into liquidation. Company B, on the other hand, is financially healthy and can afford to pay the bribe.

249 In respect of both Companies B and C, the size of the $100,000 bribe reflects the same level of greed and improper financial motivation on the part of Offender H. However, the act of giving the $100,000 results in a much larger negative impact on Company C than B because Company C is already in financial difficulty and can barely afford to pay it. In other words, the sum of $100,000 is relevant to assess the harm caused by Offender H because of its respective impact on the financial health of Companies B and C. If a sentencing court is not allowed to take into account the amount or value or gratification received as one of the factors going towards harm, it may end up underestimating the severity of Offender H’s offence in respect of Company C.

250 I am alive to the fact that a sentencing court that rightly includes the value or amount of gratification given or received as a factor going towards both harm and culpability may well be accused of double counting the same factor. In my view, the sentencing court can avoid such a criticism by stating exactly the relevance of the amount or value of the gratification to the harm caused by the offence and the culpability of the offender when evaluating these two distinct independent variables of harm and culpability.

251 I pause to mention at this juncture that there is a tremendous amount of judicial discretion involved, and hence, much subjectivity not only in identifying which are the relevant factors (some of which are enumerated in the table above at [239]) but also in weighing each of these varied and numerous factors identified to derive the appropriate level of harm or the level of culpability as inputs to the harm-culpability matrix in order to obtain the indicative starting sentence. One court may assess a particular set of facts to give rise to a level of harm falling somewhere in the higher end of the moderate section of the harm scale, whereas another court may assess the same set of facts to give rise to a level of harm falling somewhere in the middle of the severe segment of the harm scale. With such disparate inputs to the harm-culpability matrix, the output or outcome from the matrix, however well-designed, will show widely differing indicative starting sentences, which will invariably lead to widely different final sentences on the same facts. I fully acknowledge this problem. This problem will persist until at some future point of time, a “scoring system” of sorts (alluded to at [139]–[142]) is designed (as a possible future refinement to the matrix methodology) that can assist to determine more precisely and less subjectively the indicative level of harm or culpability from the numerous and varied offence-specific factors set out in the table above at [239].

252 At the same time, I also recognise that it is difficult (but not impossible) to design a “scoring system” to quantify broadly the various offence-specific factors going to harm and culpability by assigning weights and points to each factor, and then have an aggregate score to depict where along the spectrum or scale of harm and culpability a particular factual situation will fall into (eg, whether a particular degree of harm is to be pegged at the high end of moderate harm, or at the low end of severe harm in the harm continuum).

253 In the absence of such a “scoring system”, the assessment will remain largely a generally holistic and broad-brush assessment of the degree of harm and culpability by reference to all the relevant offence-specific factors set out in the table above at [239].

254 After the court identifies and assesses the relevant offence-specific factors to ascertain the specific degree of severity for the harm and culpability parameters to be applied as pinpoint inputs to the framework (Step 1), the framework will throw up a definitive indicative starting sentence as a pinpoint output for the court to consider (Step 2). That is the purpose of the framework: to provide a pinpoint output in the form of an indicative starting sentence after the court provides pinpoint inputs to the framework in the form of precise levels for the harm and culpability parameters in a Double Variable Framework. This is an ideal situation. More often than not, the court provides a set of range inputs to the framework due to: (a) uncertainties in its assessment of the twin parameters of harm and culpability; and/or (b) its inability to be more precise in its evaluation of the weights for all the numerous offence-specific factors relevant to the given factual matrix of the case when determining the severity of the harm and culpability parameters. In a situation where the court can only provide a set of range inputs to the framework (ie, a certain assessed range for the severity of the harm parameter and a certain assessed range for the severity of the culpability parameter), the framework will inevitably throw up a range output for the indicative starting sentence for the court to consider. A range input begets a range output. In such a situation, the court must exercise its discretion to select the most appropriate indicative starting sentence from the range of indicative starting sentences thrown up by the framework (Step 3).

255 This section of the judgment is only concerned with explaining how to arrive at an indicative starting sentence via a method which adheres to the general principles identified above at [147]–[172]. To this end, I have crafted my own modified matrix (“the Modified Harm-Culpability Matrix”) which is directly based on the Prosecution’s harm-culpability matrix. The Modified Harm-Culpability Matrix is then translated into a more simplified and user-friendly form which I term the “Contour Matrix”.

256 Although they look different, the Contour Matrix and the Modified Harm-Culpability Matrix are essentially one and the same. They share the same content (ie, indicative sentencing ranges and indicative starting sentences), are based on the same underlying principles and apply in much the same way. In particular, the matrices both accord equal weight to harm and culpability and are used to determine the indicative starting sentence per charge for an offender who claims trial to purely private sector corruption offences under ss 6(a) and 6(b) of the PCA. Hence, there is (and should be) no practical difference whether the Contour Matrix or Modified Harm-Culpability Matrix is applied in any given case.

257 The following sections are meant to be a quick and easy guide to Steps 2 and 3 of the Modified Framework. For an explanation of how I have derived and finalised the contents of the Modified Harm-Culpability Matrix and the Contour Matrix, see Annex A. For an explanation of how the matrices are capable of accommodating uncertainty in the sentencing analysis, see Annex B.

258 I recognise that the Modified Harm-Culpability Matrix may seem foreign to readers accustomed to the traditional harm-culpability matrix. As such, I will first produce a simplified version of the Modified Harm-Culpability Matrix, which I term the “simplified matrix”. This simplified matrix is essentially the same as the Modified Harm-Culpability Matrix. I begin with it as it bears the most resemblance to traditional harm-culpability matrices and allows me to explain the key features of the Modified Harm-Culpability Matrix and, subsequently, how it differs from the Prosecution’s harm-culpability matrix.

259 The simplified matrix mirrors the harm-culpability matrix proposed by the Prosecution, except that: (a) the indicative sentencing ranges specifically use the words “Up to” and “and FINE” to account for the possibility of a combination sentence under ss 6(a) and 6(b) of the PCA (to meet the concern expressed at [213] above); (b) it includes a “Towards Infinity” section to account for extreme levels of harm and culpability (to meet the concern expressed at [209] above); and (c) the indicative sentencing ranges in some boxes overlap with other boxes to some extent. Focussing on (c), this is necessary in order to prevent cliffs and discontinuities in the indicative sentencing ranges (see [222]–[235] above).

260 The simplified matrix is used in the same way as a normal harm-culpability matrix. That being said, when applying this simplified matrix, one must bear in mind that: (a) harm and culpability are continuous independent variables; (b) the labels of slight, moderate, severe harm; or low, medium, high culpability, are merely very loose labels delineating the differing degrees of harm and culpability respectively; and (c) every single point in the matrix represents a single combination of a certain level of harm and culpability – in other words, one must be spatially aware when using the simplified matrix, which comprise within it an infinite number of combinations of infinite degrees of severity for both the harm and culpability parameters in a Double Variable Framework.

