ArticlePDF Available

DOES STUDENT WORK REALLY AFFECT EDUCATIONAL OUTCOMES? A REVIEW OF THE LITERATURE: DOES STUDENT WORK REALLY AFFECT EDUCATIONAL OUTCOMES?

Authors:

Abstract and Figures

We review the theories put forward, methodological approaches used and empirical conclusions found in the multidisciplinary literature on the relationship between student employment and educational outcomes. A systematic comparison of the empirical work yields new insights that go beyond the overall reported negative effect of more intensive working schemes and that are of high academic and policy relevance. One such insight uncovered by our review is that student employment seems to have a more adverse effect on educational decisions (continuing studies and enrolment in tertiary education) than on educational performance (test and exam scores). *** A DISCUSSION PAPER VERSION OF THIS STUDY IF FREELY DOWNLOADABLE HERE: https://ideas.repec.org/p/iza/izadps/dp11023.html
Content may be subject to copyright.
doi: 10.1111/joes.12301
DOES STUDENT WORK REALLY AFFECT
EDUCATIONAL OUTCOMES? A REVIEW
OF THE LITERATURE
Brecht Neyt and Eddy Omey
Ghent University
Dieter Verhaest
KU Leuven
Stijn Baert*
Ghent University
Abstract. We review the theories put forward, methodological approaches used and empirical
conclusions found in the multidisciplinary literature on the relationship between student employment
and educational outcomes. A systematic comparison of the empirical work yields new insights that
go beyond the overall reported negative effect of more intensive working schemes and that are of high
academic and policy relevance. One such insight uncovered by our review is that student employment
seems to have a more adverse effect on educational decisions (continuing studies and enrolment in
tertiary education) than on educational performance (test and exam scores).
Keywords. Education; Self-selection; Review; Student employment
JEL codes. I21; J22; J24
1. Introduction
Student employment is the norm for a large number of youth in many OECD countries, both in secondary
and tertiary education (Marsh and Kleitman, 2005; Beerkens et al., 2011). For instance, for students in
tertiary education, the student employment rate is around 49% in the United States (USA) and 47% in
Europe (Beerkens et al., 2011). One important reason why many students combine study and work is that
it provides them with an income, which may help them to satisfy their consumption aspirations (Watts and
Pickering, 2000; Baert et al., 2016). However, research in multiple disciplines has shown that the effect
of students’ work decisions may go beyond the short term. For example from the broad field of sociology,
several studies show that student employment is correlated with problem behaviour among youth, such as
alcohol use, delinquency and drug use (Steinberg et al., 1993; McMorris and Uggen, 2000; Safron et al.,
2001). In addition, from the field of psychology, Steinberg and Dornbusch (1991) find that combining
study and work is associated with psychological and psychosomatic stress. Finally, studies in labour
Corresponding author contact email: Stijn.Baert@UGent.be; Tel: +32486492752.
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
DOES STUDENT WORK REALLY AFFECT EDUCATIONAL OUTCOMES? 897
economics and the sociology of work have extensively investigated the impact of student employment on
later labour market outcomes, finding mainly non-negative results (Ehrenberg and Sherman, 1987; Carr
et al., 1996; Ruhm, 1997; Hotz et al., 2002; Parent, 2006; Baert et al., 2016, 2017).
One aspect of student employment has been investigated across various disciplines in the social and
behavioural sciences: its impact on educational behaviour and performance (Carr et al., 1996; Warren
et al., 2000; Stinebrickner and Stinebrickner, 2003; Bachman et al., 2011). The central position of these
outcomes in the academic literature on the impact of student labour supply decisions on later outcomes
in youth is not surprising. First, it is highly relevant to examine the effect of student employment on
educational attainment since the trade-off between starting a student job and using this time for studying
is a decision every adolescent has to make (Bozick, 2007; Triventi, 2014). Second, if student employment
affects educational attainment, it indirectly affects all later outcomes in life that are (partly) determined by
this attainment (e.g. labour market success, wealth and happiness; Hartog and Oosterbeek, 1998; Blundell
et al., 1999; Chiswick et al., 2003). For these reasons, the impact of student employment on educational
attainment is also highly relevant from a policy point of view. Policymakers should take this potential
impact into account when making decisions about whether to encourage (particular forms of) student
employment.
This paper summarizes two decades of literature on the relationship between student employment and
educational outcomes. In general, research on this subject has experienced a rapid growth in the past
two decades, calling for a structured overview of the main findings of these studies. In particular, since
previous studies adopt various approaches to account for the biggest methodological challenge when
empirically investigating the relationship between student work and educational outcomes, that is the
endogeneity problem, it is interesting to compare their results by method used. Nevertheless, to the best
of our knowledge, the present study is the first to survey this body of research.
In general, that is when equally weighting all studies, we find that the effect of student work on
educational outcomes is non-positive. In addition, we find that the consequence of combining study and
work is more adverse for students in tertiary education than for students in secondary education. We
argue that this is due to the more challenging nature of tertiary education and to the different effect
of student work during secondary versus tertiary education with respect to attitudes towards school and
intertemporal preferences. Additionally, whereas combining study and work has an especially detrimental
effect on students’ educational choices (continuing studies and enrolment in tertiary education), the effect
of student work on students’ educational performance (test and exam scores) is a lot less worrisome. When
narrowing the focus of our review to studies that, in our opinion, used the most convincing approaches
to control for the endogeneity problem inherent to estimating the effect of student work on educational
outcomes, the contrast between the effect of student work on different outcome variables is even stronger.
The remainder of this paper is structured as follows. In the next section, we briefly sketch out the
main theories, cited in various disciplines, depicting the relationship between student employment and
educational attainment. In Section 3, we describe the endogeneity of these outcomes and the different
ways in which previous studies have tried to account for this problem. In Section 4, we present an overview
of the empirical findings, with a focus on how the results converge and diverge by country and educational
level, outcome variable, type of student job and student characteristics. In this section, we also compare
the results yielded by different statistical methods used to control for the endogeneity bias. Section 5
formulates the main takeaway messages from our review for scholars and policymakers.
2. Theoretical Mechanisms
In this section, we briefly introduce the main theories found in multiple disciplines, providing support for
a relationship between student employment and later educational outcomes. These theories help explain
the empirical findings in the literature, which we discuss extensively in Section 4. Studies that examine
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
898 NEYT ET AL.
this relationship are primarily interested in whether working while studying is a complement to or a
substitute for education that is whether it improves or worsens educational attainment, respectively. In the
following paragraphs, we consecutively present the leading theories that advocate both of these views.
On the one hand, according to Human Capital Theory (Becker, 1964), student employment can be a
complement to education due to the additional skills and knowledge obtained while working. There are
several reasons why student work may lead to such an increase in human capital. First, student employment
enables the acquisition of new general and transferable skills such as work values, communication skills
and a sense of time management (Rothstein, 2007; Staff and Mortimer, 2007; Buscha et al., 2012).
As these non-academic skills and achievements are increasingly important in college admissions and
employers’ hiring decisions (Bruggink and Gambhir, 1996; Rosenbaum, 2002; Ashworth et al., 2017;
Baert and Vuji´
c, 2018), combining study and work may substantially contribute to the accumulation
of human capital necessary for post-secondary education success.1Second, combining study and work
may offer students the opportunity to apply in practice what they have learned in school (Hotz et al.,
2002; Geel and Backes-Gellner, 2012). Third, student employment may change students’ intertemporal
preferences and increase their future-orientedness, thereby motivating them to work harder in school in
order to achieve a certain career goal (Oettinger, 1999; Rothstein, 2007).
On the other hand, building on the Theory of the Allocation of Time (Becker, 1965), Zero-Sum Theory
suggests that student employment and education are substitutes. More formally, this theory states that
students have fixed time resources and that student employment strongly constrains students’ use of these
resources. As a consequence, time resources used to work cannot be used for activities that enhance
academic performance (e.g. studying, doing homework and attending classes; Bozick, 2007; Kalenkoski
and Pabilonia, 2009, 2012; Darolia, 2014). As the reduced time spent on these activities subsequently
worsens academic performance (Stinebrickner and Stinebrickner, 2004, 2008; Arulampalam et al., 2012),
student employment may have a detrimental effect on educational outcomes.
However, there are several reasons why student work may not substantially crowd out time spent
on activities that enhance academic performance, and why this (potential) crowding out may not be
as detrimental to academic achievement as Zero-Sum Theory suggests. First, previous time use studies
show that spending one more hour on student work does not necessarily translate into spending 1 hour
less on study activities (Triventi, 2014). In other words, student workers may cut back on leisure time
without reducing the time they dedicate to school-related activities (much). Indeed, Schoenhals, Tienda
and Schneider (1998), Warren (2002) and Kalenkoski and Pabilonia (2009, 2012), find that time spent
working does not reduce the time spent on school-related activities in a one-to-one relationship. Working
students also scale down the time spent on non-school-related activities (e.g. time spent with family
or friends and time spent watching television or in front of a computer). Second, Babcock and Marks
(2011) show that time allocated to attending classes and studying has decreased substantially over the past
decades for students in USA tertiary education. Consequently, time spent working may rather crowd out
the increasing time spent outside (formal) education instead of the decreasing time spent inside (formal)
education. Third, there may exist an interaction between Human Capital Theory and Zero-Sum Theory.
That is, applying the law of diminishing returns to both learning by studying and learning by doing (Sen,
1966), the marginal benefits (marginal costs) of student work with respect to gaining human capital are
the highest (lowest) for the first hours of employment. Indeed, on the one hand, these first hours may
be the most essential ones with respect to gaining human capital through a student job. On the other
hand, the first hours of student work may crowd out the least important hours with respect to gaining
human capital through studying. This last point also exemplifies why there may be a non-linear effect
of student work on educational achievement that is the effect may be positive up to a threshold of hours
worked and turn negative when that threshold is exceeded.2
Another theory that supports a negative association between student work and educational success is
Primary Orientation Theory (Warren, 2002; Bozick, 2007; Baert et al., 2017), often cited in the field
of sociology. This theory suggests that the worse academic performance of working students compared
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
DOES STUDENT WORK REALLY AFFECT EDUCATIONAL OUTCOMES? 899
to non-working students is related to their primary orientation being toward work rather than toward
school. In other words, it reflects a disengagement from school that existed before the decision to
work was made, rather than a negative effect due to student employment itself. Therefore, instead of
providing an explanation for a causal, negative effect of student work, this theory reveals a potential
selection problem that one wants to control for in empirical analyses.3Indeed, Bozick (2007), Staff and
Mortimer (2007) and Triventi (2014) hypothesize that when pre-existing differences between working
and non-working students, such as their primary orientation, are properly controlled for, the difference
in academic performance between these two groups disappears. We elaborate more generally on this
selection problem in the next section.
In Section 4, where we discuss the empirical findings in the literature, we distinguish between studies
focussing on the effect of student work during secondary education and those focussing on the effect of
student work during tertiary education. Based on the aforementioned theories, there are several reasons
why student employment is expected to be less of a substitute for education for students in tertiary
education. First, as noted earlier, over the last decades students in (United States) tertiary education
have spent less and less time on study-related activities, potentially invalidating the application of Zero-
Sum Theory here. However, this evolution was not found for students in secondary education (Zick,
2010). Therefore, the assumption that working crowds out time spent on activities that foster academic
performance may be less valid for students in tertiary education. Additionally, as students in tertiary
education have more flexibility in their schedules, the main assumption of Zero-Sum Theory may again
be violated for them. Indeed, their classes are usually not compulsory, they often have flexibility in
planning their academic workload by choosing between different courses (Triventi, 2014), and they can
decide to spread their studies over more years than the default number in order to lighten the workload
per year (Darolia, 2014). Second, a selection effect with respect to students’ primary orientation may be
less of an issue for students in tertiary education, since only more school-oriented students will choose
to commence this form of non-compulsory education. More work-oriented students will not enter this
type of education, but rather pursue labour market opportunities (Bozick, 2007). Third, most students in
tertiary education have already combined study and work in secondary education, so they should be more
adept at mixing these two activities (Bozick, 2007; Staff and Mortimer, 2007).
