On the economics of plagiarism

Abstract

Cheating and plagiarism can involve the transgression of intellectual property rights across many areas of life. When a direct
financial benefit from such practices is identifiable, the opportunity to seek legal redress is available via civil court
action. When it is undertaken by a public official it may constitute malfeasance. Yet in the case of breaches of university
regulations (from the growing number of student cheating and plagiarism incidents) subsequent legal intervention may be characterised
by situations where the university is the defendant and the alleged plagiarist is the plaintiff (seeking compensation for
interrupted study and/or tarnished reputation). University defences can flounder around the issue of proving intent to deceive.
What can they do to try to prevent such occurrences? This paper uses economic analysis to examine such issues. Economic models
of plagiarism motivated primarily by (i) time-saving and (ii) dishonesty are developed to help frame the discussion. Both
model approaches overlap in their implications, namely, ensuring that sufficient resources are devoted to monitoring coursework
(to increase the probability that cheating and plagiarism are detected) and of providing sufficiently clear and severe institutional
penalties (to counter-balance any expected benefits that the student may perceive to be available from cheating and plagiarism).
Policy proposals are raised for further debate and consideration.

Full text

On the Economics of Plagiarism
Alan Collins
*
, Guy Judge
*
, Neil Rickman
**
*
Department of Economics, University of Portsmouth
**
Department of Economics, University of Surrey
Correspondence Address:
Prof. Alan Collins
Department of Economics
University of Portsmouth
Richmond Building, Portland Street,
Portsmouth, PO1 3DE. U.K.
Tel: ++44(0)23 9284 4128
Fax: ++44(0)23 9284 4037
Email: alan.collins@port.ac.uk

2
On the Economics of Plagiarism
1
Abstract: Cheating and plagiarism can involve the transgression of intellectual property rights
across many areas of life. When a direct financial benefit from such practices is identifiable, the
opportunity to seek legal redress is available via civil court action. When it is undertaken by a
public official it may constitute malfeasance. Yet in the case of breaches of university regulations
(from the growing number of student cheating and plagiarism incidents) subsequent legal
intervention may be characterised by situations where the university is the defendant and the alleged
plagiarist is the plaintiff (seeking compensation for interrupted study and/or tarnished reputation).
University defences can flounder around the issue of proving intent to deceive. What can they do to
try to prevent such occurrences? This paper uses economic analysis to examine such issues.
Economic models of plagiarism motivated primarily by (i) time-saving and (ii) dishonesty are
developed to help frame the discussion. Both model approaches overlap in their implications,
namely, ensuring that sufficient resources are devoted to monitoring coursework (to increase the
probability that cheating and plagiarism are detected) and of providing sufficiently clear and severe
institutional penalties (to counter-balance any expected benefits that the student may perceive to be
available from cheating and plagiarism). Policy proposals are raised for further debate and
consideration.
JEL codes: I21, H42, K42.
Keywords: plagiarism, cheating, deception, Higher Education.
1
We are grateful to an anonymous referee for his critical but constructive comments on an earlier draft of this paper. In
addressing these points we believe that our argument has been significantly strengthened.

3
INTRODUCTION
Cheating and plagiarism can involve the transgression of intellectual property rights across many
areas of life. A recent high profile example is the case of Acquaviva v Madonna reported in BBC
(2005). In that instance a Belgian songwriter (Salvatore Acquaviva) won a court battle against pop
star Madonna after accusing her of plagiarising one of his songs for her 1998 hit single, “Frozen”.
Another case relates to an instance of plagiarism by staff at the UK Government Foreign and
Commonwealth Office in a report relating to the potential terrorism risks posed by the former Iraqi
regime (Rangwala, 2003). There have also been cases of journalists who have had to resign after
being found guilty of plagiarism (Byrne, 2003). Plagiarism among academic economic researchers
has also been subject to detailed consideration (Enders and Hoover 2004).
When a direct financial benefit from such practices is identifiable, the opportunity to seek legal
redress is available via civil court action. When it is undertaken by a public official it may constitute
malfeasance. Yet in the case of breaches of university regulations (from the growing number of
student cheating and plagiarism incidents) subsequent legal intervention, may actually be
characterised by situations where the university is the defendant and the alleged plagiarist is the
plaintiff (seeking compensation for interrupted study and/or tarnished reputation). See, for example,
Clark versus University of Lincolnshire and Humberside (2000). The case shows the difficulties in
proving intent to deceive.
The extent of plagiarism among students is now a growing concern within higher education
institutions (HEIs) around the world (Larkham and Manns, 2002). In the academic community
anecdotal evidence alongside formal and casual empiricism suggests that instances of plagiarism
have increased significantly (Furedi 2004). A recent survey of UK university students found that

