Case Update: Public Prosecutor v Chow Chian Yow Joseph Brian [2016] SGHC 18 – SGHC applies regression analysis to criminal sentencing

Public Prosecutor v Chow Chian Yow Joseph Brian [2016] SGHC 18

Significance: Singapore High Court judge cites John Stuart Mill’s “On Liberty”, applies regression analysis to plot graphs of benchmark sentences for a criminal offence, and lays down sentencing principles on offence of evading National Service (NS) without a valid exit permit under the Enlistment Act.


The respondent was absent from Singapore without a valid exit permit for a total of 6 years and 27 days, thus liable to a conviction under section 32(1), punishable under section 33(b), of the Enlistment Act (Cap. 93). In that time, he did his foundational course and university undergraduate studies in Australia.

Justice Chan Seng Onn begins his judgment by citing John Stuart Mill’s “On Liberty” to justify his view that “[e]ach citizen has a duty to do his fair share to sustain social arrangements from which all benefit. Parliament is to determine the nature and content of this duty, and the law is justified in using coercive power when necessary to ensure the performance of this duty”, what he calls the fair share argument: [1]-[2].

Chan J then considers that in sentencing, the court must first determine the circumstances in which a custodial sentence would be appropriate (“the custodial threshold”) to promote the criminal justice goal of general deterrence; the custodial threshold must be based on reviewing the relevant statutory provisions and their underlying policy: [20]-[21].

At [23], he cites case law which states that national security, universality and equity were the three fundamental principles underpinning the NS policy in Singapore.  As regards national security, it is for MINDEF to set when and how each eligible person has to serve his NS. As regards universality and equity, if a person can unilaterally decide to defer his NS, he can gain a real advantage over others: (i) he’d be lesser suited for a combat role as his age increases; (ii) less time available for serving post-operationally ready date (post-ORD) reservist obligations: [25]-[26].

Thus, Chan J says there should be a strong correlation between an offender’s culpability and the number of years he evaded NS by reason of being overseas without a VEP; the longer he is in default the greater is his violation of the principle of equity and universality: [27].

On the custodial threshold, Chan J looks to a ministerial statement made in Parliament in 2006 by Mr Teo Chee Hean, then Minister of Defence (“the Ministerial Statement”) as the policy underlying the offence in question: [29]-[32].

The judge notes that cases involving a short period of default of two years or less would generally not attract a custodial sentence although each case must ultimately be assessed on its own set of facts and circumstances: [32].

The court then accepted the submission of the recently promoted second Solicitor-General, Mr Kwek Mean Luck, submitting on behalf of the Prosecution that a distinction should be made between overseas defaulters who have a substantial connection to Singapore and those who do not have such a connection. The custodial threshold should not apply to the latter group. This is sensible in light of the fair share argument: [32]-[34]. Chan J then reviews some sentencing precedents on the former category: at [39]-[41].

Justice Chan then goes on a long explanation to justify what he’s about to do next, i.e. his methodological mathematical graphical plotting of benchmark sentences for such offences. His explanation is worth quoting in full (see [42]-[46]):

42 Development of benchmark sentences must have its roots in rigorous scenario analysis ie, a conditional analysis of a hypothetical offender under a wide selection of probable scenarios. A court needs to look at the entire available sentencing spectrum and determine whether and to what extent the sentences are to be spread out over this spectrum to account for varying degrees of culpability amongst offenders, ranging from the least culpable to the most culpable of them. Such analysis requires one to first identify the primary factor
that determines the length of a sentence in relation to the particular offence. The factor will then be used to plot the trend line/curve across the appropriate part of the sentencing spectrum. This is not to say that sentences are always to be determined in a two-dimensional manner. While I recognise that sentencing is
multi-dimensional, these dimensions can be accommodated subsequently by horizontal and vertical adjustments of a trend line/curve born out of a two-dimensional analysis.

43 In order for benchmark sentences to be of any practical utility, it is important to define the characteristics of a hypothetical/archetypal offender in relation to the offence. This allows the application of the benchmark as a general guide for the archetypal case and in a nuanced manner should unique circumstances be present in the case before the sentencing court.

44 The sentencing benchmark during its development should, where possible, be tested against existing precedent cases. If the existing cases do not fall within the benchmark, it falls upon the court setting the benchmark to reexamine and re-calibrate the benchmark to be set, if it is appropriate to do so, or explain the circumstances which require the benchmark to depart from the precedent(s), if they remain irreconcilable.