3. The Endogeneity of Student Work and Educational Outcomes
In this section, we discuss the substantial problem all researchers face when empirically investigating
the impact of student employment on educational outcomes: the endogeneity of both variables. The
importance of this problem originates from the fact that results can only be given a causal interpretation
if endogeneity has been adequately controlled for (Stinebrickner and Stinebrickner, 2003; Marsh and
Kleitman, 2005; Baert et al., 2016). We describe the cause of the endogeneity of student work and later
educational outcomes in Subsection 3.1. Then, in Subsection 3.2, we present various methods that are
employed to tackle the endogeneity problem.
3.1 Description of the Problem
Students who decide to combine study and work differ from those that do not combine these two
activities in more than just their work status (Warren and Lee, 2003; Singh, Chang and Dika, 2007).
These pre-existing differences between working and non-working students may also affect educational
outcomes (Rothstein, 2007). For the impact of student employment on educational outcomes to be given
a causal interpretation, one should control for these common determinants. If not, variation in educational
outcomes that should be attributed to the pre-existing differences between working and non-working
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
900 NEYT ET AL.
students will mistakenly be attributed to the difference in work status (Stinebrickner and Stinebrickner,
2003; Baert et al., 2017).
The pre-existing differences between working and non-working students can be both observable (e.g.
gender, ethnicity and parental education level) and unobservable (e.g. motivation and ability) to the
researcher. Although it is fairly easy to account for the former category of pre-existing differences by
including these factors as control variables in the estimation process, it is much harder to account for the
latter category of pre-existing differences. However, not accounting for these factors biases coefficient
estimates.
When estimating the effect of student employment on academic achievement, factors like motivation,
ability and students’ financial constraints are cited as the most important factors that are usually not
observed by researchers (Sabia, 2009; Bachmann et al., 2011; Scott-Clayton, 2012). Remarkably, for
the first two of these, that is motivation and ability, different authors argue for opposing selection effects
induced by these factors. On the one hand, several authors argue that highly motivated and able students are
more frequently involved in student work (Stinebrickner and Stinebrickner, 2003; Tyler, 2003; Beffy et al.,
2010). Following this reasoning, student workers are a positive selection of the overall population. On
the other hand, other authors argue for a negative selection, reasoning that students with lower motivation
and lower ability are more likely to combine study and work (Rothstein, 2007; Bachmann et al., 2011;
Buscha et al., 2012). With regard to financial constraints, Kalenkoski and Pabilonia (2010) and Behr and
Theune (2016) have shown that the financial wealth of a family is negatively associated with the number
of hours students work. As Sirin (2005) showed in his review of 58 studies that socioeconomic status
(which is closely linked to financial wealth of the family) is positively related to academic achievement,
one should also control for students’ financial constraints to control for a (potentially) negative selection
effect.
Consistent with Zero-Sum Theory (supra, Section 2), there may also be an issue of reverse causality.
Indeed, academic achievement might influence students’ time use and therefore both whether and how
much students work (or are allowed to work by their parents) (Sabia, 2009). Here too this endogeneity
issue could lead to both positively and negatively biased estimates. On the one hand, students with
higher marks may decide that they are capable of combining their studies with student employment,
resulting in a positive selection (Sabia, 2009). On the other hand, there may be a negative selection
bias if students who perform badly in school become discouraged in seeking academic success and
turn to student work as a more fruitful allocation of their time (Warren and Lee, 2003; Bozick,
2007).
In previous research, several methods have been developed to account for these sources of endogeneity.
We discuss these methods in the next subsection. Additionally, in Subsection 4.2, we report empirical
evidence on the direction of the endogeneity bias.
3.2 Methodological Approaches to Tackling the Problem
In this subsection, we sum up five categories of methods that are used to control for the endogeneity
problem described in the previous subsection. Column (5) in Table 1 summarizes the main methodological
approaches of each of the articles included in the present review study – we return to the selection of
these articles in Section 4.
As reviewed by Ruhm (1997), a first generation of studies treated student employment as (nearly)
exogenous. They examined descriptive statistics and conducted simple regressions (controlling for a small
set of observable factors besides student work). The contributions listed in Table 1 using ordinary least
squares (OLS), linear probability models (LPM) and logit regression models are, from a methodological
point of view, close to these first-generation studies as their primary strategy is to absorb as much
observable heterogeneity influencing both student work decisions and later educational outcomes as
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
DOES STUDENT WORK REALLY AFFECT EDUCATIONAL OUTCOMES? 901
Table 1. Summary of the Literature.
(1) Study (2) Country (3) Main outcome
variable(s)
(4) Main explanatory
variable(s)
(5) Main methodological
approach
(6) Main result(s)
A. Studies using data on student work during secondary education
Apel et al. (2008) USA (National Longitudinal
Survey of Youth;
1997–2003).
GPA and continuing studies. Any student work dummy
and hours worked per
week.
FE model combined with IV
approach (instrument:
state child labour laws).
Negative effect on continuing
studies only.
Baert et al. (2017) Belgian (Study Hive on
Transition from School to
Work Data; 1999–2009).
Graduating and tertiary
education enrolment.
Student work (during the
summer and academic
year) dummies.
Dynamic discrete choice
model with unobserved
heterogeneity.
Negative effect on tertiary
education enrolment when
working during both the
summer and the academic
year only.
Baum and Ruhm
(2016)
USA (National Longitudinal
Survey of Youth; 1979 and
1997)
Tertiary education
enrolment and tertiary
education graduation.
Hours worked per week. LPM. Negative effect on tertiary
education graduation only.
Buscha et al. (2012) USA (National Education
Longitudinal Study;
1988–1992).
Math and reading scores. Student work (of different
types) dummies and hours
worked per week.
DiD/DiDiD approach
combined with matching
approach.
No effect.
Dustmann and van
Soest (2007)
UK (National Child
Development Study; 1974).
Credits achieved and
continuing studies.
Hours worked per week. IV approach (instruments:
local unemployment rate
and parental income).
Negative effect for males
only.
Eckstein and
Wolpin (1999)
USA (National Longitudinal
Survey of Youth;
1979–1991).
GPA and continuing studies. Hours worked per week. Dynamic discrete choice
model with unobserved
heterogeneity.
Negative effect, albeit small.
Kalenkoski and
Pabilonia (2009)
USA (American Time Use
Survey; 2003–2006).
Minutes spent doing
homework per day.
Minutes worked per day. SEM. Negative effect.
Kalenkoski and
Pabilonia (2012)
USA (American Time Use
Survey; 2003–2008).
Minutes spent doing
homework per day.
Any student work dummy. SEM. Negative effect.
Lee and Staff
(2007)
USA (National Education
Longitudinal Study;
1988–1992).
Continuing studies. Intensive student work
dummy.
Matching approach. Negative effect.
Lee and Orazem
(2010)
USA (National Longitudinal
Survey of Youth;
1997–2002).
GPA, graduating and tertiary
education enrolment.
Hours worked during
secondary education.
IV approach (instruments:
individual date of birth,
state truancy laws and
local demand for low-skill
labour).
Negative effect on tertiary
education enrolment.
Positive effect on
graduating.
(Continued)
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
902 NEYT ET AL.
Table 1. Continued
(1) Study (2) Country (3) Main outcome
variable(s)
(4) Main explanatory
variable(s)
(5) Main methodological
approach
(6) Main result(s)
Marsh and
Kleitman (2005)
USA (National Education
Longitudinal Study;
1988–1992).
GPA, standardized test
scores, highest degree,
qualifications achieved,
months of college and
study engagement
variables. a
Hours worked per week. OLS. Negative effect.
McCoy and Smyth
(2007)
Ireland (National Survey of
Schools; 1994).
GPA and continuing studies. Any student work dummy
and hours worked per
week.
Matching approach. Negative effect.
McNeal (1997) USA (High School and
Beyond Study; 1980–1982).
Continuing studies. Student work (of different
types) dummies and hours
worked per week.
Logit model. Negative effect when working
in farming, doing
gardening work,
performing odd jobs or
working as a babysitter
only.
Montmarquette
et al. (2007)
Canada (Statistics Canada
School Leavers Survey;
1991 and 1995).
GPA and continuing studies. Hours worked per week. Dynamic discrete choice
model with unobserved
heterogeneity.
Negative effect (when
working more than
15 hours per week) on
continuing studies for
males only.
Oettinger (1999) USA (National Longitudinal
Survey of Youth;
1979–1983).
GPA. Weeks worked per year and
hours worked per week.
FE model. Negative effect. Less adverse
for whites than for blacks.
Parent (2006) Canada (Statistics Canada
School Leavers Survey;
1991 and 1995).
Graduating. Hours worked per week. IV approach (instruments:
local unemployment rate
and provincial
unemployment rate of
25–44-year-olds).
Negative effect (when
working more than
10 hours per week).
Payne (2003) UK (England and Wales Youth
Cohort Study; 1998–2000).
Qualifications achieved. Hours worked per week. OLS and logit model. Negative effect (when
working more than
15 hours per week).
Quirk et al. (2001) USA (National Educational
Longitudinal Study;
1988–1992).
GPA. Hours worked per week. SEM. Negative effect when working
more than 12 hours per
week. Positive effect when
working less than 12 hours
per week.
(Continued)
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
DOES STUDENT WORK REALLY AFFECT EDUCATIONAL OUTCOMES? 903
Table 1. Continued
(1) Study (2) Country (3) Main outcome
variable(s)
(4) Main explanatory
variable(s)
(5) Main methodological
approach
(6) Main result(s)
Rothstein (2007) USA (National Longitudinal
Survey of Youth; 1997).
GPA. Hours worked per week. FE model and IV approach
(instruments:
county-level
unemployment rate,
average wage rate for
teens and state laws
regarding teen
employment).
No effect.
Sabia (2009) USA (National Longitudinal
Study of Adolescent Health;
1995–1996).
GPA and study engagement
variables. b
Any student work dummy
and hours worked per
week.
FE model. No effect.
Schoenhals et al.
(1998)
USA (National Education
Longitudinal Study; 1988
and 1990).
GPA and study engagement
variables. c
Student work categorical
variablehand hours
worked per week.
OLS. Negative effect on attendance
only.
Singh (1998) USA (National Educational
Longitudinal Study; 1990).
GPA and standardized test
scores.
Hours worked per week. SEM. Negative effect, albeit small.
Singh et al. (2007) USA (School and Social
Experiences Questionnaire;
2002).
GPA. Hours worked per week. OLS. Negative effect.
Staff and
Mortimer
(2007)
USA (Youth Development
Study; 1988–2003).
Graduating in tertiary
education.
Student work categorical
variable.i
Logit model. Non-workers and steady
workers have better
outcomes than sporadic
workers.
Staff et al. (2010) USA (Monitoring the Future
Project; 1992–1997).
GPA and study engagement
variables. d
Hours worked per week
(actual and desired).
RE model. Negative effect.
Tyler (2003) USA (National Educational
Longitudinal Study; 1990
and 1992).
Math and reading scores. Hours worked per week. IV approach (instrument:
child labour laws).
Negative effect.
Warren (2002) USA (self-administered
pencil-and-paper
questionnaire; 1999).
Study engagement
variables. e
Student work categorical
variablejand hours worked
per week.
Cross tabulation. Negative effect.
Warren and Lee
(2003)
USA (National Educational
Longitudinal Study; USA
Census; 1990 and 1992).
Continuing studies. Hours worked per week. Non-linear hierarchical
model.
Negative effect (when
working more than
20 hours per week).
Warren et al.
(2000)
USA (National Education
Longitudinal Study; 1990
and 1992).
GPA. Any student work dummy
and hours worked per
week.
SEM. No effect.
(Continued)
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
904 NEYT ET AL.
Table 1. Continued
(1) Study (2) Country (3) Main outcome
variable(s)
(4) Main explanatory
variable(s)
(5) Main methodological
approach
(6) Main result(s)
Weller et al. (2003) USA (Safe and Drug-Free
Schools Program; 1995).
GPA and study engagement
variables. f
Hours worked per week. MANCOVA, ANCOVA
and MANOVA.
Negative effect.
B. Studies using data on student work during tertiary education
Arano and Parker
(2008)
USA (administrative college
data and self-administered
online questionnaire;
2005).
