Judicial Disparity, Deviation, and Departures from Sentencing Guidelines: The Case of Hong Kong

• DOI: http://doi.org/10.1111/jels.12260
• Published date: 01 September 2020
• Author: Sayaka Ri,Kevin Kwok‐yin Cheng,Natasha Pushkarna
• Date: 01 September 2020

Journal of Empirical Legal Studies

Volume 17, Issue 3, 580–614, September 2020

Judicial Disparity, Deviation, and

Departures from Sentencing Guidelines:

The Case of Hong Kong

Kevin Kwok-yin Cheng,*Sayaka Ri, and Natasha Pushkarna

Analyzing sentencing disparity calls for more calibrated measures to capture the nuances

of judicial discretion within jurisdictions that adopt strict sentencing guidelines. This arti-

cle uses an unconventional outcome variable, percent deviation, to investigate guideline

digressions in a nested, multilevel model. Percent deviation is calculated based on the dif-

ference between the guidelines’ “arithmetic starting point” and the actual starting point

that a judge adopts. Two equations were used to measure percent deviation from the arith-

metic starting point before and after adjustment for guilty plea sentence reductions.

Extracting data on drug trafficking cases from an open-source database from the Hong

Kong Judiciary (n= 356), we illustrate how percent deviation can be employed as a

measure of inter-judge disparity using hierarchical linear models (HLMs). Our findings

suggest that approximately 8 to 10 percent of the deviation in sentence length can be

attributed to judges’ differential sentencing behaviors. The deviation is affected by case

characteristics as well as judicial characteristics. Due to the wide guideline ranges, depar-

tures from said guidelines’ ranges are not common. This indicates that the guideline

ranges mask the deviation and inter-judge disparity that exist and recur.

I. Introduction

The balance between consistency and judicial discretion lies at the heart of sentencing.

Different jurisdictions have varying forms of sentencing guidelines that promote unifor-

mity and restrain judicial disparities. For instance, the Sentencing Council for England

and Wales provides sentencing guidelines for most common offenses. The court first

defines the category of the offense with reference to the offender’s culpability and the

*Address correspondence to Kevin Kwok-yin Cheng, Room 604, Faculty of Law, Lee Shau Kee Building, The Chi-

nese University of Hong Kong, Shatin, New Territories, Hong Kong SAR; email: kevincheng@cuhk.edu.hk. Cheng

is Associate Professor and Assistant Dean (Research) at the Faculty of Law, The Chinese University of Hong Kong;

Ri is Research Assistant at the Faculty of Law, The Chinese University of Hong Kong; Pushkarna is Postdoctoral

Fellow at the Faculty of Law, The Chinese University of Hong Kong.

We thank Antonio Da Roza, the editor, and the anonymous referees for their helpful comments and feedback

on earlier drafts of this article. This research was generously supported by a General Research Fund (GRF) by the

Research Grants Council, University Grants Committee (Project No. 14609918).

580

harm caused. The starting point of the sentence, which is within a category range, is then

ascertained. A non-exhaustive list of aggravating and mitigating factors is reviewed to

determine if there should be any upward or downward adjustment from the established

starting point of sentence. The court should also consider any sentence reduction for a

guilty plea (Sentencing Council 2012). In the U.S. Federal Sentencing Guidelines, there

is a two-dimensional grid that defines sentencing ranges. Since the 2005 U.S. Supreme

Court case of United States v Booker,

1

the guidelines are no longer mandatory but are now

advisory. In Hong Kong, where the present study takes place, sentencing guidelines are

set out by the appellate court. The Hong Kong Court of Appeal specifies guidance with

respect to sentences of particular offenses for the lower trial courts.

2

The Court of Appeal

has previously stated:

Fairness to convicted persons require[s] a substantial degree of consistency in the punishment

to be imposed for similar offences … While there can never be perfect consistency, since the

facts of each case, and the record, character and circumstances of convicted persons differ, jus-

tice requires that there should not be wide variations in punishment because of individual views

of judges or magistrates …

3

Despite this, sentencing studies have underscored the prominence of inter-judge

disparity in sentence outcomes (Anderson & Spohn 2010; Bushway et al. 2012;

Hester 2017; Krasnostein & Freiberg 2013; Marder & Pina-Sa´nchez 2018; Roberts &

Ashworth 2016; Yang 2014). Inter-judge disparity is defined as the relative differences in

average sentence lengths for comparable caseloads assigned to different judges

(Anderson et al. 1999). Efforts to quantify disparity have largely been devoted to measur-

ing disparities in terms of sentence length (Albonetti 2002; Anderson & Spohn 2010;

Bushway et al. 2012; Fleetwood et al. 2015; Nutting 2017; Pina-Sa´nchez et al. 2020; Pina-

Sa´nchez & Linacre 2016; Tillyer et al. 2015). Prior to the Supreme Court’s decision in

Booker, a handful of studies incorporated expected sentence length as a control variable

for sentencing models (Bushway & Piehl 2001; Engen & Gainey 2000; Griswold 1987).