4
about one quarter of them admitted to plagiarising and 16% said that that they had done so more
than once (BBC 2004).
Plagiarism is also a problem in North America: McCabe (2003) found that 38% of students in the
23 US universities he sampled admitted to plagiarism (compared with only 10% in a similar but
smaller survey conducted three years earlier).
In the UK an MBA student recently faced expulsion from her university after it was discovered that
she had paid an essay writing company to produce coursework for her (Curtis 2003). Many
universities now have substantial sections on their web pages warning students about plagiarism
(for general advice see, for example, JISC 2005). Some make use of software services such as those
offered by turnitin.com or the JISC Plagiarism Advisory Service to check a sample of submitted
coursework for plagiarism (JISCPAS 2004).
The costs of such activity are potentially large. As well as damaging the integrity of degree awards
in a global marketplace (and thus potentially leading to some measure of trade deflection),
plagiarism diverts scarce academic resources away from research and teaching, towards monitoring
and detection activities. Thus, there is significant (and timely) policy relevance, in addition to
natural academic interest, in opening up economic debate on this theme. By directing some
economic research attention to plagiarism in HEIs it may be possible to help explain its apparently
increasing prevalence and better devise policies to combat this phenomenon.
An alternative view may be, however, that plagiarism and cheating have now become much more
accepted behavioural norms (Szabo and Underwood, 2003), such that their presence carries through
useful skills into work environments. This is not to say that deception practices being utilised by
students is a new phenomenon. In his novel Porterhouse Blue Tom Sharpe (1974) details some past
practices that featured in the days before student photo-identity cards. He presents a fictional

5
account of the means by which students, who through an intermediary (Scullion), could employ a
more academically gifted student to do coursework for them or sit their examinations. The fee for
this service varied according to the desired class of examination/coursework mark. There are
numerous other anecdotal tales of similar practices and educational corruption from around the
world. Yet while such practices are not new it is the case that technical change (principally in terms
of electronic library services and Internet search engines) and market evolution (the emergence of
plagiarism or cheat service websites and firms) has made the practice of plagiarism significantly
less time-intensive, cheaper and of more widespread availability.
It is contended herein that value may be added to the study and development of plagiarism and
counter-plagiarism tactics and policies by the adoption of an economic approach. This paper
articulates the rationale for this assertion and brings to bear some of the relevant conceptual tools
for analysing this growing phenomenon. Drawing on such thinking, key policy implications are
raised along with a series of empirical questions that beg to be addressed so that policy may be
more fully informed by firmer evidence. Section I discusses the changing nature of Higher
Education and the reasons why students may be motivated to plagiarise. Section II presents and
analyses various economic models of plagiarising behaviour. Section III considers counter-
plagiarism policies in Higher Education. Section IV provides some concluding remarks and
identifies some questions for future research.
I. THE CHANGING NATURE OF HIGHER EDUCATION
AND THE MOTIVATION TO PLAGIARISE
The Good Old Days?
Prior to the widespread use of the Internet, plagiarism principally entailed physically assembling
hard copies of sources to plagiarise from and then transcribing and integrating them into a coherent,
hopefully seamless essay format. Since this method still requires the use of some significant

6
analytical thought in order to synthesise a successful plagiarised essay, it may be that the
incremental effort required to fashion the work into a non-plagiarised essay would not be great. In
this way the state of technology could be argued to have inhibited the extent of plagiarism to some
degree. Furthermore, university education prior to the widespread use of the Internet might also be
reasonably described as characterised not only by much smaller student cohorts with more generous
student grants, but also a much higher proportion of more academically motivated students. Such
students would thus be less likely to perceive a need to resort to plagiarism and would be more
likely to subscribe to the academic honour codes or social customs associated with their presence in
the university community. Not only would the smaller student intakes make the expected
probability of detection seem higher, as detection effort would clearly be spread among a lower
number of students, but also the smaller cohorts would be more likely to socially interact and form
social capital, such that academic honour codes and other customs associated with a sense of
collegiality would more easily be reinforced.
Changing academic honour codes?
There is a growing body of work providing economic analyses of the extent to which individuals
are willing to follow social customs such as academic honour codes. Such codes effectively form, in
the terminology of Gifford (1999), a 'mental public good' that does not wear out with use but is
actually strengthened by increasing use. Yet it is also diminished or destroyed by misuse or non-
use. Following Akerlof (1980) observance or otherwise of an academic honour code could be
viewed in terms of the standard postulates of utility maximisation, where an individual is motivated
by some implicit cost-benefit calculation. Therein Akerlof focuses on a possible prisoner's dilemma
with two clearly sub-optimal scenarios. In the extreme cases, these are where the academic honour
code is obeyed even though all agents (the students) are disadvantaged by so doing and the other
where all agents, despite widespread welfare losses, do not obey the honour code.
Another factor that may well have contributed to changing attitudes among students is the
increasing sense of university corporatism. By this we mean a process whereby academics are