GPA. Hours worked per week. IV approach (instrument:
students’ financial
resources).
Negative effect for freshmen
and for sophomores
(juniors) (seniors) when
working more than 9 (19)
((15)) hours per week.
Positive effect for
sophomores (juniors)
((seniors)) when working
less than 9 (19) ((15))
hours per week.
Bachman et al.
(2011)
USA (Monitoring The Future
Project; 1976–2003).
Years of college. Hours worked per week. Matching approach. Negative effect (when
working more than
15 hours per week).
Baert et al. (2017) Belgium (self-administered
online questionnaire;
2017).
Credits achieved. Hours worked per week. OLS. Negative effect when being
work-oriented. No effect
when being study-oriented.
Beerkens et al.
(2011)
Estonia (Survey of Students’
Socio-Economic Situation;
2008).
Graduating (without delay). Hours worked per week. Logit model. Negative effect, albeit small
(when working more than
25 hours per week).
Beffy et al. (2010) France (French Labor Force
Surveys; 1992–2002).
Graduating. Any student work dummy
and hours worked per
week.
IV approach (instruments:
local unemployment rate
for low-skilled youth and
father’s social status).
Negative effect.
Behr and Theune
(2016)
Germany (Absolventenpanel;
2001).
Graduating (without delay). Any student work dummy. Matching approach. Negative effect.
Body et al. (2014) France (self-administered
online questionnaire;
2012).
Graduating. Hours worked per week. IV approach (instruments:
students’ lifestyle, social
category of parents,
financial support and
nationality).
Negative effect (when
working more than 8 hours
per week). Less adverse in
the public sector than in the
private sector.
(Continued)
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
DOES STUDENT WORK REALLY AFFECT EDUCATIONAL OUTCOMES? 905
Table 1. Continued
(1) Study (2) Country (3) Main outcome
variable(s)
(4) Main explanatory
variable(s)
(5) Main methodological
approach
(6) Main result(s)
Bozick (2007) USA (Beginning
Post-secondary Students
Longitudinal Study; 1996
and 1998).
Continuing studies. Hours worked per week. Logit model. Negative effect (when
working more than
20 hours per week).
Darolia (2014) USA (National
Longitudinal Survey of
Youth; 1997–2008).
GPA and credits achieved. Hours worked per week. FE model combined with IV
approach (instruments:
lagged outcome and
student work variables,
house prices and credit
scores).
Negative effect on credits
achieved only, for
full-time students only.
Derous and Ryan
(2008)
USA (self-administered
online questionnaire;
2008).
GPA and study engagement
variables. g
Hours worked per week. OLS. Negative effect on study
engagement only.
DeSimone (2008) USA (College Alcohol
Study; 1993–2001).
GPA. Hours worked per week. IV approach (instruments:
paternal schooling and
Jewish upbringing).
Negative effect.
Kalenkoski and
Pabilonia (2010)
USA (National
Longitudinal Survey of
Youth; 1997).
GPA. Hours worked per week. SEM. Negative effect, albeit small.
McVicar and
McKee (2002)
UK (Status Zero Survey;
1993–1998).
Qualifications achieved. Any student work dummy. Bivariate probit model. Negative effect (when
working more than
15 hours per week).
Moulin et al. (2013) Canada (Youth in Transition
Survey; 1999–2007).
Graduating. Hours worked per week. Cox proportional hazards
model.
Negative effect (when
working more than
25 hours per week).
Rochford et al.
(2009)
Ireland (Paid Part-Time
Employment
Questionnaire; 2009).
Validated scales on course
performance, personal
and professional
development, college
experience and grades
achieved.
Hours worked per week. OLS. Negative effect.
Scott-Clayton and
Minaya (2016)
USA (Beginning
Post-secondary Students
Longitudinal Study;
2001–2009).
GPA and graduating. Federal Work Study
Program participation
dummy.
Matching approach. Negative effect on GPA only.
(Continued)
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
906 NEYT ET AL.
Table 1. Continued
(1) Study (2) Country (3) Main outcome
variable(s)
(4) Main explanatory
variable(s)
(5) Main methodological
approach
(6) Main result(s)
Stinebrickner and
Stinebrickner
(2003)
USA (administrative college
data; 1989–1997).
GPA. Hours worked per week. IV approach (instrument: job
assignments).
Negative effect.
Theune (2015) Germany (Absolventenpanel;
2001).
Graduating (without delay). Student work categorical
variable.k
Cox proportional hazards
model.
Negative effect.
Triventi (2014) Italy (Eurostudent Survey;
2004).
Credits achieved. Student work categorical
variable.l
Treatment model with a latent
factor determining both
variables and exclusion
restrictions.
Negative effect.
Wen z a nd Yu
(2010)
USA (Winona State
University Student Sample;
2004–2008).
GPA. Hours worked per week. FE model. Negative effect, albeit small.
The following abbreviations are used: ANCOVA (analysis of covariance), DiD (difference-in-differences), DiDiD (difference-in-differences-in-differences),
FE (fixed effects), GPA (grade point average), IV (instrumental variable), LPM (linear probability model), MANCOVA (multivariate analysis of covariance),
MANOVA (multivariate analysis of variance), OLS (ordinary least squares), RE (random effects), SEM (structural equation modelling), UK (United Kingdom)
and USA (United States).
aIndicators of time spent on homework, frequency of absenteeism, school preparation, college preparations and number of colleges applied to.
bIndicators of whether students pay attention in class, finish their homework on time, get along with fellow students, skip class and expect to attend college.
cIndicators of school attendance, hours spent doing homework per week and hours spent reading per week.
dIndicators of whether students expect to attend college, try their best, do not complete assignments, misbehave at school, skip class and participate in school
activities.
eIndicators of being late for school, skipping class, getting in trouble for not following school rules, going to class without a pencil, pen or paper, going to
class without books, going to class without doing one’s homework and time spent on homework per week.
fIndicators of being late for school, skipping class, sleeping in class, cheating and time spent on homework per week.
gStudy attitude scale introduced by Weinstein et al. (1987).
hCategories: (1) Never been employed, (2) not currently employed but have been employed during the school year, (3) not employed this school year but have
been employed during the summer, (4) employed prior to last summer and (5) currently employed.
iBased on the total duration and average number of hours of student work, respondents are classified into five categories: (1) non-workers, (2) sporadic workers,
(3) occasional workers, (4) steady workers and (5) most invested workers.
jCategories: (1) Never been employed, (2) employed in the past, but not at the moment and (3) currently employed.
kCategories: (1) Never been employed, (2) sometimes worked while studying and (3) always worked while studying.
lCategories: (1) Never been employed, (2) low-intensity workers and (3) high-intensity workers.
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
DOES STUDENT WORK REALLY AFFECT EDUCATIONAL OUTCOMES? 907
possible (Baert et al., 2016). However, some pre-existing differences between working and non-working
students are unobservable in survey and administrative data and, as a consequence, cannot be controlled
for in these regressions. As mentioned in the previous subsection, this may lead to biased empirical
evidence.
A second, more advanced way of controlling for observables is through matching. The studies included
in this review that apply this method all use propensity score matching (PSM). The objective of PSM is to
compare each working student with a similar non-working student. This is achieved through a three-step
procedure (Buscha et al., 2012; Behr and Theune, 2016; Scott-Clayton and Minaya, 2016). In the first
step, for each individual in the sample the probability of working as a student is predicted based on
various covariates, that is the propensity score. Frequently used covariates in this respect are gender,
ethnicity, parental education level, socio-economic background and previous academic performance.
Next, working and non-working students are matched based on their propensity score that is students with
similar propensity scores are linked. In the final step, the educational outcomes of these linked students
are compared to each other. The matching method assumes that selection of students into student work is
random conditional on the covariates used to calculate the propensity score (‘Conditional Independence
Assumption’). However, similarly to what was argued in the previous paragraph, this assumption may
not be satisfied in practice, due to unobservable differences between working and non-working students
that cannot be used to calculate the propensity scores.
In a third approach, longitudinal data are exploited to also control for differences between student
workers and non-workers that cannot be observed in the analysed data. Most studies in this category
– especially those published in the field of economics – control for individual fixed effects (Sabia,
2009; Wenz and Yu, 2010; Darolia, 2014). By adding fixed effects (FE) to a regression model, time-
invariant unobserved heterogeneity between working and non-working students can be controlled for.
However, various authors state that it is doubtful that unobserved heterogeneity between working and
non-working students is constant over time (Oettinger, 1999; Stinebrickner and Stinebrickner, 2003).
For example Oettinger (1999) argues that the timing of college admission decisions gives students in
secondary education an incentive to increase their academic effort before these decisions are made
and reduce it afterwards. This time-varying academic effort is a potential determinant of both student
work and educational attainment for which FE models cannot control. In addition, in these models, the
parameters of interest are identified only through the within-student dimension of the data that is based
on students with variation in their work activities during the period of observation. A close alternative is
the estimation of a random effects model, as in Staff, Schulenberg and Bachman (2010). In this model,
individuals’ unobservables are integrated out as random draws from a restricted distribution instead of
being conditioned upon as FE. Other methods exploiting longitudinal data to control for unobserved
heterogeneity are event studies estimating Cox proportional hazard models (Moulin et al., 2013; Theune,
2015) and difference-in-differences (DiD) estimations – Buscha et al. (2012) combine the latter method
with matching. However, just as FE models, all these methods make assumptions about the time evolution
of the unobserved differences between workers and non-workers.
A fourth approach to control for the endogeneity of student work and later educational outcomes is
jointly modelling these outcomes and using exogenous variation in predictors of student work decisions
to identify their causal effect on educational outcomes. A popular method in this respect – frequently
used in the contributions of economists – is instrumental variable (IV) estimation. For this method, a
two-stage least squares (2SLS) regression is estimated. In the first stage, student employment is predicted
by regressing it on an IV (and other control variables). In the second stage, this prediction is used as
the independent variable explaining the educational outcome of interest. An adequate IV for student
employment is a variable that satisfies two conditions: (i) it is highly correlated with student employment
and (ii) it does not directly correlate with educational outcomes. Frequently used IVs when estimating the
impact of student employment on educational attainment are local labour market conditions (Parent, 2006;
Dustmann and van Soest, 2007; Rothstein, 2007; Beffy et al. 2010; Lee and Orazem, 2010) and interstate
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
908 NEYT ET AL.
variation in prevailing labour laws (Tyler, 2003; Rothstein, 2007; Apel et al., 2008; Lee and Orazem, 2010).
Condition (ii), in particular, cannot be easily guaranteed with respect to these instruments (Oettinger,
1999; Stinebrickner and Stinebrickner, 2003; Buscha et al., 2012). For instance, as discussed in Baert
et al. (2016), local labour market conditions during secondary or tertiary education may affect students’
decision on whether or not to drop out. Moreover, IV estimates only isolate a local average treatment effect
(LATE), that is they only capture the effect of student work for individuals who are affected by the chosen
instrument (Angrist et al., 2000). Another method in this fourth category that is widely used—across
fields—is simultaneous equation modelling (SEM). In this method, student employment, educational
outcomes and other (potentially) related outcomes are modelled as a system of regression equations (Quirk
et al., 2001; Kalenkoski and Pabilonia, 2009, 2010, 2012). Again, identification of causal relationships
between these outcomes requires that variables can be found that only predict particular outcomes while
being left out of the equations for other outcomes (‘exclusion restrictions’). Again, local labour market
conditions are often used as exclusive predictors of student work outcomes. Two final tools within this
fourth category, both of which are closely related to SEM, are the bivariate probit model used by McVicar
and McKee (2002) and the treatment model proposed by Triventi (2014), to which we return below.
A fifth and final approach is the dynamic discrete choice modelling outlined in Eckstein and Wolpin
(1999), Montmarquette, Viennot-Briot and Dagenais (2007) and Baert et al. (2017). Similarly to the fourth
approach, within dynamic discrete choice models, all relevant school and work outcomes and decisions
are jointly modelled (as discrete choices). However, the modelled outcomes are explicitly allowed to
differ for a finite number of unobserved heterogeneity types in the data. Just as in random effects models,
the distribution of these types is identified by the multiple outcomes observed for each individual. A
crucial assumption in these models is the orthogonality of the unobserved and observed – and, therefore,
included – determinants of the first modelled outcome. This is also a strong assumption.