261 The simplified matrix contains the same indicative sentencing ranges and applies in the same manner as the Modified Harm-Culpability Matrix, except that it omits a number of features. My explanation on how to derive an indicative starting sentence from the Modified Harm-Culpability Matrix (and, to a certain extent, the Contour Matrix), can be applied to the simplified matrix.

262 Having set out some of the key features by way of the simplified matrix, I will now introduce the main features of the Modified Harm-Culpability Matrix before explaining how it can be simplified and transformed into the Contour Matrix. After that, I will provide a step-by-step guide to using both matrices.

263 The Modified Harm-Culpability Matrix is as follows:

264 The core features of the Modified Harm-Culpability Matrix are as follows:

(a) It is subdivided into four broad bands of “Slight” or “Low”, “Moderate” or “Medium”, “Severe” or “High” and “Infinity” to mark out generally the various continuously increasing levels of harm and culpability. The gridlines in the matrix are dotted lines (as opposed to solid lines) to reflect the understanding that the broad bands are merely labels for the differing levels of harm and culpability.

(b) The indicative sentencing ranges for each box is represented by the words in red. They represent the range of indicative sentences for all possible combinations of differing severities of harm and culpability within their respective boxes.

(c) Harm and culpability (as determined by the various offence-specific factors) are regarded as two continuous independent variables that separately and independently affect the indicative sentence. Much like a traditional harm-culpability matrix, an increase in culpability is represented by an upward vertical movement along the canvas of the matrix, while an increase in harm is represented by a rightward horizontal shift along the canvas of the matrix.

(d) The extreme top and right of the matrix corresponding to “Towards Infinity” harm and culpability are labelled “Up to Max” to show that there is an indicative sentencing range of up to the maximum statutorily prescribed sentence if the offence committed is exceptionally egregious. In other words, if the offence is extremely egregious in terms of culpability, the indicative sentence can theoretically reach the statutorily mandated maximum sentence even if harm is low, and vice versa. I should add that if culpability and harm are both zero, no offence may be disclosed in the first place. There would thus be no sentence. Unlike the other nine boxes in the matrix, the portions corresponding to infinity are shaded orange to show that only exceptional cases will fall within those areas of the matrix.

(e) The slight harm, low culpability box is the only box in the matrix which has an indicative sentencing range of fines only. This is to reflect the general position that many purely private sector corruption cases that do not present with strong aggravating factors (eg, corruption of a foreign public official, involvement of a strategic industry, or exceptional amounts of corrupt gratification given or received) may ordinarily be dealt with via the imposition of a fine. The word “FINE” (which is used in the matrix to refer to a fine from $0 to $100,000) is not used in that box because it is not possible to have a fine of $0.

(f) Apart from the slight harm, low culpability box, all other boxes include the words “and FINE” to reflect Parliament’s intention that a sentencing court may impose a custodial sentence only or a custodial sentence plus a fine for an offence under ss 6(a) and 6(b) of the PCA.

(g) The black dots in the Modified Harm-Culpability Matrix reflect single points within the matrix, ie, combinations of different degrees of harm and culpability. A single point anywhere within the matrix represents one specific combination of a certain degree of harm and culpability. The labels attached to the individual black dots reflect the indicative sentence at that specific point. The dots are placed strategically at the top-most end, middle and bottom-most end of each box (ie, highest, middle and lowest combination of harm and culpability that falls within each box) to provide guidance as to the indicative starting sentences at each of those key points.

(h) The yellow line represents the custodial threshold. It coincides with the upper and right-most tip of the slight harm, low culpability box.

(i) The light blue section reflects where cases falling within Romel categories 2 and 3 will generally fall into, while the dark blue sections reflect where cases falling within Romel category 1 will generally fall.

(j) When the dots having the identical indicative starting sentences are joined together by a continuous line, one gets the lines of equal indicative starting sentences that look like the contour lines of equal elevation in a topographic map, ie, Elevation Lines of Equal Sentences (see [186]–[188] above). It is clear that as these lines of equal indicative starting sentences are situated further and further away from the bottom-leftmost corner of the blue portion of the harm-culpability matrix, the lines will represent higher and higher indicative starting sentences. For a particular point (representing a combination of a specific degree of harm and culpability being subjectively determined based on the facts) that happens to fall somewhere in between two contour lines of equal indicative sentences, a rough and ready visual approximation will provide the indicative starting sentence for that intermediate point.

265 This matrix provides not only an indicative sentencing range within each box if the matrix were to be divided into nine boxes, but also a specific and more definitive indicative starting sentence if the court is able to determine more precisely where the specific degree of harm and culpability lie along the respective harm and culpability axis based on the factual matrix of the case in question and an assessment of the relevant offence-specific factors.

266 If a sentencing court is only able to say that the harm and culpability fall within a range and is unable to concretely determine the harm and culpability with any greater precision, then the same matrix will be able to provide a possible range of indicative starting sentences. For a detailed explanation of how to deal with such uncertainty, see Annex B.

267 The Modified Harm-Culpability Matrix is primarily geared towards helping a sentencing court determine the indicative starting imprisonment sentence. It does not provide guidance as to the appropriate indicative quantum of fine to be levied on an offender except to say that at the top- and rightmost tip of the severe harm, high culpability box, the indicative sentence is the maximum imprisonment term of 60 months’ and a $100,000 fine. The absence of specific guidance is because of the difficulty recognised at [216] above. The quantum of fines to be levied in addition to or in place of an imprisonment term (or some portion of it) is thus left to the sentencing court.

268 It is important to explain the concept of the custodial threshold in this specific context lest the Modified Harm-Culpability matrix be misunderstood as reserving the entire sentencing spectrum of fines up to $100,000 for a small segment of offenders.

269 In the context of offences for which the sentences of fine and imprisonment are mutually exclusive, a custodial threshold refers to a situation where the court will no longer impose a fine because of the severity of the offence but will impose a custodial sentence only as per the statute. However, under s 6 of the PCA, Parliament has provided for the possibility of a custodial term, a fine and a combination sentence, ie, a sentence of imprisonment with an added fine.