7
merely seen as employees and students are seen first and foremost as 'customers' purchasing a
service/degree certificate which is subject to formal quality assurance procedures in a manner
somewhat akin to purchases of any other goods, such as chocolate biscuits.
Alternative motivations for plagiarism?
For some individuals, participation in plagiarism may simply arise where deception behaviour is an
intrinsic personality trait and thus a significant source of utility. However it is contended that much
plagiarism can principally be explained in terms of a perceived need to save academic time (effort)
in the face of competing uses of time: leisure and part-time work. Indeed the BBC story of the UK
survey cited above reported that this was one of the two main reasons given by students for
cheating. The time-saving nature of plagiarism-supporting innovations may thus be welcomed by a
subset of those students who, for example, face severe income constraints and need to engage in
significant amounts of part-time work to support their university education. The academic
opportunity cost of such part-time work hours could thus be mitigated to some extent by recourse to
time saving but deception-based academic practices. To capture this type of behaviour we develop a
series of models based on the assumption that students are plagiarising in order to save time.
The dramatic growth of overseas student numbers (many with part-time work needs) has generated
additional recruits to the likely pool of those who could more easily be tempted to engage in
plagiarism. This is understandable given that the academic work hours needed to achieve a given
level of educational attainment is necessarily increased for some students by the need also to engage
in substantial language translation effort. There may also be generational and cross-cultural
differences in the perception of plagiarism as wrongdoing. Another possibility is that students feel
under pressure to obtain at least an Upper Second Class award on their degree courses so that they
can proceed to Masters level courses or to compete effectively in the job market. Some students
who feel that they are not able to guarantee marks at this level, no matter how much honest work
that they do, may be tempted to plagiarise in order to overcome academic deficiencies. In any case

8
the second reason given by students in the UK plagiarism survey was that they found it easier to
cheat than to do the work themselves. Accordingly we develop a second type of model where
cheating is a substitute for honest work in the ‘marks production function’.
To conclude this section we argue that the decision to participate in plagiarism today is not
principally influenced by the durability of social customs or academic honour codes. It should be
thought of more in terms of an orthodox illegitimate activity that can readily be viewed in a
framework that partially exploits the work of Becker (1965,1968) as a starting point.
II ECONOMIC MODELS OF PLAGIARISING BEHAVIOUR
Galles et al (2003) observe that the typical prescription to the problem of plagiarism is to step up
monitoring efforts. They conclude this is both difficult and costly and accordingly advance the case
for a preferred strategy. This consists of tougher and more consistent sanctions. In this paper, an
attempt is made to address and frame the same problem in a way that more explicitly fleshes out the
specific motivations for cheating and plagiarism activity in order to inform policy discussion at a
somewhat finer scale of resolution.
As we have noted, students may be motivated to engage in plagiarism for a number of reasons.
Perhaps the most important is simply to save time. Contemporary students may have to divide their
time between three activities: study, leisure and part-time work. Stevens and Weale (2004) provide
some figures on the average hours per week devoted to each of these activities from a survey of UK
university students conducted in 2000 and 2001, and an analysis of how the part-time wage rate can
affect the allocation of time between each activity. To the extent that students can save time by
plagiarising they can keep study time in check and make more time for the other two activities. An
alternative view is that students seek to improve the mark that they are awarded for their work and
that they do this by attempting to pass off someone else’s work as their own. Here we analyse