Below, we discuss the effect of student work on educational outcomes as identified by clusters of
studies with the same methodological approach.
4. Convergences and Divergences in the Empirical Findings
In this section, we summarize the findings of studies that were published, as a journal paper or a
discussion paper, between 1997 and 2017 and that empirically investigate the relationship between
student employment and later educational outcomes. This review is the fruit of a systematic search. In
a first step, the abstracts of all articles, indexed in Web of Science or Google Scholar, including the word
groups ‘student work’, ‘student job’ or ‘student employment’, were screened regarding their relevance.
This provided us with an initial list of studies for our review. In a second step, we explored (i) the articles
included in the references of these studies and (ii) the articles citing these studies in Web of Science or
Google Scholar. This second step was re-iterated whenever an additional relevant paper was found.
A schematic overview of these studies can be found in Table 1. In Subsection 4.1, we briefly discuss the
overall non-positive impact of student work on educational behaviour and educational performance. Then,
in Subsection 4.2, we elaborate on the extent to which different methods used within and between studies
yield diverging results. This also gives an indication of the direction of the endogeneity bias discussed
in Section 3. Additionally, we indicate which studies, in our opinion, are the most convincing in terms of
identifying a causal effect and we discuss their results with extra attention. Finally, in Subsection 4.3, we
discuss moderators – in a broad sense – of the effect of student work on educational outcomes. Therefore,
in this subsection, we first discuss convergences within clusters of studies, as grouped by (i) the country
where their data were gathered and by (ii) whether they focus on student work during secondary or
during tertiary education. Then, we discuss heterogeneous effects of student employment on educational
outcomes by (iii) (educational) outcome variable, (iv) type of student job and (v) student characteristics.
Here too, we zoom in on the studies that applied the most ambitious approaches to estimate causal effects.
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
DOES STUDENT WORK REALLY AFFECT EDUCATIONAL OUTCOMES? 909
4.1 Overview of the Main Findings
A first look at Table 1 reveals that mainly a non-positive relationship between student employment
and academic outcomes is found in previous research. More specifically, 31 of the 50 studies (i.e.
62.00%) included in our review report a negative effect of student employment on educational outcomes.
Five of them explicitly highlight, however, that this effect is rather small. In addition, 12 studies (i.e.
24.00%) report both negative and neutral effects, depending on the type of educational outcome (infra,
Subsection 4.3.3), type of student job (infra, Subsection 4.3.4) or type of student (infra, Subsection 4.3.5)
considered. So, in total, 43 studies (i.e. 86.00%) provide evidence of at least some negative association.
Of the remaining seven studies, four find no significant effect and three report both negative and positive
associations.
From this first look at the literature, it appears that student employment and education are substitutes
rather than complements. However, this general picture may reflect effects driven by an endogeneity bias
and it may conceal interesting convergences and divergences that can be observed when investigating the
literature more carefully. Both of these concerns are addressed in the next two subsections.
4.2 Direction of the Endogeneity Bias
In this subsection, we explore the direction – and to some extent also the size – of the endogeneity bias
by comparing results that are obtained by using different methods. In Subsection 4.2.1, we summarize
relevant information in this respect from studies that present both elementary estimated results and results
obtained using more sophisticated methods. Then, in Subsection 4.2.2, we compare the empirical findings
for clusters of studies based on the (main) method they use – in this subsection, we follow the same
structure as in Subsection 3.2.
4.2.1 Comparing Methods within Studies
When examining studies that apply multiple methods (i.e. different methods are used to analyse the
same data), the results of these different methods vary substantially. However, these studies provide no
unambiguous conclusion on whether and to what extent more elementary models yield negatively or
positively biased effects of student employment on educational attainment. In other words, based on this
within-study evidence, it remains inconclusive whether student workers are a positively or negatively
selected subpopulation of the population of students, respectively (supra, Subsection 3.1).
More specifically, on the one hand, some studies provide evidence of a positive selection effect, that is
their results based on elementary approaches are less negative than those based on approaches controlling
for unobserved heterogeneity (Triventi, 2014). For example Stinebrickner and Stinebrickner (2003) report
both positive and neutral effects of student work on educational attainment based on OLS models and a
robustly negative effect when using an IV approach. Similarly, the OLS estimates of Tyler (2003) indicate
that student work only slightly decreases students’ math and reading scores, whereas estimates using an
IV approach provide evidence of a substantial decrease in these outcomes. Finally, Sabia (2009) finds a
positive relationship between student work and grade point average (GPA) based on OLS estimates, but
does not find a significant relationship when estimating an FE regression model.
On the other hand, Rothstein (2007) and Buscha et al. (2012) report evidence of a negative selection into
student work. In the former study, a negative impact found based on OLS regressions becomes negligible
when estimating an FE regression model and even turns completely insignificant once an IV approach is
used. The latter study reports a negative effect of part-time work on math scores when applying a matching
strategy. However, when combining this approach with a DiD strategy, taking into account unobservable
heterogeneity between working and non-working students, this negative effect disappears.
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
910 NEYT ET AL.
Throughout the other (sub)sections of this review (and in Table 1), for studies that applied multiple
methods, the outcome yielded by the most ambitious method with respect to controlling for the endogeneity
problem is the one we take into account.
4.2.2 Comparing Methods between Studies
Fifteen studies included in our review estimate elementary models (cross tabulations, variance analysis,
OLS, LPM and logit regressions) to analyse the impact of student employment on educational outcomes.
All of them report non-positive effects. More specifically, nine studies report a consistently negative
effect (Warren, 2002; Payne, 2003; Warren and Lee, 2003; Weller et al., 2003; Marsh and Kleitman,
2005; Bozick, 2007; Singh et al., 2007; Rochford et al., 2009; Beerkens et al., 2011), while six studies
report both negative and neutral effects (McNeal, 1997; Schoenhals et al., 1998; Staff and Mortimer,
2007; Derous and Ryan, 2008; Baum and Ruhm, 2016; Baert et al., 2017).
Next, five studies rely solely on a matching approach to control for the endogeneity of student work and
educational outcomes. Four of them report a negative relationship between these variables (Lee and Staff,
2007; McCoy and Smyth, 2007; Bachman et al., 2011; Behr and Theune, 2016), while one study finds
both negative and neutral effects depending on the outcome variable used (Scott-Clayton and Minaya,
2016).
Overall, these two approaches, which only control for differences between student workers and non-
workers that are observable in their data (supra, Subsection 3.2), yield non-positive results. Thirteen of
them report a consistently negative impact. This proportion does not substantially diverge from what was
found for the total set of studies, as discussed in Subsection 4.1. So again, this exercise does not allow a
firm conclusion to be drawn with respect to the direction of the selection effect in this context.
Further, ten studies exploit the longitudinal nature of their data to control for individual unobserved
heterogeneity. Four of them rely on a purely FE approach. Among them, Oettinger (1999) and Wenz and
Yu (2010) find a negative impact, while Rothstein (2007) and Sabia (2009) find no significant impact. Two
additional studies, Apel et al. (2008) and Darolia (2014), combine a FE approach with an IV approach
and both find mixed effects that is negative and neutral effects, depending on the outcome variable
considered.4Staff et al. (2010) are the only authors who rely on a RE model and find a negative impact
of hours worked per week as a student on GPA and study engagement. The two studies estimating Cox
proportional hazards models report a negative effect (Moulin et al., 2013; Theune, 2015). As mentioned
earlier, Buscha et al. (2012) combine PSM with DiD. Using this approach, no significant impact of student
employment on educational attainment is found.
Fourth, 19 of the 50 studies in our review jointly model student work and later educational outcomes,
thereby exploiting the adoption of exogenous predictors of the former outcome. Twelve of them rely for
their estimation of the causal effect of student work on educational outcomes on an IV approach. Of the 10
studies not combining this approach with a control for FE, six report a negative effect (Stinebrickner and
Stinebrickner, 2003; Tyler, 2003; Parent, 2006; DeSimone, 2008; Beffy et al., 2010; Body et al., 2014),
three report effects with diverging signs and significance (Dustmann and van Soest, 2007; Arano and
Parker, 2008; Lee and Orazem, 2010), depending on the outcome variables under investigation and type
of student work and one reports no significant effect (Rothstein, 2007). Further, of the six studies relying
on a SEM approach, two report a substantial negative effect (Kalenkoski and Pabilonia, 2009, 2012),
two a small negative effect (Singh, 1998; Kalenkoski and Pabilonia, 2010), one negative and positive
effects depending on the number of hours worked (Quirk et al., 2001) and one a neutral effect (Warren
et al., 2000). Finally, McVicar and McKee (2002) estimate a bivariate probit model and Triventi (2014)
estimates a treatment model in which the student work decision and later number of credits acquired are
jointly explained, with the unemployment rate and age only determining the first variable and a latent
factor determining both of them. They find a negative relationship between student employment and
credits or qualifications achieved, respectively.
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
DOES STUDENT WORK REALLY AFFECT EDUCATIONAL OUTCOMES? 911
Finally, three studies employ dynamic discrete choice modelling. Of these studies, only Eckstein
and Wolpin (1999) find homogeneously adverse educational outcomes for students with more intensive
working schemes, although they report the effect is rather small. In contrast, Baert et al. (2017) report a
negative effect of student work during secondary education only in a very specific case that is with respect
to tertiary education enrolment for pupils who work both during the summer and the academic year. In
addition, Montmarquette et al. (2007) report a negative effect with respect to continuing studies during
secondary education for males only.
Nine studies used data and methodological approaches that convinced us of their ability to identify
causal effects.5First, the two studies combining a FE approach with an IV approach, that is Apel et al.
(2008) and Darolia (2014) are particularly convincing, as they do not only control for time-invariant
unobserved heterogeneity between working and non-working students by using a FE approach, but also
account for the potentially time-varying endogenous relationship between student work and educational
outcomes by using IVs.6In addition, as these authors provide formal statistics to support the predictive
power of their instruments and as we are persuaded by their argumentation of why these instruments do
not impact academic success, we regard their IV approach as convincing. Second, Buscha et al. (2012)
combined PSM with a DiD approach, and were thus able to control in a flexible way for both observable
(through a flexible PSM approach) and unobservable (through a persuasive DiD approach) heterogeneity
between working and non-working students, allowing them to estimate causal effects. Third, among
the studies that relied on IVs (or exclusions restrictions) only to identify causal effects, Tyler (2003),
Rothstein (2007) and Triventi (2014) are particularly convincing in our opinion, because they too provide
a thorough (and convincing) discussion concerning the strength and exogeneity of their instruments.7
The other studies that relied on an IV approach only were less convincing in this respect.8Finally, we
classify the three studies using dynamic discrete choice models, that is Eckstein and Wolpin (1999),
Montmarquette et al. (2007) and Baert et al. (2017), as convincing because we believe these models’
assumption that observable and unobservable determinants of academic success are uncorrelated at the
moment of the first modelled outcome is reasonable.
Of these nine studies we perceive as most convincing, three (i.e. 33.33%) find evidence of a negative
effect of student employment on educational outcomes, although one study reports the effect is rather
small. Additionally, four (i.e. 44.44%) report both negative and neutral effects, depending on the type
of educational outcome (infra, Subsection 4.3.3), type of student job (infra, Subsection 4.3.4) or type of
student (infra, Subsection 4.3.5) considered. Finally, two studies report neutral effects. So, the findings of
these studies are somewhat less negative than the overall findings in Subsection 4.1, providing evidence
for a negative selection effect.9This contrasts with the results of the within-study evidence in Subsection
4.2.1, where the direction of the endogeneity bias was unclear.
4.3 Heterogeneous Effects
In this section, we report on various dimensions of heterogeneity in the empirical evidence. First, we
focus on dimensions that are fixed at the study level, that is country and education level of analysis.
So, when breaking the results down by these factors, we focus on between-study differences. Next, we
explore dimensions of heterogeneity in the relationship between student work and educational outcomes
that vary both between and within studies: type of educational outcome, type of student work and type of
student (worker).