270 As such, the concept of a custodial threshold is slightly trickier. It is trite that fines can also have punitive and deterrent effects on offenders. Furthermore, a sentence of a fine may be as punitive as a sentence of imprisonment, depending on the quantum of the fine and the length of imprisonment in default of payment of the fine – for example, a fine of $100,000 (in-default 1 months’ imprisonment) may be regarded as being more punitive than 2 days’ imprisonment. In a similar vein, a sentence of $100,000 (in-default 1 months’ imprisonment) and 2 days’ imprisonment may be regarded as being more punitive than 4 days’ imprisonment alone. This means that if one were to line up all possible sentences under s 6 of the PCA in ascending order of severity, there may be situations in which a sentence of a very high fine with a high in-default sentence is much more severe (and hence, more suitable to an offender who has committed a more serious offence) than a pure low sentence of imprisonment. In the absence of a concretely defined correlation between the severity of imprisonment terms and fines with in-default sentences (eg, that $10,000 fine in-default 3 days’ imprisonment is equivalent to 1 day’s imprisonment, or $100,000 fine in-default 30 days’ imprisonment is equivalent to 10 days’ imprisonment), it is not possible to fully determine what this exact line up of sentence severity will be.

271 As mentioned previously, the Modified Harm-Culpability Matrix is primarily focussed on determining the appropriate indicative imprisonment sentence for an offender. While it does not need to be able to help courts determine the appropriate fine for an offender (eg, $14,500, $17,500 or $95,000), it cannot completely ignore and exclude the possibility of a fine. Doing so would mean that it would fall foul of the Completeness principle.

272 As such, it is necessary to rethink what a “custodial threshold” means. In my view, a custodial threshold in the context of cases where a combination of a fine and imprisonment term is a possibility, refers to an inflection point after which the severity of the offence mandates that the indicative starting sentence can no longer be a fine (not even of $100,000) alone. It must, at the minimum, include also an additional imprisonment term.

273 Correspondingly, for areas below the custodial threshold, the sentencing court will impose a fine most of the time. However, the sentencing court is not precluded from imposing a short custodial sentence in place of a fine (eg, if it is clear that the offender will be unable to pay the fine) such as four days’ imprisonment. It is also not precluded from imposing a very short custodial sentence and a smaller fine such as three days’ imprisonment and a $10,000 fine in-default 2 days’ imprisonment.

274 For a fuller explanation of various aspects of the Modified Harm-Culpability Matrix, as well as how I have derived and backtested its contents, see Annex A.

275 It will be much easier to understand and use the Modified Harm-Culpability Matrix when one completely ignores the indicative sentencing range stipulated for each box and simply concentrates on the individual indicative starting sentences for all the dots placed at various strategic points in the matrix. One can then interpolate between the indicative starting sentences stipulated for each of these dots (or the nearest adjacent dots) to obtain the indicative starting sentence if the case in question (with its own level of harm and culpability) falls somewhere between these dots. I can further simplify the Modified Harm-Culpability Matrix by removing the indicative sentencing ranges (in red) in each of the boxes and joining up the points of equal indicative sentences to derive the contour lines of equal sentence in its place (akin to the contours of elevation in a topographical map). I refer to these as contour lines. The resultant matrix is the Contour Matrix.

276 As can be seen below, the Contour Matrix retains the main features of the Modified Harm-Culpability Matrix (see above at [264]). Contour lines are represented by continuous non-dotted lines on the Contour Matrix. The indicative starting sentence for all points on the same contour line is denoted by the words in black (eg, 44 months and FINE). There are thicker and thinner contour lines. The thicker lines represent contour lines derived by joining up the dots at the top-most, middle and bottom-most tips of each box in the Modified Harm-Culpability Matrix. The thinner lines are defined via interpolation, ie, they are drawn at the midpoint between two thicker lines. Correspondingly the indicative starting sentence attached to each thinner line is the mid-point indicative starting sentence between the two thicker lines. For example, between “12 months and FINE” and “28 months and FINE”, the mid-point indicative starting sentence is “20 months and FINE”. The word “FINE” again refers to a fine from $0 to $100,000.

277 The process of interpolation becomes even clearer should a case in question fall in between these diagonal contour lines. These contour lines show increasing indicative starting sentences as the diagonal lines are situated further and further away from the bottom-leftmost point of the matrix, as they obviously must. The indicative starting sentences in this model are distributed such that they increase smoothly with corresponding increases in both culpability and harm. Even if the harm at any level is kept constant, and only culpability increases (or vice versa), the indicative starting sentences can be seen also to increase smoothly. There are no gaps and discontinuities when one moves in any direction in this model. When one moves along a contour line of equal indicative starting sentence, one actually finds that one independent variable (ie, culpability or harm) increases while the other independent variable decreases, and with equal weightage ascribed to both independent variables, there will be no net change in overall criminal culpability thereby resulting in no change in the indicative starting sentence. This again makes logical sense.

278 I now summarise how the Contour Matrix and Modified Harm-Culpability Matrix are used as part of the Modified Framework in Steps 2 and 3. First, after identifying and assessing the offence-specific factors going to harm and culpability at Step 1, the sentencing judge will broadly determine where the specific level of harm caused by the offender and where the specific level of his culpability lie along the respective spectrum of harm and culpability in the Modified Framework (eg, moderate harm and medium culpability or very, very, high harm and low culpability). If there is uncertainty as to the offender’s level of harm or culpability in the sentencing judge’s assessment of the offence-specific factors, the sentencing judge will have to determine what is the appropriate range of the levels of harm and culpability (eg, between slight to moderate harm, and between medium to high culpability). When these range inputs are applied, the Modified Framework will throw up an output comprising a range of values for the indicative sentence (ie, an indicative sentencing range). For more details on this, see Annex B.

279 Second, the sentencing judge will have to exercise his or her discretion to select an appropriate indicative starting sentence from the indicative sentencing range result thrown up by the use of the Modified Framework (this is Step 2 of the Modified Framework):

(a) The Modified Harm-Culpability Matrix: the indicative sentencing range refers to the red words within each box of the matrix (eg, up to 28 months and FINE).

(b) The Contour Matrix: the indicative sentencing range refers to the indicative sentences prescribed with reference to the contour lines within which the broad levels of harm and culpability fall. The indicative sentencing range (as an output from the Contour Matrix) will be narrower or broader depending on how precisely or imprecisely the sentencing judge has been able to assess the levels of the harm and culpability parameters (as inputs into the Contour Matrix).

(i) For example, if the sentencing judge finds that harm is very broadly in the moderate range and culpability is very broadly in the medium range, the two contour lines flanking these levels are those corresponding to “Custodial Threshold” at the bottom left corner of the middle box and “up to 28 months and FINE” at the top right corner of the middle box. The indicative sentencing range is thus up to 28 months’ imprisonment plus the option of imposing a fine up to $100,000.