9
various models that attempt to capture the essentials of these different types of behaviour.
(a) Time saving plagiarism models
Case 1: Simple plagiarism by a student
Assume that the student has a certain amount of time T available for leisure or for study. Let s be
the amount of time devoted to study and L is the remaining leisure time. We assume that time
devoted to study improves the student grade and hence contributes to utility. But we also assume
that the student enjoys utility from leisure time. We can introduce a budget constraint as well as a
time constraint and assume that the student works to supplement her income, which can be used to
buy consumer goods and services.
The student is aware that there is a risk that they will be caught cheating. She therefore chooses her
study hours, work hours, consumption of ordinary goods and services and expenditure on
plagiarism so as to maximize expected utility, which must include the expected disutility that she
will suffer if she is caught cheating.
Let m = exogenously provided income (grant, money from parents etc.)
Let h = number of hours worked part-time
Let w = the hourly wage rate
Let x = amount of consumer goods purchased
Let p = price of consumer goods
Let c = cost of one item of plagiarised material.
Let q = the number of items of plagiarised materials that the student purchases.
Let f = the time freed up by each unit of plagiarism.
Let r = the risk (probability) of the student plagiarist being caught (detected).
Let K(q) = the disutility or cost of being caught plagiarising. This could be via a punishment such as
the reduction of a degree award or the need to resubmit work with a capped mark, or even the
anticipated loss of future earnings due to less favourable references etc. It is assumed that this cost

10
increases with the number of items that the student has been caught plagiarising – here we assume
that if a student is caught plagiarising once she will be fully investigated so that all other instances
of plagiarism will be identified.
The formal model may be stated as follows:
0;0)(),,(][max  cqpxwhmhsLfqTtosubjectqrKxsLUUE
For this problem the Lagrangian is
)()()(),,(
21
cqpxwhmhsLfqTqrKxsLUV 

In this model the student must choose how much time to spend studying (s), how much time to
spend on paid employment (h), how much of her budget to spend on ordinary goods and services
(x) and how much to spend on plagiarised material (q). The necessary conditions for maximising
expected utility subject to the time and budget constraints are
2
:
[1.1]
0
1


ss
UV
[1.2]
0
21


wV
h
[1.3]
0
2


pUV
xx
[1.3]
0
21


cfrKV
qq
[1.5]
0
1
 hsLfqTV

[1.6]
0
2
 cqpxwhmV

Solving equations [1.1] to [1.4] yields the following condition
[1.7]
c
rKfU
p
U
qs
x


If we first assume that the risk of being caught plagiarizing is zero (or that, at the margin, there is no
disutility suffered from being caught plagiarizing) then this would reduce to
2
In this model the consumption of plagiarised material does not contribute directly to utility, However it can free up
time that can then either be spent on study, or in paid employment which can augment income and thus allow for

11
f
c
U
p
U
sx

which could also be written as
c
p
fU
U
s
x

The student would consume a combination of normal goods and plagiarised material up to the point
where the ratio of the cost of these items is equated with the ratio of the marginal utility of
consuming normal goods to the marginal utility derived from an additional unit of study, weighted
by the extra amount of time freed up by purchasing an extra unit of plagiarised material.
However, interpreting the full condition given in [1.7], when there is some chance of the student
being caught plagiarizing this will bring down the expected benefits of the time saved – the more so
the larger is the marginal disutility suffered when being caught plagiarizing.
Rearranging equation [1.5], substituting into [1.7] and solving for q we find










q
ThsL
UrKU
p
c
Sqx
giving
[1.8]
 
qx
S
rKU
p
c
ThsLU
q



So, again first considering the situation of no risk of being caught, we find
[1.9]
 
x
S
U
p
c
ThsLU
q


But the introduction of a risk of apprehension, or a significant punishment for being caught
cheating, will lower the amount of plagiarism undertaken since the denominator in [1.8] is that
much larger than in [1.9]
Case 2: The student engages in two forms of plagiarism with different costs, time saved and risks of
being caught.
additional consumption of goods and services.

12
Plagiarism can take a variety of forms. Here we distinguish two types. Type 1 plagiarism is where
the student constructs their assessed work from a variety of publicly available (free) sources. These
could take the form of online materials that have been downloaded and cut and pasted into the
submitted work, or they could be taken from printed material that is in the public domain. Although
this material is free there is a small cost to the student in assembling the material. However
provided that they are not caught this cost is outweighed by the time freed up by this form of
cheating. Even if they are caught it might be worth taking the risk if the probability of detection and
the penalty are not too high.
Type 2 plagiarism is where the student pays privately for someone else to do the work, either
mediated by an online cheat site, or through some other private arrangement (say paying a fellow
student to do the work).
So let q
1
denote the quantity of type 1 plagiarised material
let c
1
denote the cost of creating one unit of this type of material
let f
1
denote the time freed up by this form of plagiarism
let r
1
denote the risk of being detected engaging in this form of plagiarism
let q
2
denote the quantity of type 2 plagiarised material purchased
let c
2
denote the cost of purchasing one unit of this type of material
let f
2
denote the time freed up by this form of plagiarism
let r
2
denote the risk of being detected engaging in this form of plagiarism
So the problem becomes
)()(),,(][max
2211
qKrqKrxsLUUE 
0;0
22112211
 qcqcpxwhmhsLqfqfTtosubject
and the Lagrangian becomes