4.3.1 By Country of Analysis
About three-quarters of the studies included in this review are conducted in North America (37 studies),
of which three are in Canada and the rest in the United States. The 13 remaining studies are carried out
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
912 NEYT ET AL.
in Europe: three in the United Kingdom (UK), two in Belgium, two in France, two in Germany, two
in Ireland, one in Estonia and one in Italy. The results are substantially more negative for studies based
on European data than for studies based on North American data. All seven studies finding either no
significant effect or both negative and positive associations are based on data for North America. In total,
21 of the 37 North American studies versus 10 of the 13 of the European studies report an overall negative
effect. However, the differences in the results for these two regions of analysis are related to the findings
in the next subsections, as studies conducted in North America examine more often the effect of student
work during secondary education on students’ test and exam scores (where results are less negative; infra
Subsection 4.3.2 and Subsection 4.3.3), whereas European studies focus more on the effect of student
work during tertiary education on outcomes such as graduating (where results are more negative; infra
Subsection 4.3.2 and Subsection 4.3.3).
The number of more convincing studies (supra, Subsection 4.2.2) in North America and Europe is in
line with the total number of studies in these regions. Indeed, seven of the nine most convincing studies
in terms of estimating a causal effect – again, a clear majority – are studies based on North American
data. When focussing solely on these nine studies, the finding that results are more negative for studies
based on European data is confirmed.
4.3.2 By Educational Level
A next comparison we make is between studies that examine students in secondary education (30
studies) and those that examine students in tertiary education (20 studies). Clearly, the evidence of a
negative relationship between student work and later educational outcomes is more pronounced in the
latter studies. For the studies on student work during secondary education, 16 report a negative effect.
Further, five studies do not find a significant relationship between student employment and educational
outcomes. Additionally, nine studies find mixed effects (including two reporting negative and positive
results), depending on the educational outcome, particular student work engagement or subset of students
considered. For the studies conducted in tertiary education, 15 find a negative effect, while only four
studies report both negative and zero effects and only one both positive and negative effects.
This pattern of a more negative effect for students in tertiary education is also found when only
considering the most convincing studies discussed in Subsection 4.2.2. Indeed, the two more ambitious
studies that find neutral effects of student work examined student work in secondary education. In contrast,
the convincing studies in tertiary education always find at least some negative relationship.
The finding that results are more adverse for students in tertiary education contrasts with our theoretical
expectations discussed in Section 2. Moreover, we are not aware of any explanation for this pattern put
forward in the literature. We believe, however, that this finding is sensible for two reasons. First, due to
the more challenging nature of studies at college or university (compared to those in secondary school),
combining study and work during tertiary education may be less feasible. Second, combining study and
work in tertiary education may change students’ attitudes toward school and intertemporal preferences.
This may cause their present discounted value of continuing school to decrease and, as a consequence, the
probability to quit tertiary education earlier than they anticipated before experiencing the student work to
increase (supra footnote 3). This reasoning is less valid for students in secondary education, as they are
to a lesser extent confronted with choosing between continuing school and joining the labour market.
4.3.3 By Educational Outcome
In this subsection, we distinguish between four categories of outcome variables used as dependent variable
in the studies included in Table 1: educational engagement, educational choices, test and exam scores and
educational attainment. While the first two categories measure students’ behaviour, the last two categories
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
DOES STUDENT WORK REALLY AFFECT EDUCATIONAL OUTCOMES? 913
measure students’ performance. Both are interrelated: behaviour affects performance (Lillydahl, 1990)
and vice versa (Triventi, 2014). Many studies combine outcomes from different categories so that summing
the number of studies per category mentioned below yields a number higher than 50.
First, nine studies consider the impact of student employment on educational engagement.10 Apart
from one study finding a neutral effect, all these studies report a negative relationship between student
work and study engagement. This is remarkable, as the effect of student employment on educational
engagement is mostly examined for students in secondary education (where the effect is, in general, less
adverse than in tertiary education, as mentioned in Subsection 4.3.2). This finding could be interpreted
as evidence of the key idea behind Zero-Sum Theory that is that time spent working crowds out time
spent on activities that enhance academic performance. However, none of the nine studies we considered
as the most convincing to estimate a causal relationship (in Subsection 4.2.2), examined the impact of
student work on educational engagement. Therefore, the almost unambiguous negative effect found on
study engagement could be due to these studies not properly controlling for (among other unobservables)
students’ primary orientation (supra, Section 2).
Second, 12 studies look at the effect of student work on educational decisions: nine focus on (not)
dropping out of school11—in Table 1 consistently referred to as the ‘positive’ continuing studies – and
three on tertiary education enrolment. Eight of them report homogeneously negative findings for this
outcome. Three other studies also find a negative relationship, but for males or particular student jobs
only. One study reports a neutral effect. Four of the nine studies we perceive to use the most ambitious
approach to estimate a causal relationship examine the effect of student work on educational choices.
Consistent with the general picture, they all report a negative effect of student employment on this
outcome, although one study finds this result only when students work both during the academic year
and during the summer and one reports this effect only for males who work more than 15 hours per
week.
The third – and most popular – category of outcome variables used is the scores that students obtain
for standardized tests or exams. Indeed, this kind of variable is used in 27 of the 50 studies in our review.
In particular, GPA is used in 24 of the studies.12 Only in 15 of the 27 studies within this category a
homogeneously negative impact is found. Additionally, 10 studies find no significant effect of student
work on test and exam scores. Two studies, Quirk et al. (2001) and Arano and Parker (2008), report both
negative and positive effects, depending on the number of hours worked. When we focus on the studies that
we labelled as most convincing, seven investigate the effect of student employment on students’ scores.
Only two of these seven studies report a negative effect (Eckstein and Wolpin, 1999; Tyler, 2003), of
which one study qualifies this effect to be small in magnitude (Eckstein and Wolpin, 1999). This provides
even stronger evidence for the finding that the negative effect of student work on test and exam scores
is less pronounced than when looking at the general picture in Subsection 4.1. Weller et al. (2003) and
Rothstein (2007) hypothesize that these less adverse outcomes with respect to test and exam scores could
be due to working students choosing less demanding courses or academic tracks. Likewise, Bachman
et al. (2011) suggest that working intensively during high school may negatively affect the quality of post-
secondary institutions attended. These adverse effects would not be reflected in the scores that students
obtain.
Finally, 20 studies include outcome variables capturing educational attainment.13 Fourteen of them
report a robustly negative impact of student work on these variables, while three report a negative impact
only for particular subsets of students, two report zero effects and one reports a positive impact. This
distribution does not deviate substantially from the overall pattern discussed in Subsection 4.1. However,
when focussing on the eight studies that focus on secondary or tertiary education graduation without
taking into account the delay in realizing this outcome, only half of them report a robustly negative
impact. The less adverse effects found on the probability of graduating unconditionally may again be due
to students choosing less demanding courses or academic tracks (Weller et al., 2003; Rothstein, 2007;
Bachman et al., 2011), which makes graduating easier and therefore more probable. Moreover, looking
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
914 NEYT ET AL.
at the probability of graduating may conceal an additional detrimental effect of student work, namely that
working students may take more years to graduate. That is why three studies consider the probability of
graduating without schooling delay as an outcome. These studies unanimously find a negative impact of
student work on this variable (Beerkens et al., 2011; Theune, 2015; Behr and Theune, 2016). Three of the
nine studies we perceived to be the most convincing examine the effect of student work on educational
attainment. One reports a neutral effect on graduating in secondary education (Baert et al., 2017), while
two report a negative effect on credits achieved in tertiary education (Darolia, 2014; Triventi, 2014), of
which one only for full-time students (Darolia, 2014). Although no convincing study directly examines
whether student work leads to study delay, the lower number of credits obtained when combining study
and work indicates that student workers may indeed prolong their time to degree completion.
From the findings in this subsection, we conclude that the apparent highly heterogeneous effect of
student work in Section 4.1 is more homogeneous within each category of outcome variables, especially
when focussing on the most convincing studies.
4.3.4 By Student Job Characteristics
Many previous studies only find adverse effects of student employment on educational attainment when
students work intensively. The threshold value of working intensively is not well defined: it ranges from
working more than 8 hours per week as a student (Body et al., 2014) to working more than 25 hours per
week (Beerkens et al., 2011; Moulin et al., 2013). Oettinger (1999), McVicar and McKee (2002), Payne
(2003), Warren and Lee (2003), Parent (2006), Bozick (2007), Lee and Staff (2007), Montmarquette et al.
(2007), Bachman et al. (2011), Beerkens et al. (2011), Moulin et al. (2013) and Body et al. (2014) only
find a negative impact on educational outcomes when students work more than a certain number of hours
per week. Quirk et al. (2001) and Arano and Parker (2008) even find a positive effect on GPA when
working less than a certain number of hours per week (while the effect reverses when working more than
this number of hours per week). These results are clearly in line with Zero-Sum Theory. Somewhat in
contrast, Staff and Mortimer (2007) find evidence of a non-linear relationship. In their study, non-workers
and steady workers have better outcomes than sporadic workers. From the most convincing studies,
Montmarquette et al. (2007) only find a negative effect on continuing studies when students work more
than 15 hours per week. The other studies do not examine (or do not report) the non-linearity of the effect
of student employment.
In addition, some other dimensions of heterogeneity in the effect of student work on educational
outcomes by student job characteristics are explored in the literature. For instance, Baert et al. (2017)
compare the effect of student employment during both the school holidays (in the summer) and the
academic year with student employment during the school holidays only. They find a negative effect of
student work on tertiary education enrolment only when students are (also) employed during the academic
year. Additionally, Darolia (2014) and Triventi (2014) show that the impact of student employment is
more adverse for full-time students and students who work intensively, respectively. These results can
be interpreted as support for Zero-Sum Theory, since there is only a negative effect when student
employment substantially coincides with schoolwork. Next, Body et al. (2014) report less adverse effects
on the probability of passing the academic year when students are employed in the public sector. They
argue that this is due to more flexible working hours in this sector, allowing students to cut back on hours
worked when work is demanding at school. Also this interpretation – if correct – can be seen as support
for Zero-Sum Theory. Finally, McNeal (1997) reports heterogeneous effects of student employment
depending on the particular type of job exercised. He finds that combining study and work has a negative
impact on the probability of continuing studies only when students work in ‘less mundane and structured’
(McNeal, 1997, p. 219) jobs such as farming, gardening or babysitting – in line with Human Capital
Theory, these jobs might be less complementary to what is learned in school.
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
DOES STUDENT WORK REALLY AFFECT EDUCATIONAL OUTCOMES? 915
4.3.5 By Student Characteristics
A final source of heterogeneity in the results is reported by authors who compare the impact of student
work on educational outcomes for distinct groups of students (based on characteristics other than their
student job). First, by measuring heterogeneous effects by gender, Dustmann and van Soest (2007) find a
negative effect of student employment on exam performance and continuing studies only for males. The
latter result is also reported by Montmarquette et al. (2007). In neither of these studies do the authors
provide an explanation for why this heterogeneity between males and females might exist. Second – but
related – Oettinger (1999) finds that student employment has a more adverse effect on GPA for (ethnic)
minorities. The author does not formulate an explanation for this.
Third, Lee and Staff (2007) compare groups of students based on their predisposition for intensive
work. They find a negative effect of student employment on the probability of staying in secondary
education only for students with low to middle propensities for working more than 20 hours per week.
They argue that for students with these low propensities for intensive work, employment may detract
from school and pull them out of school prematurely. Contrarily, this would not be the case for students
with high propensities for student work, as these already feel the push out of school and doing a student
job may not pull them away from school any further.
Finally – but related to the former two dimensions of heterogeneity – Warren (2002) and Baert
et al. (2017) measure and take into account students’ primary orientation. More specifically, Warren
(2002) confirms a key assumption underlying Primary Orientation Theory by showing that work-oriented
students both work more hours as a student worker and have worse educational outcomes. Baert et al.
(2017) directly explore the validity of Primary Orientation Theory by comparing the effect of hours of
student work on the percentage of courses passed for students with a primary orientation toward school
and students with a primary orientation toward work. They find only a negative association between
student work and educational attainment for work-oriented students.
5. Conclusion
In this paper, we have reviewed what has been written in the scientific literature on the impact of student
work on educational outcomes since 1997. In this last section, we first formulate three takeaway messages
from our review for researchers and then discuss the policy relevance of the convergences in the literature.