(ii) Similarly, if the sentencing judge finds that the level of harm is on upper end of moderate and culpability is on the lower end of medium, then the two contour lines flanking these levels are those corresponding to “6 months and FINE” and “20 months and FINE”. The indicative sentencing range is thus “6 to 20 months’ imprisonment and FINE”.

280 Third, in the exercise of his sentencing discretion, the sentencing judge should make an assessment as to where exactly the offender’s offence lies within the area of the matrix.

281 Fourth, after determining the offender’s exact position (per charge) within each box, the sentencing judge can then arrive at the appropriate indicative starting sentence by interpolating from the nearest points or lines labelled with an indicative starting sentence (“strategic points”). This is Step 3 of the Modified Framework.

(a) The Modified Harm-Culpability Matrix: the strategic points are the dots within the matrix. The sentencing judge should pick the dot that is nearest to the specific point on which the offender falls. The indicative starting sentence of the offender will be similar to that of the dot nearest to the offender’s point. For example, if the nearest dot is the one in the middle of the moderate harm, moderate culpability box, the nearest indicative starting sentence is 12 months’ imprisonment plus fine up to $100,000. If the offender’s point is leftwards and downwards from the point (ie, lower in terms of harm and culpability), then the appropriate indicative sentence is somewhere close to but below 12 months’ imprisonment plus fine up to $100,000. If the offender’s point is rightwards and upwards from the point (ie, higher in terms of harm and culpability), then the appropriate indicative sentence is somewhere close to, but above 12 months’ imprisonment plus fine up to $100,000.

(b) The Contour Matrix: the strategic points are not “points” per se but rather, contour lines. The sentencing judge should identify the contour line that is closest to the specific point on which the offender falls. The indicative starting sentence of the offender will be similar to that of the line nearest to the offender’s point. For example, if the nearest contour line is the one labelled “12 mths and FINE”, the nearest indicative sentence is 12 months’ imprisonment plus the option of imposing a fine up to $100,000. If the offender’s point is leftwards and downwards from the contour line (ie, lower in terms of harm and culpability), then the appropriate indicative starting sentence is somewhere close to but below 12 months’ imprisonment plus fine up to $100,000. If the offender’s point is rightwards and upwards from the point (ie, higher in terms of harm caused and culpability), then the appropriate indicative sentence is somewhere close to, but above 12 months’ imprisonment plus fine up to $100,000.

282 Fifth, after determining the indicative starting sentence for the imprisonment term, the sentencing judge can then consider if it is appropriate to impose in addition a fine up to $100,000, and if so, what the quantum should be. This is also part of Step 3 of the Modified Framework.

283 As can be seen, this is materially similar to how a traditional harm-culpability matrix is used. The main difference is in Step 3 of the Modified Framework which provides specific guidance to a sentencing court as to how the indicative starting sentence is derived from the matrix. The key to this is spatial awareness and an understanding that every single point in the matrix represents a particular combination of the different degrees of severity or values of the respective harm and culpability parameters, as the two independent variables in this Double Variable Framework.

284 At Step 4, the court considers the offender-specific factors which do not directly relate to the commission of the offence in question and are generally applicable across all criminal offences. As stated in Wong Chee Meng ([10] supra), the following non-exhaustive considerations will be relevant at this stage of the analysis:

285 These factors are not exhaustive of the considerations that might be relevant to sentencing for an offence committed under ss 6(a) and 6(b) of the PCA.

286 Finally, after determining the sentences for each charge, the sentencing court must have regard to the totality principle in determining the global sentence for the offender.

287 I turn to apply this framework to the 28 charges facing each appellant.

288 There are two main offence-specific factors that go towards harm.

289 First, significant harm was caused to Koh who was unable to extricate himself and Chia Lee from the corrupt profit-sharing arrangement. ¹⁰⁶

(a) For six long years, Koh (and by extension Chia Lee) was forced to partake in an arrangement which did not make commercial sense. When the parties first entered into the arrangement, the profit from the sale of industrial flour was US$23 per metric ton, but Koh’s share was only US$3 per metric ton (ie, less than 1/7 of the total profits) even though Chia Lee was responsible for all the administrative work involved in the sale. Furthermore, Koh’s share of the profits remained constant at US$3 per metric ton even though profits from the industrial flour business increased to US$40, US$50 and even US$60 per metric ton. The reality of the arrangement was that as the business flourished, Koh and Chia Lee would be handling ever larger quantities of administrative work for the same meagre remuneration.

(b) The profit-sharing arrangement exposed Chia Lee to the risk of ever-increasing tax liability as the business flourished. Koh was well aware that Chia Lee’s tax liability for the industrial flour would grow as the business flourished. ¹⁰⁷ Koh also knew that the constant “profit” of US$3 per metric ton would be insufficient to cover the increased tax liability after a certain point and would thus result in losses. While there is no proof that Chia Lee had ever paid an increased tax liability as a result of the arrangement, ¹⁰⁸ the fact remains that as the industrial flour business flourished, Koh’s actions would have exposed Chia Lee to the risk of ever-increasing tax liability (and an even higher financial penalty if it had been caught evading taxes), while the so-called “profit” of a meagre US$3 per metric ton to Koh and Chia Lee remained constant.

(c) Crucially, Koh’s financial situation became dire by June 2005. This was made worse by the continual extraction of bribes from him, namely the bribes corresponding to charges 19 to 28. ¹⁰⁹ The financial situation deteriorated to the point that Koh stopped calculating the appellants’ share of the profits from the industrial flour business, and simply paid them whatever he could afford. ¹¹⁰

290 Second, harm was also caused to the appellant’s principal. Sojitz Japan brought a civil suit against the appellants in respect of the damage suffered by the companies due to the appellants’ corrupt conduct and was awarded judgment against them. The Japanese judgment was registered with the Singapore High Court in 2014 by Sojitz Singapore. ¹¹¹ The appellants were held to be jointly and severally liable to pay a total sum of approximately $875,248.51, ¹¹² but they eventually reached an agreement with Sojitz Singapore to pay a sum of $200,000 as full and final settlement of the Japanese judgment debt. ¹¹³ The judgment sum of $875,248.51 is a good gauge of the harm that was caused to the appellants’ principal.