13
)()()()(),,(
2112221112211 s
qcqcpxwhmhsLqfqfTqKrqKrxsLUV 

The student must choose s, h, x and q
1
and q
2
and the necessary conditions for maximising expected
utility subject to the time and budget constraints are:
[2.1]
0
1


s
UVs
[2.2]
0
21


wV
h
[2.3]
0
2


pUV
xx
[2.4]
0
2111111


cfKrV
qq
[2.5]
0
2221222


cfKrV
qq
[2.6]
0
2211
1
 hsLqfqfTV

[2.7]
0
2211
2
 qcqcpxwhmV

But now
2
2221
1
1111
2
c
Krf
c
Krf
p
U
qq
x






Substituting for 
1
and rearranging we find
[2.8]
2
222
1
111
c
KrfU
c
KrfU
p
U
qsqs
x




As before if we first consider a case where there is no risk of being caught plagiarizing, then
students will arrange matters such that
[2.9]
2
2
1
1
c
fU
c
fU
p
U
SSx


14
It is likely that c
1
< c
2
(privately paid for plagiarised material will cost more than material taken
from public sources) and we also expect f
1
< f
2
(downloaded material will take more time to put
together than purchased material), so these factors may tend to offset each other leading to time-
saving students making use of both types of plagiarised work. Interpreting the full condition [2.9]
we expect r
2
< r
1
(privately plagiarised material will be harder to detect than material taken from
public sources which can usually be tracked down via web searches or using tools such as
turnitin.com or with help from the JISC Plagiarism Advisory Service (JISCPAS 2004)). Probably
the university authorities would take stronger action against students who indulge in the more
blatant type 2 form of cheating so we expect K
q2
> K
q1
. Thus while the risk of being caught may be
lower, the anticipated punishment would be greater and so these two factors may also tend to offset
each other. For these reasons we cannot say outright whether more of one type of plagiarism will be
consumed than the other; it depends not only on the costs of plagiarism but also the risks of being
caught, the expected scale of punishment and the relation of all of these factors to the marginal
utility of study time. However increasing both the risk of being apprehended and the punishment
for plagiarism would reduce the amount of either type of plagiarism.
(b) A mark enhancement model
Here we focus on students who plagiarise in order to improve their mark or grade, rather than to
save time. The student may choose two inputs to produce marks (
Y
): honest work (
W
) and
cheating (
C
).
The ‘marks production function’
),( CWY
is well-behaved with positive but diminishing marginal
products. If cheating occurs, let the probability of its detection be
]1,0[)( Cp
,
with
0)0(,0)( 

pCp
. If the student is caught cheating, she incurs a penalty of Z < 0. Both types
of effort are costly to the student:
),( CW

, which is increasing and convex in both arguments.

15
Ignoring the possibility of
0W
(perhaps we might say that
0),0( CY
: even the world’s most
severe plagiarist must copy/print/hand in their work and this may be thought of as ‘honest’ work)
the student’s problem is
0..),(),()](1[)(
max
,
 CtsCWCWYCpZCp
ch

Let the Kuhn-Tucker multiplier be
0

and let partials be denoted with
subscripts 1 and 2. The first-order conditions are:
[3.1]
0)](1[
1
 YCp
[3.2]
0)](1[(.)])[(
22



YCpYZCp
[3.3]
0C

Assume that the constraint does not bind (i.e. the student cheats and  = 0) dividing (3.2) by (3.1)
yields
[3.4]
))((
2
1
2
1
YZCp
Y
Y





Equation (3.4) says that a wedge is driven between the usual tangency between the MRTS
2
1
Y
Y

and MRT (
2
1



) by the penalty imposed for (detected) cheating. Note that the second term on the
right-hand-side is positive (when incorporating the ‘−’ in front of it). This adjustment moves the
student towards higher W and lower C, i.e. the penalty improves incentives for honesty.
Note that there are two ways to reduce the prospect of cheating (and, ultimately, generate a corner
solution with C = 0):

16
1. Increase the severity of penalties and monitoring (increase
)(Cp

and Z).
2. Modify assessment procedures to affect the relative sizes of
1
Y
and
2
Y
, for example by using the
full marking ranges (i.e. increase
1
Y
). This flattens the marks isoquant and moves the student
around the production possibility frontier. Measures such as interim reports for dissertations or
keeping logs of work on the development of assessed work can reduce the substitutability between
h and c in the marks production function.
Figure 1 illustrates the student’s optimum position and how it is affected by a penalty and the
probability of detection.