First, the empirical evidence summarized in this paper is, to a substantial extent, in line with Zero-
Sum Theory. Indeed, in general, we find that in previous studies mainly a non-positive effect of student
employment on educational outcomes is found and therefore that student work appears to be a substitute
for education. In particular, studies report that more intensive working schemes yield worse educational
outcomes. Moreover, the finding that student work seems to have a more adverse effect on educational
engagement than on educational performance and seems to be more adverse when being done during the
academic year (compared to during the summer holidays) and in the private sector (compared to the public
sector) can be linked to Zero-Sum Theory. Also, the observation that rather than affecting the overall
probability of graduating, student work negatively affects graduation without delay is consistent with this
theory. However, to test Zero-Sum Theory in a direct way – and therefore to test whether spending 1
hour more on student work translates into spending less time on study activities – data on students’ time
use need to be analysed. Several studies examined such data for students in secondary education (Warren
et al., 2000; Weller et al., 2003; Kalenkoski and Pabilonia, 2009, 2012) and indeed report evidence in
line with Zero-Sum Theory. However, so far no similar study on time use has been carried out for student
workers and non-workers in tertiary education. Research on this subject could uncover the extent to which
support can be found for Zero-Sum Theory for students in this type of education.
Second, as reviewed, multiple studies discussing zero (or positive) effects of student work on GPA
and graduating hypothesize that the more modest evidence for these outcomes might be due to working
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
916 NEYT ET AL.
students choosing less demanding tracks or attending lower-quality schools and colleges. However, as far
as we know, no study to date has investigated thoroughly the impact of student employment on school and
track choice. We believe this would be a perfect complement to the reviewed literature. Somewhat related,
it could also be interesting to investigate whether, in line with Human Capital Theory, the association
between student work and educational outcomes is more positive when students work in a job related to
their field of study. In this respect, Beffy et al. (2009) and Geel and Backes-Gellner (2012) examined the
impact of field-related student employment on later labour market success, and found a higher surplus of
this kind of student work.
Third, with regard to empirical approach, we believe that this body of literature would greatly benefit
from further attempts to control for the endogeneity problem inherent to estimating the effect of student
work on (educational) outcomes. In general, we advise researchers to build on the studies that we
consider the most convincing in terms of estimating causal effects. More specifically, we believe using
longitudinal data, potentially combined with an IV approach (e.g. by exploiting regional and/or temporal
variation in (child) labour laws; supra Subsection 4.2.2) could help researchers to identify causal effects.
Furthermore, authors relying on a structural model should weaken their assumption that observable and
unobservable determinants of student work are uncorrelated at the start of their model. This can be
done by estimating educational decisions and/or outcomes as early as possible in the observed students’
lifetimes. In particular, future studies should make sure to control for students’ – potentially time-varying
– primary orientation. Indeed, the two studies that do take the primary orientation of students into account
find suggestive evidence of a high correlation between primary orientation and both student work and
educational outcomes. As a consequence, we encourage future contributors to this literature to exploit data
in which students’ primary orientation is documented. Somewhat related, causal relationships between
student work and several crucial educational outcomes have not yet been convincingly estimated. More
specifically, we recommend future research to examine the causal effect of student work on educational
engagement and students’ time to degree.
Besides their academic relevance, the empirical findings reviewed in this study also have implications
for policy. Because previous studies mainly report negative effects of (substantial) student employment
on educational engagement and educational choices, bluntly encouraging student work seems not to
be justified. In general, it seems to be important that students supply labour to the extent that they do
not prioritize their student job(s) over their studies. In particular, the risks of student work that directly
interferes with their studies – such as intensive work schemes during the academic year, in particular
in sectors that limit students’ flexibility in adjusting their (study) schedule – should be made visible
to students. Nevertheless, the impact of student work on educational outcomes should be considered
together with its impact on other socio-economic outcomes, at the micro and macro level. For instance, as
mentioned in our introduction, studies examining the impact of student employment on later labour market
outcomes mainly find non-negative effects (Ruhm, 1997; Parent, 2006; Baert et al., 2016). Therefore,
more broadly, we advocate that authorities actively inform students about all assets and risks related to
student work, including its trade-off with educational outcomes.
Notes
1. Therefore not surprisingly, schools are increasingly occupied with teaching students these
competences by focussing on active learning (e.g. through group assignments, class discussions
and gamification; OECD, 2018).
2. The three reasons mentioned here for why Zero-Sum Theory may be less valid in practice may differ
in significance by type of student. For example, student work may not substantially crowd out time
spent on study-related activities for part-time students (in tertiary education), as they have greater
flexibility in their schedules (Darolia, 2014).
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
DOES STUDENT WORK REALLY AFFECT EDUCATIONAL OUTCOMES? 917
3. As mentioned above, student work may also cause a change in intertemporal preferences and future-
orientedness. As a consequence, an indirect causal effect via primary orientation is also possible.
4. Although Rothstein (2007) uses both a FE approach and an IV approach in her study, she does not
combine these two approaches in a single econometric model.
5. Stinebrickner and Stinebrickner (2003) applied a convincing IV approach in which they exploit
quasi-experimental variation in hours worked by USA students in a single college. However, we do
not consider this study as persuasive because of the limited generalizability of its findings, which the
authors themselves acknowledge. This is due to the very unique nature of the student work program
they examined, in which students are obliged to work for at least 10 hours per week in exchange for
full-tuition subsidies.
6. Apel et al. (2008) use state child labour laws as an instrument. Darolia (2014) uses lagged outcome
and student work variables, house prices and credit scores as instruments.
7. Tyler (2003) uses child labour laws as an instrument. Rothstein (2007) uses the county-level
unemployment rate, the average wage rate for teens and state laws regarding teen employment
as instruments. Triventi (2014) uses local youth unemployment rate and age as instruments.
8. Among them, Parent (2006), Dustmann and van Soest (2007), Beffy et al. (2010) and Lee and Orazem
(2010) rely on first-generation instrumental variables related to regional labour market conditions
(see Subsection 3.2), without integrating this approach into a broader, state-of-the art strategy, such
as the one used by Triventi (2014). The other studies rely on state truancy laws, parental (financial)
status, Jewish upbringing and student financial status, date of birth, lifestyle and/or nationality as
instrumental variables, without a convincing discussion to support these instruments’ adequacy
(Dustmann and van Soest, 2007; Arano and Parker, 2008; DeSimone, 2008; Beffy et al., 2010; Lee
and Orazem, 2010; Body et al., 2014).
9. On the other hand, one could argue that this finding is not surprising given that studies with FE and IV
approaches are overrepresented among these nine studies and that such approaches are less efficient
in terms of standard errors than simpler approaches. However, also compared to the other studies
relying on FE and IV approaches, the nine most convincing studies provide less negative evidence.
10. This category consists of measures of students’ time spent on homework, class attendance, class
preparation, conduct at school and tertiary education aspirations (see footnotes of Table 1 for more
details).
11. For the eight studies in secondary education, this was measured by whether or not students chose to
drop out of school at a certain age, varying between age 15 and age 18. For the one study in tertiary
education, this was measured by whether students chose to drop out after the first year. Although
this variable is closely related to graduating (as never dropping out of school leads to graduation),
these two variables are not identical, as students may choose to drop out of school at a later age than
examined by these studies.
12. The three remaining studies use as their outcome variables math and reading scores (twice) and
validated scales on course performance and grades achieved.
13. Three (nine) studies look at the probability of graduating from secondary (tertiary) education. One
(three) study (studies) investigate(s) the credits achieved in secondary (tertiary) education. Two (one)
studies (study) look(s) (among other outcomes) at the qualifications obtained in secondary (tertiary)
education. Two studies look at the time in college.
References
Angrist, J.D., Graddy, K. and Imbens, G. (2000) The interpretation of instrumental variables estimators in
simultaneous equations models with an application to the demand for fish. Review of Economic Studies
67(3): 499–527.
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
918 NEYT ET AL.
Apel, R., Bushway, S.D., Paternoster, R., Brame, R. and Sweeten, G. (2008) Using state child labor laws to
identify the causal effect of youth employment on deviant behavior and academic achievement. Journal
of Quantitative Criminology 24(4): 337–362.
Arano, K. and Parker, C. (2008) How does employment affect academic performance among college students.
Journal of Economics 34(2): 65–82.
Arulampalam, W., Naylor, R.A. and Smith, J. (2012) Am I missing something? The effects of absence from
class on student performance. Economics of Education Review 31(4): 363–375.
Ashworth, J., Hotz, V.J., Maurel, A. and Ransom, T. (2017) Changes across cohorts in wage returns to schooling
and early work experiences. NBER Working Paper Series, 24160.
Babcock, P. and Marks, M. (2011) The falling time cost of college: evidence from half a century of time use
data. Review of Economics and Statistics 93(2): 468–478.
Bachman, J.G., Staff, J., O’Malley, P.M., Schulenberg, J.E. and Freedman-Doan, P. (2011) Twelfth-grade
student work intensity linked to later educational attainment and substance use: new longitudinal evidence.
Developmental Psychology 47(2): 344–363.
Baert, S., Marx, I., Neyt, B., Van Belle, E., and Van Casteren, J. (2017) Student employment and academic
performance: an empirical exploration of the primary orientation theory. Applied Economics Letters 25(8):
547–552.
Baert, S., Neyt, B., Omey, E., and Verhaest, D. (2017) Student work, educational achievement and later
employment: a dynamic approach. IZA Discussion Paper Series, 11127.
Baert, S., Rotsaert, O., Verhaest, D., and Omey, E. (2016) Student employment and later labour market success:
no evidence for higher employment chances. Kyklos 69(3): 401–425.
Baert, S. and Vuji´
c, S. (2018) Does it pay to care? Volunteering and employment opportunities. Journal of
Population Economics 31(3): 819–836.
Baum, C.L. and Ruhm, C.J. (2016) The changing benefits of early work experience. Southern Economic Journal
83(2): 343–363.
Becker, G.S. (1964) Human Capital: A Theoretical and Empirical Analysis, with Special Referenceto Education.
New York: National Bureau of Economic Research.
Becker, G.S. (1965) A theory of the allocation of time. Economic Journal 75(299): 493–517.
Beerkens, M., M¨
agi, E., and Lill, L. (2011) University studies as a side job: causes and consequences of massive
student employment in Estonia. Higher Education 61(6): 679–692.
Beffy, M., Foug`
ere, D. and Maurel, A. (2009) L’impact du travail salari´
edes´
etudiants sur la r´
eussite et la
poursuite des ´
etudes universitaires. Economie et Statistique 422(1): 31–50.
Beffy, M., Foug`
ere, D. and Maurel, A. (2010) The effect of part-time work on post-secondary educational
attainment: new evidence from French data. IZA Discussion Paper Series, 5069.
Behr, A. and Theune, K. (2016) The causal effect of off-campus work on time to degree. Education Economics
24(2): 189–209.
Blundell, R., Dearden, L., Meghir, C., and Sianesi, B. (1999) Human capital investment: the returns from
education and training to the individual, the firm and the economy. Fiscal Studies 20(1): 1–23.
Body, K.M.D., Bonnal, L., and Giret, J.F. (2014) Does student employment really impact academic
achievement? The case of France. Applied Economics 46(25): 3061–3073.
Bozick, R. (2007) Making it through the first year of college: the role of students’ economic resources,
employment, and living arrangements. Sociology of Education 80(3): 261–285.
Bruggink, T.H. and Gambhir, V. (1996) Statistical models for college admission and enrollment: a case study
for a selective liberal arts college. Research in Higher Education 37(2): 221–240.
Buscha, F., Maurel, A., Page, L., and Speckesser, S. (2012) The effect of employment while in high school on
educational attainment: a conditional difference-in-differences approach. Oxford Bulletin of Economics
and Statistics 74(3): 380–396.
Carr, R.V., Wright, J.D., and Brody, C.J. (1996) Effects of high school work experience a decade later: evidence
from the national longitudinal survey. Sociology of Education 69(1): 66–81.
Chiswick, B.R., Lee, Y.L., and Miller, P.W. (2003) Schooling, literacy, numeracy and labour market success.