45 It must also be borne in mind that in a two-dimensional analysis, the degree of culpability of an offender at any point of the sentencing curve is represented by that point on the vertical y-axis in relation to the extent of the primary or most significant variable factor (ie, the period without a VEP) present in offender’s case as represented by that point on the horizontal x-axis. Therefore, it must also be decided if the sentence is to increase linearly across the spectrum for that variable factor or whether (as in the present case) the seriousness of the offence increases at an increasing rate for equal increments of that variable factor such that the gradient of the sentencing curve should be steeper over a certain part of the spectrum. If that is indeed the case, the gradient of the two-dimensional plot will change along the curve as one moves further
towards the right along the horizontal x-axis of the sentencing graph.

46 Lastly, I am of the view that it is pointless to debate whether sentencing is an art or a science. To achieve logical consistency and fairness in sentencing, the mental iterative and analytical processes to derive an appropriate sentence  for a set of facts and circumstances in relation to the commission of a particular offence cannot purely be an art or a science. A judicious exercise of discretion by combining both approaches is more likely to produce the most appropriate and fair sentence. Science should be used as a guide where possible to aid in the art of sentencing. In my view, it is short-sighted to avoid completely the use of any analytical tools and scientific aids if they are capable of playing a useful guiding role to assist in the process of reasoning that must take place when deriving an appropriate and fair sentence for a particular case having regard to
all the relevant facts and circumstances. Parliament sets out the sentencing range for a particular offence. The court has to exercise its sentencing discretion within that sentencing spectrum. Accordingly, a court should not shy away from envisaging various possible scenarios and analysing what should be the appropriate sentence for each of the various scenarios given the sentencing spectrum and then deciding how it should exercise its discretion to determine the sentence for a particular case that is broadly consonant within the whole possible spectrum. There must be a good deal of “reason” and “logic” in reasoning. Unless chance favours the judge, it is generally difficult to achieve parity, consistency and fairness when exercising the sentencing discretion without being assisted or guided by an analytical framework that is grounded in logic, especially when multiple factors come into play to affect the determination of the appropriate sentence.

Chan J then considers the factors he would determine to have correlation on the benchmark sentence:

(1) number of years the offender evaded NS without a Valid Exit Permit (VEP);

(2) whether the offender voluntarily surrendered to MINDEF or was arrested;

(3) whether the offender claimed trial or pleaded guilty;

(4) performance during full-time NS.

Then he described:

(i) a base case (a paradigm exemplar): began evading NS obligations from start point of 16 years 6 months old, arrested, pleaded guilty, unexceptional performance in NS;

(ii) a more culpable offender:began evading NS obligations from start point of 16 years 6 months old, arrested, claims trial, unexceptional performance in NS;

(iii) less culpable offender:began evading NS obligations from start point of 16 years 6 months old, voluntarily surrendered, pleaded guilty, unexceptional performance in NS.

Then he plots his graph and shows us his mathematical, analytical skills.


He goes on to explain the sentencing discount applied under the different scenarios. That’s not all. Further analytical steps have to be taken:

“65 … [the Court would then have] to inquire on the age when the offender commenced being overseas without a VEP (“offence commencement age”). The following steps will then flow from the aforementioned inquiry:

(a) If the offence commencement age is 16 years and six months, the court may look directly to the relevant sentencing curve in Graph 1 that would most closely apply to the offender to seek a guide on an appropriate sentence for an offender; and

(b) If the offence commencement age is after 16 years and six months, the court may still look to the relevant sentencing curve in Graph 1 that would most closely apply to the offender to seek a guide on the baseline for sentencing the offender. The court would then have to take his age into account to increase the baseline sentence, as an offence commencement age much later than 16 years and six months would for the same number of “years without VEP” (unless shown otherwise) result in the likelihood of lower physical fitness of the offender over the course of the entire NS cycle (including his post-ORD reservist obligations).

66 The court would then have to consider if there are any other aggravating or mitigating factors in order to calibrate an appropriate sentence.”

Then he sets out a further table on sentencing discount for exceptional performance during NS: [69]. Enough said.

In case you did not know, Justice Chan Seng Onn is one of those guys we envy for being very brilliant and quite possibly a polymath. He obtained his B.Sc in Engineering from the University College of London in 1976, his M.Sc in Industrial Engineering from the National University of Singapore in 1981, his LLB. from the National University of Singapore in 1986 and his LLM. from Cambridge University in 1987 He began his career in the Singapore Legal Service in 1987. He was appointed Judicial Commissioner in 1997, Solicitor-General in 2001, and High Court Judge in 2007.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.