17
W
C
‘PPF’ (slope = -
2
/
1
)
‘iso-marks’ (slope = -Y
2
/Y
1
)
A
B
The student will move from A to B when
[i] p’(C) > 0,
[ii] Z-Y < 0,
[iii] Y
1
increases (so the iso-marks curve flattens)
Figure 1 The trade-off between honest work and cheating

18
III COUNTER-PLAGIARISM AND HIGHER EDUCATION POLICY
If the theoretical models presented above are a reasonable characterisation of the decision to
plagiarise then in what way should counter-plagiarism policy be shaped and how are current
developments in the Higher Education sector likely to impact on the future extent of plagiarism?
While it is of course possible to alter the 'technology' of some student assessment through defensive
assessment strategies and greater reliance on orthodox unseen examination papers, this does not
necessarily address all the skills and knowledge domains where academics deem it appropriate to be
assessed via essays and dissertation type submissions. Thus, if these types of assessments are to be
retained for some significant proportion of the overall degree assessment profile, some attention
needs to be directed to the costs of monitoring assessed work and increasing the probability of
detecting plagiarism, and of setting a sufficiently severe punishment for those caught plagiarising.
The Expected Probability of Detection
It is clear that actual and prospective students are able to observe large or growing numbers of
students taking any given degree course or course module. Arguably, merely featuring in a large
cohort may contribute to the perception of some students that the likelihood of plagiarism being
detected is small or falling. This can arise from some simple and plausible supposition. Consider
that with a large number of essay submissions and operating within the timescales generally agreed
for coursework returns, it may be considered that only a sample of these essays could realistically
be subjected to detailed plagiarism detection scrutiny. Further contributing to this perception may
be the accumulation of information and experience through, or within cohorts, regarding the likely
distribution of plagiarism detection effort amongst academic staff. If students discerned such
variations in detection effort then this could also potentially lead to strategic selection of course
module options.

19
Students' expectations regarding the probability of detection will also be conditioned in some part
by their perceptions of their own stock of plagiarism capital. There may be observed considerable
variations in the level of skill applied to producing a plagiarised output but that such skills may
rapidly develop with experience and/or via demonstration. At one extreme, extremely hard-to-detect
plagiarism might be characterised as exhibiting considerable time, effort, knowledge of sources and
other subject-specific skills that are none too different to those required by keen hard working
students not producing plagiarised coursework submissions. It may be characterised by (i)
extensive use of off-line sources (especially books) as opposed to principally on-line sources, (ii)
extensive re-writing with a thesaurus close to hand, and (iii) setting aside time to review the
plagiarised output. More often, however, as evidenced by its easy detection in a high proportion of
cases, plagiarism is simply employed as a time-saving strategy where (i) to (iii) may not feature
significantly, or at all. Unaided time-saving led plagiarism will often be sloppy patchwork quilts of
essay text cobbled together from a number of free web-based sources. Such attempts are easily
detected using a word string in an Internet search engine such as Google. The task is rendered even
easier if the text extracted from such web sources is of a high academic standard but is poorly
interwoven with original text (a characteristic of many weaker students who plagiarise or for those
plagiarists whose first language is not English).
However it is likely to be more efficient to make use of specialised plagiarism detectives than to
expect individual academics to track down plagiarised work. If coursework is submitted in
electronic form it can be checked for plagiarism against an existing database of material, both
online and offline. This could be at the institutional (university) level or at a national level through
the JISCPAS service. Checking for plagiarised material in this way also ensures fairness and
consistency in plagiarism detection. (More informal systems of plagiarism detection at a
departmental or individual academic level invite strategic behaviour on behalf of academic staff
who may have reasons to be more or less vigilant in their detection and reporting of plagiarism
cases. This can be analysed in a model that adopts a principal –agent framework to the management