Economic Record 79(245): 165–181.
Darolia, R. (2014) Working (and studying) day and night: heterogeneous effects of working on the academic
performance of full-time and part-time students. Economics of Education Review 38(1): 38–50.
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
DOES STUDENT WORK REALLY AFFECT EDUCATIONAL OUTCOMES? 919
Derous, E. and Ryan, A.M. (2008) When earning is beneficial for learning: the relation of employment and
leisure activities to academic outcomes. Journal of Vocational Behavior 73(1): 118–131.
DeSimone, J.S. (2008) The impact of employment during school on college student academic performance.
NBER Working Paper Series, 14006.
Dustmann, C. and van Soest, A. (2007) Part-time work, school success and school leaving. Empirical Economics
32(2–3): 277–299.
Eckstein, Z. and Wolpin, K.I. (1999) Why youths drop out of high school: the impact of preferences,
opportunities, and abilities. Econometrica 67(6): 1295–1339.
Ehrenberg, R.G. and Sherman, D.R. (1987) Employment while in college, academic achievement, and post-
college outcomes: a summary of results. Journal of Human Resources 22(1): 1–23.
Geel, R. and Backes-Gellner, U. (2012) Earning while learning: when and how student employment is beneficial.
Labour 26(3): 313–340.
Hartog, J. and Oosterbeek, H. (1998) Health, wealth and happiness: why pursue a higher education? Economics
of Education Review 17(3): 245–256.
Hotz, V.J., Xu, L.C., Tienda, M., and Ahituv, A. (2002) Are there returns to the wages of young men from
working while in school? Review of Economics and Statistics 84(2): 221–236.
Kalenkoski, C.M. and Pabilonia, S.W. (2009) Does working while in high school reduce US study time? Social
Indicators Research 93(1): 117–121.
Kalenkoski, C.M. and Pabilonia, S.W. (2010) Parental transfers, student achievement, and the labor supply of
college students. Journal of Population Economics 23(2): 469–496.
Kalenkoski, C.M. and Pabilonia, S.W. (2012) Time to work or time to play: the effect of student employment
on homework, sleep, and screen time. Labour Economics 19(2): 211–221.
Lee, C. and Orazem, P.F. (2010) High school employment, school performance, and college entry. Economics
of Education Review 29(1): 29–39.
Lee, J.C. and Staff, J. (2007) When work matters: the varying impact of work intensity on high school dropout.
Sociology of Education 80(2): 158–178.
Lillydahl, J.H. (1990) Academic achievement and part-time employment of high school students. Journal of
Economic Education 21(3): 307–316.
Marsh, H.W. and Kleitman, S. (2005) Consequences of employment during high school: character building,
subversion of academic goals, or a threshold? American Educational Research Journal 42(2): 331–369.
McCoy, S. and Smyth, E. (2007) So much to do, so little time part-time employment among secondary students
in Ireland. Work, Employment & Society 21(2): 227–246.
McMorris, B.J. and Uggen, C. (2000) Alcohol and employment in the transition to adulthood. Journal of Health
and Social Behavior 41(3): 276–294.
McNeal, Jr, R.B. (1997) Are students being pulled out of high school? The effect of adolescent employment
on dropping out. Sociology of Education 70(3): 206–220.
McVicar, D. and McKee, B. (2002) Part-time work during post-compulsory education and examination
performance: help or hindrance? Scottish Journal of Political Economy 49(4): 393–406.
Montmarquette, C., Viennot-Briot, N., and Dagenais, M. (2007) Dropout, school performance, and working
while in school. Review of Economics and Statistics 89(4): 752–760.
Moulin, S., Doray, P., Laplante, B., and Street, M.C. (2013) Work intensity and non-completion of university:
longitudinal approach and causal inference. Journal of Education and Work 26(3): 333–356.
OECD. (2018) Teaching for the Future: Effective Classroom Practices to Transform Education. Paris: OECD
Publishing.
Oettinger, G.S. (1999) Does high school employment affect high school academic performance? Industrial and
Labor Relations Review 53(1): 136–151.
Parent, D. (2006) Work while in high school in Canada: its labour market and educational attainment effects.
Canadian Journal of Economics/Revue Canadienne d’ ´
Economique 39(4): 1125–1150.
Payne, J. (2003) The impact of part-time jobs in years 12 and 13 on qualification achievement. British
Educational Research Journal 29(4): 599–611.
Quirk, K.J., Keith, T.Z., and Quirk, J.T. (2001) Employment during high school and student achievement:
longitudinal analysis of national data. Journal of Educational Research 95(1): 4–10.
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
920 NEYT ET AL.
Rochford, C., Connolly, M., and Drennan, J. (2009) Paid part-time employment and academic performance of
undergraduate nursing students. Nurse Education Today 29(6): 601–606.
Rosenbaum, J.E. (2002) Beyond Empty Promises: Policies to Improve Transitions Into College and Jobs.
Washington D.C.: Office of Vocational and Adult Education, US Department of Education.
Rothstein, D.S. (2007) High school employment and youths’ academic achievement. Journal of Human
Resources 42(1): 194–213.
Ruhm, C.J. (1997) Is high school employment consumption or investment? Journal of Labor Economics 15(4):
735–776.
Sabia, J.J. (2009) School-year employment and academic performance of young adolescents. Economics of
Education Review 28(2): 268–276.
Safron, D.J., Schulenberg, J.E., and Bachman, J.G. (2001) Part-time work and hurried adolescence: the links
among work intensity, social activities, health behaviors, and substance use. Journal of Health and Social
Behavior 42(4): 425–449.
Schoenhals, M., Tienda, M., and Schneider, B. (1998) The educational and personal consequences of adolescent
employment. Social Forces 77(2): 723–761.
Scott-Clayton, J. (2012) What explains trends in labor supply among US undergraduates? National Tax Journal
65(1): 181–210.
Scott-Clayton, J. and Minaya, V. (2016) Should student employment be subsidized? Conditional counterfactuals
and the outcomes of work-study participation. Economics of Education Review 52: 1–18.
Sen, A. (1966) Education, vintage, and learning-by-doing. Journal of Human Resources 1: 3–21.
Singh, K. (1998) Part-time employment in high school and its effect on academic achievement. Journal of
Educational Research 91(3): 131–139.
Singh, K., Chang, M., and Dika, S. (2007) Effects of part-time work on school achievement during high school.
Journal of Educational Research 101(1): 12–23.
Sirin, S.R. (2005) Socioeconomic status and academic achievement: a meta-analytic review of research. Review
of Educational Research 75(3): 417–453.
Staff, J. and Mortimer, J.T. (2007) Educational and work strategies from adolescence to early adulthood:
consequences for educational attainment. Social Forces 85(3): 1169–1194.
Staff, J., Schulenberg, J.E., and Bachman, J.G. (2010) Adolescent work intensity, school performance, and
academic engagement. Sociology of Education 83(3): 183–200.
Steinberg, L. and Dornbusch, S.M. (1991) Negative correlates of part-time employment during adolescence:
replication and elaboration. Developmental Psychology 27(2): 304–313.
Steinberg, L., Fegley, S., and Dornbusch, S.M. (1993) Negative impact of part-time work on adolescent
adjustment: evidence from a longitudinal study. Developmental Psychology 29(2): 171–180.
Stinebrickner, R. and Stinebrickner, T.R. (2003) Working during school and academic performance. Journal of
Labor Economics 21(2): 473–491.
Stinebrickner, R. and Stinebrickner, T.R. (2004) Time-use and college outcomes. Journal of Econometrics
121(1): 243–269.
Stinebrickner, R. and Stinebrickner, T.R. (2008) The causal effect of studying on academic performance. BE
Journal of Economic Analysis and Policy 8(1): 1–53.
Theune, K. (2015) The working status of students and time to degree at German universities. Higher Education
70(4): 725–752.
Triventi, M. (2014) Does working during higher education affect students’ academic progression? Economics
of Education Review 41: 1–13.
Tyler, J.H. (2003) Using state child labor laws to identify the effect of school-year work on high school
achievement. Journal of Labor Economics 21(2): 381–408.
Warren, J.R. (2002) Reconsidering the relationship between student employment and academic outcomes: a
new theory and better data. Youth and Society 33(3): 366–393.
Warren, J.R. and Lee, J.C. (2003) The impact of adolescent employment on high school dropout: differences
by individual and labor-market characteristics. Social Science Research 32(1): 98–128.
Warren, J.R., LePore, P.C., and Mare, R.D. (2000) Employment during high school: consequences for students’
grades in academic courses. American Educational Research Journal 37(4): 943–969.
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
DOES STUDENT WORK REALLY AFFECT EDUCATIONAL OUTCOMES? 921
Watts, C. and Pickering, A. (2000) Pay as you learn: student employment and academic progress. Education +
Training 42(3): 129–135.
Weinstein, C.E., Palmer, D.H., and Schulte, C. (1987) Learning and Study Strategy Inventory. Clearwater: H
& H Publishing Company.
Weller, N.F., Cooper, S.P., Basen-Engquist, K., Kelder, S.K., and Tortolero, S.R. (2003) School-year
employment among high school students: effects on academic, social and physical functioning.
Adolescence 38(151): 441–459.
Wenz, M. and Yu, W. (2010) Term-time employment and the academic performance of undergraduates. Journal
of Education Finances 35(4): 358–373.
Zick, C.D. (2010) The shifting balance of adolescent time use. Youth & Society 41(4): 569–596.
Journal of Economic Surveys (2019) Vol. 33, No. 3, pp. 896–921
C
2018 John Wiley & Sons Ltd.
... While student finances and employment have often been examined for domestic students in the international context, a comprehensive literature review notes a lack of studies analyzing how the relation of the job to the area of study affects educational outcomes (Neyt et al., 2019). In Germany, no previous studies have focused on master's students. ...
... Employment can be beneficial: Students can earn money necessary to cover their living and study costs, establish work and personal contacts, learn efficient time-use and soft skills, gain (job-related) work experiences, and improve their (field-specific) language skills. Based on the human capital theory (Becker, 1964), scholars have argued that working can positively affect educational and labor market outcomes (Neyt et al., 2019;Staneva, 2020). ...
... However, students allocate their overall time to study-related activities (studying and attending classes), employment, and other activities (unpaid work, leisure time). Based on the theory of the allocation of time (Becker, 1965), many scholars derive the zero-sum hypothesis, which states that increasing the time spent in one activity necessarily decreases the time for another activity (Darolia, 2014;Neyt et al., 2019;Triventi, 2014;Vögtle & Hámori, 2020). An increase in time for employment decreases study-related time and should consequently decrease educational outcomes. ...
Article
Full-text available
Even though there is an increasing number of degree-mobile students in Europe, not much is known about the effect of student employment on academic performance and study progress for international students. International students broadly engage in student employment during their studies. They differ in several characteristics from native students (e.g., by financial situation, language skills, and time spent on studying) and are a heterogeneous group (e.g., by country of origin, educational background, and intention to remain in the destination country). This study explores whether student employment and different dimensions of employment (e.g., study-related employment, employment amounts) affect the semester grade point average and the share of achieved credit points per semester. Using the first four semesters of a longitudinal study of international students in Germany and hybrid panel models (n = 1625), the study shows that students with a higher study-related employment tendency across semesters have, on average, better semester grades. When estimating the within-student effect, it is demonstrated that changes to student employment and different employment dimensions do not change the semester grades. In contrast, starting employment or increases in employment amounts (e.g., more hours per week) decreases the share of achieved credit points per semester. However, only specific student groups (e.g., students studying mathematics, natural sciences, and engineering) experience a delay in their study progress due to higher employment intensities.
... Recent work by Baert et al. (2018) suggest that the peer review literature is inconclusive in relation to the penalty of student employment and educational performance.) Their own research, however, provided further support for the findings of a negative impact on degree outcomes as a result of student working (Behr and Theune 2016;Curtis 2007;Ford, Bosworth, and Wilson 1995;Humphrey 2006;Neyt et al. 2019;Triventi 2014). This is contrasted with a growing number of studies focused on student employability outcomes improving as a result of working alongside university, with the acquisition of transferable skills having a measurable effect on graduate employment outcomes (Qenani, MacDougall, and Sexton 2014;Rothwell, Herbert, and Rothwell 2008). ...