291 That said, while the corrupt conduct of the appellants has caused loss to the Japan Company and the Singapore Company, there is no evidence that they suffered any additional loss by virtue of the fact that Chia Lee remained the sole distributor of edible flour. Chia Lee had been the sole distributor of edible flour for Nippon Flour Mills since 1978, ¹¹⁴ and was held in high regard by the Japan Company, Singapore Company and Nippon Flour Mill. ¹¹⁵ It continued working with Nippon Flour Mills even after the termination of the appellants’ employment on 26 February 2010 ¹¹⁶ – in fact, in December 2013, Chia Lee was still trading in industrial flour. ¹¹⁷

292 In my judgment, the aggregate harm is at the lower end of severe (ie, when viewed cumulatively for the purpose of the aggregate or total sentence). However, for the purposes of calibrating the individual charges, the harm accruing from the individual instances of corruption (each involving much smaller amounts of bribe when compared to the aggregate amount of bribes) is perhaps between the low and the middle portions of the moderate level depending on the individual amounts of bribe for each charge. The bribe quantum is relevant to harm because it represents the amount of money that Chia Lee and Koh earned rightfully from their labour in selling industrial flour, but which was instead diverted to the appellants.

293 There are four main factors that go towards culpability.

294 First is the amount of gratification involved in each charge. The amounts involved ranged from $13,750 to $117,523. The amount of gratification involved is an important factor in determining the proper sentences for corruption offences (see Heng Tze Yong ([79(c)] supra) at [25]). It is also important to appreciate that the appellants were capable of accumulating such a large sum of bribes because their offending was over a long and sustained period. Even though the charges relate to payments made between February 2004 and November 2007, the reality is that the corrupt arrangement began in 2002 and only ended when Sojitz Japan took over Chia Lee’s account. ¹¹⁸ Gratification is relevant to culpability because it reflects the appellants’ greed and significant financing profit from the profit-sharing arrangement.

295 Second, the offences were sophisticated and involved considerable planning and premeditation, which I amalgamate to constitute a single aggravating factor. From the outset, the appellants had deliberately targeted Chia Lee because they knew that that they had a hold over Koh in the form of his edible flour business. ¹¹⁹ To carry out the profit-sharing arrangement, the appellants schemed to get Chia Lee, a newcomer to the industrial flour business, appointed as the new industrial flour distributor to replace the financially struggling Sin Heng Chan. They succeeded in doing so even though Koh had no expertise in the industrial flour business, which operated in a markedly different fashion from the edible flour business. ¹²⁰ Furthermore, the appellants admitted that to ensure the profitability of the corrupt scheme, they found customers for Chia Lee and negotiated with them in relation to the industrial flour business. ¹²¹

296 Third, the appellants had threatened and coerced Koh into entering and remaining in the corrupt scheme. The consistent thread in Koh’s evidence is that he deeply treasured Chia Lee’s edible flour business. It represented the sum of his life’s work which he had painstakingly built up over more than 20 years. ¹²² Unfortunately, this was used as leverage by the appellants to request for “favours” (ie, for Koh to enter and remain in the profit-sharing arrangement) in return for them safeguarding Chia Lee’s position as the sole distributor of edible flour in Singapore. ¹²³ To ensure that Koh continued doing favours for them, the appellants made deliberate efforts to keep Koh “updated” on inquiries about the edible flour business at the Singapore Company, the Japan Company and even at Nippon Flour Mills, ie, potential competitors for the role of sole distributor of edible flour. ¹²⁴ By doing so, the appellants reinforced Koh’s impression that they were powerful ¹²⁵ and capable of determining his fate, ¹²⁶ such that he had no choice but to “help” them so as to guarantee the smooth-running of the edible flour business. ¹²⁷

297 The appellants’ conduct thus fell squarely within Romel category 3 as it involved the interference with Koh’s legitimate right to choose not to be a party to the profit-sharing arrangement. Koh was also faced with the threat of having his edible flour business compromised if he did not pay bribes. This is a significant aggravating factor.

298 Fourth, there was an abuse of position and a breach of trust. The appellants occupied senior roles in the Japan and Singapore Companies and were valued employees. Masui joined the company in 1987 while Ishibe joined in 1989 – both remained in the company until their termination in 2010. These were their first jobs and they were rewarded for their steadfastness with a steady stream of promotions across the years. ¹²⁸ A significant degree of trust had clearly been reposed in both of them, which they betrayed by conceiving of, and implementing the corrupt profit-sharing arrangement that continued for several years.

299 In my judgment, culpability as a whole is at the higher end of medium.

300 For a combination of culpability at the higher end of medium on the culpability scale, and harm falling between the low end to the middle of moderate on the harm scale, depending on the individual amounts of bribe for each charge (ranging from $13,750 to $117,523), the range of indicative starting sentences based on a broad interpolation from the Contour Matrix is between 6 to 14 months’ imprisonment plus a fine up to $100,000 for the individual charges. As I have already accounted for all the other offence-specific factors that are common across the charges, the only variable left to account for is the amount of gratification given and received. The yellow square in Figure 13 below represents the general position of the 28 charges within the Contour Matrix, and their corresponding indicative starting sentences.

301 Therefore, the relevant quadrant in question to determine the range of the indicative starting sentences is approximately in the region of the upper left quadrant of the moderate harm, medium culpability box.

302 Having arrived at an indicative sentencing range, I now provide more discrete ranges of indicative sentencing according to differing amounts of gratification given and received for each charge. In my judgment, the following indicative starting imprisonment sentences will be appropriate bearing in mind all the other offence-specific factors present on the facts of this case:

303 I am further of the view that a fine of approximately 10% of the bribe quantum received per charge would be appropriate. As these fines will also have a punitive and deterrent effect on the appellants, I find it appropriate to reduce the indicative starting imprisonment term per charge by one month. The indicative starting sentences, per charge, are therefore as follows:

304 Turning to Step 4, I agree with the Prosecution and the DJ that there are no offender-specific factors warranting a downward adjustment of the indicative sentences. Hence, the imprisonment sentence and fine that I will impose for each charge is as per the above table at [303].

305 First, contrary to the appellants’ assertions, ¹²⁹ there was no “voluntary disgorgement of the value of the gratification by paying Koh US$240,000, and Sojitz Japan and Sojitz Singapore S$200,000”.

306 Ishibe argues that the appellants’ payment of $200,000 is a mitigating factor as it constituted full and final settlement of Sojitz Japan’s claim against them. I do not agree. The sum of $200,000 was paid as a settlement of a much larger judgment sum which had already been successfully obtained against them by Sojitz Japan in its civil suit against the appellants. I agree with the Prosecution that having put their principal through a full trial and an appeal to recover these monies, this payment can hardly be said to be voluntary or evidence of any contrition or remorse on the appellants’ part.

307 The payment of US$240,000 to Koh should not be seen as voluntary restitution because it was made to facilitate the appellants’ corrupt scheme by keeping the financially struggling Chia Lee afloat so that the profit-sharing arrangement could continue. Furthermore, this was a payment to the giver of gratification who, apart from being a victim, was in fact a co-conspirator in multiple corruption offences. In any event, if mitigating weight is to be placed on restitution, this would have to be in the situation where it is made to a principal in respect of loss suffered as a result of the corruption offence (see Wong Chee Meng ([10] supra) at [74]). But in this case, the US$240,000 payment was not voluntarily paid to the principal for the loss suffered and hence, I do not ascribe any mitigating weight to this payment to the giver of the gratification, a co-conspirator in the corruption offence.