20
of plagiarism).
It seems likely that for many students who choose to plagiarise, engaging in academic work may
offer less utility than other activities, or present actual disutility. Nevertheless, they require the
benefits of the academic outputs in terms of assignment grades to continue in their university
studies/lifestyle and to garner the benefits of graduation. It may thus be possible to erode the utility
of plagiarised work by weakening the perception that once graded the plagiarised work will not be
re-examined. As it stands students will be aware that academics are typically obliged to mark and
return scripts within a fixed time period as part of university quality assurance procedure. Students
will be aware that this renders it less likely that academic staff will have an adequate time
opportunity to scrutinise in sufficient detail a large sample of the assignments being graded.
Accordingly, the utility of plagiarised work could be eroded by clauses in student contracts with the
university such that student records will be amended to allow for retrospective grading in any
forensic auditing of coursework. Potentially this could feature the penalty of a formal withdrawal of
degree title, which can present significant implications for future labour market participation and
progression. Degree results could be publicly available by web searches to support employers in
verifying the durability of degree results in the light of plagiarism audits.
Costs and penalties of plagiarism
As we have noted already the costs of plagiarism for a student may be separated into those that
must be incurred whether or not the plagiarism is detected, and those that are only borne by students
who are caught plagiarizing. The first type of costs is effectively captured in our analysis via the
budget constraint, while the second enter the objective function and reduce the expected utility from
plagiarising.
There are a wealth of enterprises that effectively offer plagiarism assistance to students and whose
tariff structure relates to the degree of difficulty, degree level (year 1 undergraduate essays through

21
to Masters dissertations and PhD chapters) and level of specialised human capital required to
complete a particular assignment. Thus depending on the nature of the assignment fees may range
from around £50 to over £1500. Such enterprises raise many further interesting questions relating to
their evolution, industrial organization, and the sources of their specialised labour inputs. To what
extent does it comprise other academic staff and postgraduate research students, or ex-academic
staff and former postgraduate research students? There are any number of regulatory measures that
may be applied to restrict the functioning and market profile of these firms, yet given their largely
web-based presence, attempts to curtail trading seem likely to be doomed to failure, or even worse,
be seen to encourage a move towards even more difficult to monitor hit-and-run entry by such firms
around the world.
In terms of direct action, university networks could either exclude or be triggered to monitor use of
proscribed plagiarism and cheat sites. Yet what of the student who uses a home PC or laptop
computer? A milder yet more subtle approach to raise the expected costs of plagiarism could feature
wide public reporting to the student population of in-depth reviews of the quality of cheat site
outputs as part of an ongoing monitoring exercise in each subject area by legitimate university
academics. Students and universities could also explicitly recognise that effective plagiarism
detection is costly and that these costs need to be recovered. Student matriculation and study
contracts could feature a willingness to accept fee premia in the event of plagiarism detection and
potentially offer rebates after some suitable time interval to reward non-plagiarism. Where
plagiarism detection costs are more explicitly recognised in student contracts there would exist the
means to finance plagiarism detection more professionally within the university (a dedicated
plagiarism detection unit) or by subscribing to more formal external plagiarism detection services,
ideally motivated in part by performance-related fee contracts.
Some universities also operate penalties for plagiarism that instill leniency on a first offence (a
yellow card/red card model). The simple model set out in this paper would suggest that such an

22
approach (as discussed in Park, 2004) was folly in that it would be clearly rational to continue to
plagiarise until a yellow card was achieved. Our models suggest that universities that wish to
reduce the incidence of plagiarism need to have strict monitoring of coursework for plagiarism with
heavy penalties for those who are caught cheating in this way.
IV. CONCLUDING REMARKS AND FUTURE EMPIRICAL RESEARCH QUESTIONS
This brief paper sets out some simple models of the student decision to plagiarise in an attempt to
provoke discussion and to help trigger an economic intervention in the debate on the nature and
scale of plagiarism. It also seeks to illuminate some ways that economic thinking could influence
the design of policies to combat plagiarism. In fulsome agreement with Galles et al (2003) it is
argued that an economic contribution would be of considerable value in this policy arena since
educationists and other policy actors may not be so inclined to direct policy attention towards (i)
explicitly raising the expected probability of detection, (ii) eroding the utility of plagiarised
academic work and (iii) raising the costs of plagiarism. Education academics and university
'teaching and learning professionals' seem blinkered in their apparent persistence that the level of
plagiarism can be dramatically reduced or eradicated by merely ascribing a higher profile to
plagiarism avoidance in learning skills modules. In truth, there will always be students who choose
to plagiarise for its direct benefits and who do so with full knowledge of what plagiarism really is.
Universities and other higher education stakeholders should recognise this and support attempts to
address the nature and scale of the incentives to plagiarise.
To inform such work future research attention could usefully be deployed to provide an empirical
examination of the costs of plagiarism to the sector and perhaps the UK economy more generally.
To this end, it would also be useful to inform debate by recourse to a fuller understanding of the
shadow price of academic time in plagiarism detection and the real extent of the 'overlooking' of
plagiarism because of transaction costs, or an eye to pass rates (almost universally a key university

23
performance indicator). Further work to support counter-plagiarism policy could also usefully
explore the level and pattern of demand (by subject and type of student) of commercial services
offering plagiarism and cheating support.