Consistent across all models, deviation was defined as the proportional difference

between the actual and recommended sentence based on sentencing guidelines.

In this study, we build on previous approaches to examine inter-judge disparity by

applying percent deviation to capture and illustrate sentence deviation. Percent deviation

allows for finesse in observing the extent a judge diverged from the sentencing guidelines

by calculating the difference between the actual and recommended sentence lengths as a

proportion of the recommended sentence. As will be revealed in our study, an “arith-

metic starting point of sentence” can be calculated where there are clear sentencing

1

543 U.S. 220 (2005).

2

Although the highest court in Hong Kong is the Court of Final Appeal, the Court has stated that it is “not a sen-

tencing court” and that it is appropriate for the second highest court, the Court of Appeal, to set sentencing guide-

lines: Secretary for Justice v. Wong Chi-fung and Ors [2018] HKCFA 4.

3

Chan Chi-ming v. R [1979] HKLR 491 at 493.

Judicial Disparity, Deviation, Departure from Sentencing Guidelines 581

guidelines, such as in the case of drug trafficking where the starting point of sentence is

based on the drug type and weight of the trafficked drugs in question. The two equations

that we employ measure deviation from the arithmetic starting point based on sentencing

guidelines—one before and one after guilty plea reductions. Given how sentence reduc-

tions for guilty pleas must be explicitly considered in numerous sentencing regimes, it

was important to have a separate equation to account for them.

The equations proffered by this study are applicable to all offense types with sen-

tencing guidelines; however, this study selected drug trafficking cases to illustrate. An

original dataset derived from coding judges’ “Reasons for Sentence” was created for all

drug trafficking cases within the study period. By collecting data from court judgments,

we were able to elicit factors that judges explicitly state are relevant to their sentencing

decisions. Judges’ “Reasons for Sentence” provide detailed insight on how each judge

obtains the starting point of sentence, which aggravating and mitigating factors were

presented by the prosecution and defense, which factors were ultimately considered, and

the amount of sentence reductions, if any, awarded to defendants for their timely guilty

pleas. Through further analyses, we are able to determine the extent to which these fac-

tors influence sentencing decisions and how much judges deviate from the sentencing

guidelines. These various aspects build on the existing literature on inter-judge disparity,

with a particular focus on the legal backgrounds of judges. Additionally, the unique con-

text of Hong Kong extends the literature to that of a common law system in a Chinese

context.

II. Literature Review: Inter-Judge Disparity

Inter-judge disparity is a rich area of research, which has continued to grow to capture

newly realized influences upon a judge’s sentencing behavior (Anderson & Spohn 2010;

Bushway et al. 2012; Hester 2017; Krasnostein & Freiberg 2013; Marder & Pina-

Sa´nchez 2018; Myers 1988; Roberts & Ashworth 2016; Tonry 1996; Ulmer & Johnson 2017;

Yang 2014). Early studies focused on seemingly biased treatment based on offenders’

traits, such as their race, gender, and socioeconomic status (the basis of focal concerns

theory within sentencing that predominates in the criminology literature)

(Steffensmeier 1980; Steffensmeier et al. 1998). Looking at the “blameworthiness” of the

defendant, the judge may rely on heuristics stemming from certain identity groups being

linked to delinquency and violence (or lack thereof) (Albonetti 1997; Mustard 2001;

Steffensmeier 1980).

With increasingly comprehensive sentencing guidelines implemented through the

1980s in the United States, inter-judge disparity studies focused on whether the applica-

tion of those guidelines curtailed the unequal treatment in sentencing (U.S. Sentencing

Commission [USSC] 2019). Touted to bring greater conformity and consistency in sen-

tencing, the guidelines shifted the power to determine sentences from judges onto prose-

cutors, which led to criticisms that judges were rendered “expert clerks” (Gertner 2007:

536). Based on characteristics of the offender and the offense, the guidelines offered

decision trees for judges to follow as they review the facts of a given case (for the detailed

582 Cheng et al.