308 Second, there is no credible evidence that the appellants were remorseful for their actions or that they had cooperated with the authorities since investigations began in 2013. ¹³⁰ Only Masui raised this as an offender-specific factor that warranted a downward adjustment of the indicative sentences. However, Masui’s assertions are unsupported by any credible evidence, and he had also admitted under cross-examination that he had lied to CPIB officers during the recording of his statements. ¹³¹ The DJ devoted a number of paragraphs to the appellants’ “mendacious nature” (Decision at [72]) and also observed that their “conduct during the trial showed a singular lack of remorse on their part” (Decision at [122]).

309 Third, I am unable to accept Masui’s assertions that “care had been taken to ensure that absolutely no harm would be caused to Sojitz Japan and Sojitz Singapore as a result of the profit-sharing arrangement”. ¹³² While I note that Masui had originally recommended that Nissho Singapore take over the role of industrial flour distributor, and had only recommended Chia Lee after Nissho Japan rejected his original suggestion, ¹³³ the appellants knew that appointing and retaining Chia Lee was not in the best interests of the Singapore Company.

310 The appellants admitted that from 2002 to 2007: (a) they knew that it would be in the best interests of the Singapore Company to have a company that was experienced and financially sound as the industrial flour distributor; and (b) they were aware of other companies in Singapore who were trading in industrial flour. Nonetheless, they never recommended any other company as the distributor for industrial flour, apart from Chia Lee. ¹³⁴ This is problematic because in 2002, Chia Lee was not a suitable replacement for financially unstable Sin Heng Chan as the new industrial flour distributor. When Ishibe first approached Koh in 2002, Koh had no expertise in the industrial flour business, which operated in a markedly different fashion from the edible flour business. ¹³⁵

311 Further, in 2005, the appellants were both aware of Chia Lee’s dire financial situation. Nonetheless, they chose not to alert either the Singapore or Japan Company about it, and took no steps to find a replacement for Chia Lee despite their admitted knowledge of other suitable candidates in the industrial flour industry. ¹³⁶ It should be borne in mind that Sin Heng Chan was removed from its old position in 2002 for the exact same reason.

312 Bearing in mind the foregoing, I am of the view that on the facts of this case, there are no offender-specific factors that warrant a downward adjustment of the indicative starting sentences.

313 I turn finally to Step 5 of the Modified Framework which entails making further adjustments to account for the totality principle (see Public Prosecutor v Raveen Balakrishan [2018] 5 SLR 799 at [73]).

314 The present appeals involve one of the most egregious cases of purely private sector corruption in Singapore, both in terms of the total quantum of bribes that the appellants had systematically extracted from Koh and the coercive tactics they had employed to do so. There is also a pressing public interest concern in discouraging the corrupt criminal conduct that has been displayed by the appellants (Mohamed Shouffee bin Adam v Public Prosecutor [2014] 2 SLR 998 at [80]). I am of the view that running only two sentences consecutively would be insufficient to encompass the overall criminality of the appellants’ conduct.

315 In my judgment, I agree with the DJ that it is appropriate to run the imprisonment sentences for four charges ie, the appellants’ individual charges corresponding to C2, C6, C25 and C27 (see the charges highlighted in green in the table at [303] above) consecutively. Further, the fines for each charge are cumulative. The total sentence is thus 43 months’ and 3 weeks’ imprisonment and a fine of $200,944 for each appellant.

316 For the avoidance of doubt, I add that in arriving at the total sentence for each appellant, I have given much thought to the overall sentence after imposing combination sentences for each of the 28 charges. I am also mindful of the possibility that the appellants may default on payment of their fines and the penalties to be imposed, and have thus calibrated the in-default imprisonment terms to 12 months’ imprisonment for both the fine and the penalty (see [341] below) such that the overall sentence will be proportionate to the appellants’ overall criminal conduct even in the event of a default (see Ho Sheng Yu Garreth v Public Prosecutor [2012] 2 SLR 375 at [127], cited with approval by Hoo J in Koh Jaw Hung v Public Prosecutor [2018] SGHC 251 at [57]–[58]).

317 It is trite that fines ought not to be imposed if it is unambiguously clear that the offender is unable to pay the fine and will have to serve the imprisonment term in default of payment of the fine (Low Meng Chay v Public Prosecutor [1993] 1 SLR(R) 46 at [13]). Masui and Ishibe’s respective counsel have not pointed me to any evidence which shows that they will not be able to afford the fines. As alluded to previously (at [303] above), if no fines had been imposed in addition to the imprisonment terms for the various offences, I would not have given a one-month reduction in the imprisonment terms. Imposition of fines in addition to the imprisonment terms is therefore not disadvantageous to the appellants.

318 The penultimate issue in these appeals concerns the appropriate penalty order under s 13(1) of the PCA and much ink has been spilled on this. The DJ had ordered each of the appellants to pay a penalty of $1,025,701 (in-default 6 months’ imprisonment), being their respective halves of the total sum of gratification received, as calculated by the DJ (Decision at [132]). The appellants appealed against that order. Broadly speaking, both appellants agree with the DJ that their penalty orders ought to be based on the amount of gratification received. However, they disagree as to the exact quantum of their respective penalty orders.

319 Ishibe argues that he received only US$50,000 from Masui and the penalty order ought to be revised downwards as it is “not meant to be punitive, but simply to disgorge what one personally received from a corrupt transaction”. Ishibe’s position on the appropriate quantum of the penalty order is not entirely clear; at the conclusion of his first set of submissions filed 16 July 2019, he concludes by stating that he received “at best, slightly over S$1 million”, but in his final set of submissions filed on 11 August 2020, he concludes by saying that it is “at best, US$50,000”. ¹³⁷

320 I reject any suggestion that Ishibe should only be subject to a penalty order of US$50,000 in the light of my earlier finding that the appellants shared the bribe moneys equally at [63] above. I turn to Ishibe’s original argument which is aligned with Masui’s position on the appropriate penalty order.