24
References
Akerlof, George A. (1980) A Theory of Social Custom, of Which Unemployment May Be One
Consequence. Quarterly Journal of Economics XCIV: 749-775.
BBC (2004) Quarter of students cheating. BBC News 20
th
June 2005.
http://news.bbc.co.uk/1/hi/education/3852869.stm [date accessed 3/2/05]
BBC (2005) Madonna in plagiarism case defeat BBC News 18
th
November 2005
http://news.bbc.co.uk/go/pr/fr/-/1/hi/entertainment/music/4449580.stm [date accessed
19/11/2005]
Becker G. (1965) A Theory of the Allocation of Time, Economic Journal 75, 493-517.
Becker, Gary (1968) Crime and Punishment: An Economic Approach. Journal of Political
Economy 76(2), 169-217.
Byrne, Ciar (2003) USA Today journalist resigns in ethics row. The Guardian (9th January 2004-
01-13) http://media.guardian.co.uk/print/0,3858,4832590-105414,00.html [date accessed 15/01/04]
Curtis, Polly (2003) Cheating MBA student faces course expulsion. The Guardian (Thursday 24th
July 2003) http://education.guardian.co.uk/print/0,3858,4719394-108229,00.html [date accessed
15/01/04]
Enders, Walter and Hoover, Gary A. (2004) Whose Line Is It? Plagiarism in Economics, Journal of
Economic Literature XLII 487-493.

25
Furedi F (2004) Cheats are having a field day on campus. The Daily Telegraph (Wednesday 17
th
March) 19.
Galles, Gary, Graves, Philip E., Sexton, Robert L., and Walton, Surrey M. (2003) Monitoring Costs
and Tolerance Levels for Classroom Cheating, American Journal of Economics and
Sociology 62(4) 713-719.
Gifford, A Jnr. (1999) Being and Time: On the Nature and Evolution of Institutions. Journal of
Bioeconomics 1(1), 127-149.
Joint Information Systems Committee (JISC) (2005) Deterring, detecting and dealing with student
plagiarism. http://www.jisc.ac.uk/uploaded_documents/plagFinal.pdf [date accessed
16/02/04]
JISC Plagiarism Advisory Service (JISCPAS) (2004) Northumbria University and the Joint
Information Systems Committee.
http://online.northumbria.ac.uk/faculties/art/information_studies/Imri/Jiscpas/site/jiscpas.asp
[date accessed 15/01/04]
Larkham, P J and Mann, S (2002) Plagiarism and its Treatment in Higher Education. Journal of
Further and Higher Education 26(4) 339-349.
McCabe, D L (2003) Ethics and Teaching: Academic Integrity. Paper presented at the 2003
Conference on Ethics in Teaching and Research, Newark, New Jersey.
www.newark.rutgers.edu/gradnwk/TA%20Conference.index.html [date accessed 15/01/04]

26
Park, C (2004) Rebels without a Clause: Towards and Institutional Framework for Dealing with
Plagiarism by Students. Journal of Further and Higher Education. 28(3) 291-306.
Rangwala, Glen (2003) British Intelligence Iraq Dossier Relies on Recycled Academic Articles.
http://www.globalpolicy.org/security/issues/iraq/attack/2003/0205plagiarism.htm
[date accessed 3/02/05]
Sharpe, Tom (1974) Porterhouse Blue London, Secker & Warburg.
Stevens, P and Weale, M (2004) Lazy Students? A Study of Student Time Use. National Institute
of Economic and Social Research. http://www.niesr.ac.uk/PUBS/DPS/dp233.PDF
[date accessed 3/02/05]
Szabo, A and Underwood, J (2003) Cybercheats: Is Information and Communications Technology
Fuelling Academic Dishonesty? Active Learning in Higher Education 5(2) 180-199.
Turnitin.com (2005) http://www.turnitin.com/static/tour/tour_master.html
[date accessed 3/02/05]
Case referred to:
Clark v. University of Lincolnshire and Humberside (2000), UK Court of Appeal (Civil Division)
19
th
April 2000. Reference: [2000]3 All ER 752.