321 Citing the case of Public Prosecutor v Marzuki bin Ahmad and another appeal [2014] 4 SLR 623 (“Marzuki”), the appellants argue that the purpose of imposing a penalty equal to the gratification under s 13 of the PCA is to ensure that offenders do not retain their ill-gotten gains, rather than to impose an additional punitive measure: Marzuki at [61], [62] and [71]. ¹³⁸

322 Masui submits that the following passage from Marzuki is instructive in providing guidance as to how his penalty sum ought to be calculated (at [71]): ¹³⁹

323 On this basis, the appellants argue that the following sums ought to be deducted from the total gratification amount of $2,009,433, and, therefore, from the penalty orders on each appellant for half of that sum, ie, $1,004,716.50 ¹⁴⁰

(a) First, the sum of $200,000 paid to the Sojitz Japan in full and final settlement of the judgment sum awarded by the Japanese civil court against the appellants ought to be deducted as this was disgorged from them. Further, there is a clear nexus between the Japanese judgment and the present offences given that the judgment was concerned with their employment with the Japan Company and their eventual termination due to the breach of the company rules. ¹⁴¹ This settlement sum was borne equally by both appellants.

(b) Second, the sum of US$240,000 ought to be deducted as this was returned to Koh in June 2005. This means that it had been “disgorged” and the appellants could no longer be said to have “retained or benefitted from this sum”. ¹⁴² The appellants contributed equally to this sum.

(c) The sums of S$33,322.20 and US$138,152.48 (ie, S$171,309.07) that were in Masui’s frozen bank accounts. This gratification had been recovered by the authorities and thus disgorged as well. ¹⁴³

324 In my judgment, none of the above sums should be deducted from the appellants’ penalty orders (ie, $1,004,716.50 each). It is important to note that Marzuki was primarily concerned with a case where the bribes took the form of loans, as opposed to outright gifts of sums of money. Let me explain.

325 I begin by reproducing s 13 of the PCA:

326 The definition of “gratification” under s 2 of the PCA is a broad one and includes loans, gifts of property, services, money, etc. A plain reading of s 13 of the PCA reveals that Parliament drew a clear distinction between cases where the “gratification is a sum of money” (ie, an outright gift of money) and cases where it is not (eg, a favourable loan or revealing insider information about a particular trade). If the gratification is a sum of money, the statute states that a court “shall … order him to pay as a penalty … a sum which is equal to the amount of the gratification” [emphasis added]. In my view, the italicised words mean that once a sentencing court finds that a convicted offender accepted a sum of money as gratification, the court must impose a penalty sum equal to the amount of gratification. It is only if a court finds that the gratification did not take the form of a sum of money (eg, a loan or a service), that the court looks into the question of whether the “value of the gratification can be assessed”, and is subsequently given the limited discretion to determine the value of that gratification. I refer to this as the Plain Interpretation of s 13(1) of the PCA and I adopt it for three reasons.

327 First, the Plain Interpretation accords with common sense. There is no need for Parliament to give the court a discretion as to the value of the gratification if it is an outright gift of money as there can only be one correct value. The value of $1,004,716.50 of money is $1,004,716.50. The court is only given a limited discretion to determine the value of the gratification if there may be multiple acceptable ways to value it. For example, assuming that the gratification is a loan, it is the recipient’s ability to use that money for a period of time that constitutes the gratification rather than the actual sum loaned (Marzuki ([321] supra) at [60]). A number of factors will affect the valuation of the gratification such as whether it was repaid, when it was repaid, the choice of interest rates, etc. Further, by virtue of the mandatory word “shall”, the court is required to impose a penalty upon an offender that “shall be recoverable as a fine”, so long as it is able to quantify the value of the gratification.

328 Second, the Plain Interpretation is, in effect, an application of the well-established canon of construction: the reddendo singula singulis principle. This concerns the use of words distributively and requires that (Diggory Bailey and Luke Norbury, Bennion on Statutory Interpretation (LexisNexis, 7th Ed, 2017) at para 23.10):

329 This principle has been applied by the High Court in Jeyaretnam Kenneth Andrew v Attorney-General [2013] 1 SLR 619 to interpret Art 144(1) of the Constitution at [29]–[30]. Its approach and the resultant interpretation of Art 144(1) was wholly endorsed by the Court of Appeal in Jeyaretnam Kenneth Andrew v Attorney-General [2014] 1 SLR 345 at [11]. The Plain Interpretation is therefore supported by the express language of the statutory provision. If it was intended that the court should have a discretion to account for monies returned or otherwise disgorged, that could have been easily legislated for. The decision not to do so is, in my view, clear support for the Plain Interpretation.

330 Third, the Plain Interpretation accords with Parliament’s intention. Section 13(1) of the PCA was first introduced into Singapore law by the Prevention of Corruption Ordinance 1960 (No 39 of 1960) (“PCO 1960”). Back then, s 13 of the PCO contained only s 13(1) of the PCA which has remained unchanged since 1960, except that the phrase “any provision of this Ordinance” has been replaced by “any provision of this Act”. Section 13(2) of the PCA was introduced later in 1981. At the second reading of the Prevention of Corruption Bill (No 63 of 1960), Mr Ong Pang Boon, then Minister for Home Affairs explained that (see Singapore Parliamentary Debates, Official Report (13 February 1960) vol 12 at col 380):

331 More than half a century later in 2013, Mr Teo Chee Hean, then Deputy Prime Minister and Minister for Home Affairs echoed those words in response to parliamentary questions about, inter alia, measures that the Ministry for Home Affairs would undertake to ensure that Singapore stays free of corruption offences relating to match-fixing. He stated that (see Singapore Parliamentary Debates, Official Report (21 October 2013) vol 90) :

332 The italicised words make clear that when Parliament enacted s 13 of the PCA, it envisaged that the penalty is to be equal to the amount of gratification received by the convicted offender. There is no mention of any deductions from the amount of money that has been taken as a bribe.

333 In my judgment, the dicta in Marzuki ([321] supra) ought to be read in its proper context, ie, as applying to cases where the gratification is not an outright gift of a sum of money. In other words, Marzuki was concerned with the second category of cases under s 13 of the PCA, where the court has to consider the proper value of the gratification received. The accused in Marzuki was an assistant property executive in Jurong Town Corporation (“JTC”) tasked to conduct periodic checks and inspections at premises leased out by JTC to ensure that the lessees complied with applicable local laws and regulations and the terms of their leases. He was obliged to report any such infringements. During the course of his inspections at foreign worker dormitories owned by one Chew Wee Kiang Allen (“Allen”), he discovered that Allen had not obtained certain approvals from the Urban Redevelopment Authority and the Singapore Civil Defence Force. The accused subsequently came to an understanding with Allen that the accused would forbear from reporting the compliance lapses that he had discovered in return for loans from Allen. The accused received $31,500 worth of loans and attempted to obtain a further loan of $5,000 over the course of a year. The accused pleaded guilty to six proceeded charges under s 6(a) of the PCA which collectively involved a total loan sum of $25,000. The DJ declined to impose a penalty order under s 13(2) of the PCA in respect of the charges that were taken into consideration for sentencing, and only imposed a penalty of $25,000 under s 13(1) of the PCA in respect of the proceeded charges. On appeal, Menon CJ reduced the penalty under s 13(1) of the PCA to $5,000 because the accused had actually repaid a sum of $20,000, and thus only retained $5,000: Marzuki at [84]. A penalty under s 13(2) of the PCA was also imposed to reflect the exact unrepaid loan sums for the remaining charges: Marzuki at [85]–[86]. In this sense, it can be said that the outcome of Marzuki was in line with the Plain Interpretation because the eventual penalty sum was equivalent to the sum of the unrepaid loans, after Menon CJ determined that the unrepaid loans were, in effect, gifts of money.

334 Although some statements in Marzuki could be read as being of general application to all cases under s 13 of the PCA, I take the view that they were concerned with cases where the gratification was not an outright gift of money. In particular, I understand the dicta in respect of how the penalty sum ought to be calculated in Marzuki at [71] (see above at [322]) as being directed at the fact-scenario at hand, ie, loans. This is because the core issue in the appeal was whether gratification which took the form of a loan of money should be treated differently from gratification that took the form of a gift of money (Marzuki at [48]), when both are technically gratification taking the form of money.

335 At [60], Menon CJ distinguished between two forms of money gratification. He stated that where money gratification was given as a gift, it would plainly be correct that the sum of money received was itself the gratification, and that value of the gratification was the amount of money received, however, this would not be so clear when the money gratification was given as an interest free loan or a very low interest loan, where the value of that loan gratification has to be ascertained. However, Menon CJ also recognised that the line between money gratification given as an outright gift and money gratification given as a loan was not always clear (Marzuki at [70]). As such, it was necessary to have regard to the purpose underlying s 13 to derive an answer to this question (Marzuki at [71]). This purpose was primarily to ensure that the recipient of the gratification concerned would not be in a position to retain the benefit of the gratification.

336 Returning to the facts of the present appeals, the sums received by the appellants are outright gifts of money which they were not expected or obliged to return. In such a situation (Marzuki at [72]):

I respectfully agree, and would also add that the same would apply where the recipient is not expected to return the money received to the principal or the victim, but does so anyway voluntarily out of a sense of guilt, in which case the sentence of fine and/or imprisonment may be adjusted downwards due to the mitigatory effect of that remorse shown during the second stage of sentencing when the offender-specific factors are considered. But that remorse shown has no effect on the amount of the penalty to be imposed because in my view, the legislative language and Parliamentary intent are sufficiently clear that if the gratification is money received, then the court has no discretion except to make a penalty order equivalent to the amount of the gratification, ie, the sum of money received.

337 As such, the appropriate penalty order for each appellant is half the total gratification sum of $$2,009,433. The sum of $1,004,716.50 was an outright gift of money, and s 13(1) of the PCA requires that a value equal to the sum of money is imposed as a penalty.

338 Part of the penalty will be paid out of the amounts that have been seized and/or frozen by CPIB, and this will include the frozen bank account moneys. As such, deducting the bank account moneys from the penalty order at this point will mean that they will be double counted.

339 I also agree with the DJ that the two remaining sums are rightly characterised as expenditure which the appellants had benefitted from. The payment of US$240,000 to Koh was not a repayment or return of bribe moneys to Koh, but rather a means of facilitating the continuation of a corrupt arrangement when Chia Lee faced cashflow problems. Further, the $200,000 was paid to settle a judgment sum that would otherwise have been enforced against the appellants by other means. There is no remorse involved and hence no mitigatory effect is to be considered in so far as these two sums are concerned.

340 The appropriate penalty for each appellant under s 13(1) of the PCA is therefore $1,004,716.50. I do not think that an in-default imprisonment term of six months’ imprisonment for this sum imposed by the DJ will be sufficient to deter the appellants from evading payment of this large sum. I have added the fine of $200,944 to the penalty sum of $1,004,716.50 and imposed a higher imprisonment term of 12 months in default of payment of the aggregate sum of $1,205,660.50.

341 Accordingly, the total sentence for each appellant is 43 months’ and 3 weeks’ imprisonment, a fine of $200,944 and a penalty of $1,004,716.50 (and in default of paying the fine and penalty totalling $1,205,660.50, 12 months’ imprisonment) for each appellant.

342 I arrive finally at the last issue. Given the fact that the appellants are being sentenced under a new sentencing framework, the question arises as to whether this is a case suitable for prospective overruling.

343 In Adri Anton Kalangie v Public Prosecutor [2018] 2 SLR 557 at [70], the court clarified the principles relevant in determining the applicability of the doctrine of prospective overruling:

344 Having regard to these principles, I do not think this is an “exceptional case” that warrants the invocation of the doctrine of prospective overruling. I arrive at this view primarily because both the Modified Framework and the content of the Modified Harm-Culpability Matrix and Contour Matrix are based on existing sentencing practice. The operation of the framework also adheres to the principles laid down in precedent cases. There is no “manifestly excessive departure from the sentencing precedents in previously decided cases” or any “drastic inconsistency in sentencing”. ¹⁴⁴

345 In conclusion, I allow the appellants’ appeals against sentence. The manifestly excessive threshold has been crossed. The sentence including the fine for each of the 28 charges is set out in the table above at [303]. I adopt the approach of the DJ and I order that the sentences for the charges C2, C6, C25 and C27 are to run consecutively and the sentences for the remaining 24 charges are to run concurrently. The appellants are each sentenced to an aggregate sentence of 43 months’ and 3 weeks’ imprisonment, a total fine of $200,944 and a penalty of $1,004,716.50 (and in default of paying the fine and penalty of a total of $1,205,660.50, to serve 12 months’ imprisonment).

346 I would also like to record my thanks to the learned Deputy Public Prosecutors, Ms Jasmine Kaur, Ms Loh Hui-min and Mr Lee Jing Yan, for their assistance in respect of these appeals. At the court’s request, they had undertaken additional and very tedious work in preparing the Table of Evidence and two Tables of Profits. This required them to trace the invoices from the Singapore Company and from Chia Lee, map out the relevant payments to the bank from the bank records and compile all the relevant testimonies of Koh and the appellants, in order to verify the accuracy of the amount of bribes received by the appellants.

347 I end off with a final reminder that sentencing is a fluid and difficult evaluative exercise with very real consequences for the offenders at hand. Sentencing guidelines and frameworks are not to be applied mechanistically and indiscriminately, but flexibly and discerningly. Careful consideration of where justice may lie in each case is of primary importance.