... Student employment literature presents a very mixed picture of the impact of student employment on academic performance. Recent work by Baert et al. (2018) suggests the peer review literature is inconclusive, while previous studies (such as Behr and Theune 2016;Neyt et al. 2019;Triventi 2014) suggest a negative impact on degree outcomes as a result of student working. Whereas research by Robotham (2012) suggested that although students can spend more time at work than studying, overall they reported more positive outcomes than negative. ...
Article
Full-text available
The student workforce plays a substantial part in several low-paying industries such as retail and hospitality, and this has grown over time. However, there has been little recent research. The usual assumption is that students compete successfully with the local labour force for low-skill, part-time jobs, but there is little evidence for this. Using results from twelve employer interviews located in two cities in the United Kingdom (Bristol and Cardiff), we reconsider employers’ perspectives on taking on students. We find that, rather than seeing the labour market as an undistinguished mass of ‘arms and legs’, employers are well aware of the pros and cons of employing students, and use this information to build flexible workforces which complement the local non-student labour supply. This fits into the well documented model of the ‘core’ and ‘periphery’ workforces. We do find evidence of indirect competition, through changes in the way jobs are advertised and filled. We also note the growth in managers who have themselves worked as students may be changing the ‘frame of reference’ of those managers, further shifting the demand for student workers in the long term.
... Furthermore, on-the-job college allows individuals to develop skills in communication, work value, and time management. Studying while working allows students to apply what they have learned in college while also increasing their orientation to achieve their career objectives (Neyt et al., 2019). D intended to expand his mother's business after graduating from college. ...
... As of, the instrument can be used for the latest generation (Generation Z), which is the final year students in higher education. Thus, far, many of the studies addressing this importance of understanding the work value in final year students and graduates (De Cooman and Dries, 2012;Shujaat, 2014;Kuron et al., 2015;Ros et al., 2015;Chi et al., 2019;Doo and Park, 2019;Hampton and Welsh, 2019;Neyt et al., 2019). However, the instrument used was not specific for students in higher education. ...
Article
Full-text available
Background One of the preferences working in the Generation Z is based on their motivational work values. The relevance of job choices with the work values will contribute to student career planning. The work value instrument among generations is one of the popular instruments used to measure final year students' work value, yet few studies of the psychometric properties of non-English language versions of this instrument. This study's objectives were to adapt a questionnaire of work value in Indonesian final year university students. Methods The number of participants in this study was 316 students in Indonesia, comprised of final year students from various majors who were selected by quota sampling. The instrument consisted of 5 dimensions of value, including leisure, extrinsic rewards, intrinsic rewards, altruistic rewards, and social rewards. The reliability analysis was performed using McDonald's Omega, the evidence of validity was obtained from test content, internal structure through confirmatory factor analysis (CFA), and evidence-based in relation to other variable has conducted the correlation between work value and career development learning using the Pearson's correlation coefficient. Results The results showed that the work values instrument had good psychometric properties, including good reliability, good content validity, and internal structure. In CFA, the two-factor structure showed satisfactory model fit. Moreover, the correlation of work value with career development learning builds stronger validity evidence on this instrument. Conclusion The adapted instrument can be used practically to identify work value preferences of final year students to help them choose a work preference and setup the career planning before graduating. The result could be of interest for the researcher in work value, motivational work, and career areas in higher education. To the best of our knowledge, there have been no reports about the adaptation of work value instruments in Indonesian final year university students.
... Furthermore, the majority of economic studies consider job experience to be a type of talent (Rosen 1972, Mincer 1974and Yamaguchi 2010. Experience, according to Raelin (1997), is the capacity to make judgments in unexpected and difficult situations. ...
Article
Full-text available
Unemployment status is considered one of the divisive economic issues. This paper aims to examine the factors such as age, gender, geographical area, educational level, college major and work experience that affect unemployment status in Lebanon. As well as predicting college majors that reduce unemployment in Lebanon. Using the survey method to collect data. This study found that individuals with higher educational levels and higher work experience are set to have higher chances to be employed. Another finding is that males have more opportunities to get hired. With respect to the geographical area, citizens that live in the capital Beirut, have more chances to enter the labor market. Individuals between the age of 24 and 37 have higher chances to get recruited. By studying the college majors, this study found that individuals that have business management, computer science, art, health, nursing, nutrition and psychology as a college major could guarantee a job opportunity more than individuals with other majors. The findings would help students in choosing a college major that helps them find a job opportunity after graduating.
... However, whereas Light (2001) suggested that working while studying yielded higher future wages, Stinebrickner and Stinebrickner (2007) noted that additional study time was associated with higher academic performance. Given that work-study and additional study time may come into conflict with one another, it is unclear how tuition and the cost of attending may affect student outcomes with significant confounding factors coming from students' choice of tracks to complete their degrees (Neyt et al. 2018), in addition to student cognitive aptitudes, which predict numerous long-term outcomes throughout life (e.g., Brown et al. 2021;Deary et al. 2007;Schmidt and Hunter 2004). ...
Article
Full-text available
When seeking to explain the eventual outcomes of a higher education experience, do the personal attributes and background factors students bring to college matter more than what the college is able to contribute to the development of the student through education or other institutional factors? Most education studies tend to simply ignore cognitive aptitudes and other student characteristics—in particular the long history of research on this topic—since the focus is on trying to assess the impact of education. Thus, the role of student characteristics has in many ways been underappreciated in even highly sophisticated quantitative education research. Conversely, educational and institutional factors are not as prominent in studies focused on cognitive aptitudes, as these fields focus first on reasoning capacity, and secondarily on other factors. We examine the variance in student outcomes due to student (e.g., cognitive aptitudes) versus institutional characteristics (e.g., teachers, schools). At the level of universities, two contemporary U.S. datasets are used to examine the proportion of variance accounted for in various university rankings and long-run salary by student cognitive characteristics and institutional factors. We find that depending upon the ways the variables are entered into regression models, the findings are somewhat different. We suggest some fruitful paths forward which might integrate the methods and findings showing that teachers and schools matter, along with the broader developmental bounds within which these effects take place.
Article
Full-text available
Education and labor market outcomes of the Russian graduates are vastly studied, including their employment status, salaries, types of universities and majors they study. However, there is a lack of research of the graduates’ typical paths in education and labor market, whether they fit the conventional trajectory high school–university–permanent employment. Another question is how social background and human capital differentiate trajectories and predict a path, which graduate opts to. In this article, we address both questions. Employing the longitudinal panel study “Trajectories in Education and Careers” we examine the nine-year trajectories of those who earned a university diploma by the age of 25. In our research, we use sequence analysis to identify trajectories and regression analysis to estimate the probability to follow each trajectory depending on individual characteristics–socio-economic status and human capital–while controlling for demographic variables. We identify nine typical trajectories, including several types of a linear path, various career-oriented trajectories, a reverse and a delayed path, unstable and nontraditional trajectories. This set of education-career paths largely corresponds to foreign studies, but the Russian graduates’ trajectories turn out to be smoother. At the same time, individual factors of human capital and socio-economic status moderate education-career paths. The probability of following trajectories with prolonged education (with a post-bachelor’s degree) is significantly higher for high achieving students. Entering early career trajectories is associated with a non–cognitive skill–openness to experience. Socio-economic status differentiates two early career trajectories: "accelerated adulthood" for those with low SES and "encouraged professionalization", which is associated with high SES. Following an unstable trajectory is not associated with either socio-economic status or academic performance (cognitive skills) but is associated with low awareness about the field of study when entering a university (ill-informed choice).
Article
The transition out of adolescence signals a period of increasing personal and social responsibility. For many this means entering postsecondary or the labor market. Previous research has demonstrated that youth who do not finish high school in four years have less favorable postsecondary and labor outcomes. However, few studies compare the postsecondary and labor market outcomes within the group of students who do not finish high school in four years. The current study uses 12 years of linked-administrative data from Maryland to present the first statewide analyses comparing postsecondary and labor market outcomes for on-time graduates, GED earners, non-completers, and late graduates. The results describe an under-researched and underserved group of vulnerable students, with implications for supporting students during high school to improve the postsecondary and labor market transition.
Article
Purpose This paper investigates the effects of working during university education on labour market outcomes of university graduates. Design/methodology/approach Based on data from three successive cohorts of graduates from the Spanish region of Catalonia, the authors estimate the effect of having worked in different types of jobs before graduation, classified according to work intensity and the match with the field of study, on the probability of being employed, having a permanent contract or having a job that requires the specific degree four years after graduation. The authors employ a multinomial endogenous treatment model that enables controlling for self-selection into pre-graduation working activities. Findings Pre-graduation work activities that are related to the field of study are generally beneficial for employability and job stability. Work experiences unmatched with the degree's content are detrimental for graduates' job–education match. Originality/value This is the first paper that jointly considers the role of work intensity (part-time vs full-time) and the relationship with the field of study in a framework that accounts for self-selection into different types of jobs. The authors also contribute to the literature by estimating the effect of pre-graduation jobs not only on the chances of being employed four years after graduation but also on two important aspects of job quality: having a permanent contract and having a job that requires the specific degree.
Article
Purpose This paper aims to study the relationship between working while studying in college/university and education mismatch among employed youth in the Zambian labour market. Design/methodology/approach The study uses data from the 2014 School-to-Work Transition Survey and a multinomial logit model to examine three education-mismatch categories: undereducated, matched and overeducated. The paper also examines heterogeneities by education level and gender and uses empirical and subjective approaches of education mismatch. Findings The evidence shows that employed youth who worked while studying have a higher likelihood of having well-matched jobs. The subgroup analysis by education level reveals no significant relationship between working while studying among employed youth with higher education (secondary and above). However, employed youth with lower education (primary and lower) are less likely to be mismatched for the job. The linkage between the education system and the labour market needs to be strengthened to support a smoother school-to-work transition for youth. Additional support to enable exposure to the right type of work during youth's college or university studies could increase job match and reduce labour market inefficiencies. Originality/value The paper provides insights into a significant challenge faced by youth in developing countries, i.e. finding a suitable job for youth's level of education.
Article
Full-text available
We investigate whether volunteering has a causal effect on individual employment opportunities. To this end, a field experiment is conducted in which volunteering activities are randomly assigned to fictitious job applications sent to genuine vacancies in Belgium. We find that volunteers are 7.3 percentage points more likely to get a positive reaction to their job applications. The volunteering premium is higher for females but invariant with respect to the number of engagements.
Article
Full-text available
This study empirically assesses the thesis that student employment only hurts academic performance for students with a primary orientation toward work (versus school). To this end, we analyse unique data on tertiary education students’ intensity of and motivation for student employment by means of a state-of-the art moderation model. We find, indeed, only a negative association between hours of student work and the percentage of courses passed for work-oriented students. This finding may explain the contradictory results in the literature neglecting this factor.
Article
Full-text available
We investigate the impact of student work experience on later hiring chances. To completely rule out potential endogeneity, we present a field experiment in which various forms of student work experience are randomly disclosed by more than 1000 fictitious graduates applying for jobs in Belgium. Theoretical mechanisms are investigated by estimating heterogeneous treatment effects by the relevance and timing of revealed student work experience. We find that neither form of student work experience enhances initial recruitment decisions. For a number of candidate subgroups (by education level and occupation type), even an adverse effect is found.
Article
We examine whether the benefits of high school work experience have changed over the last 20 years by comparing effects for the 1979 and 1997 cohorts of the National Longitudinal Survey of Youth. Our main specifications suggest that the future annual earnings benefits of working 20 h per week in the senior year of high school have fallen from 17.4% for the earlier cohort, measured in 1987–1989, to 12.1% for the later cohort, in 2008–2010. The gains have diminished by similar amounts for men and women but much more substantially for those who do not later attend college than for those who do. We further show that most of the differential between cohorts can be attributed to the way that high school employment is related to subsequent adult work experience and occupational attainment.
Article
Student employment subsidies are one of the largest types of federal employment subsidies, yet little is known about their impact. We provide a framework highlighting the likelihood of heterogeneity in program effects, depending upon whether recipients are marginal or inframarginal workers. We then utilize a matching approach to estimate the effects of the Federal Work-Study program, leveraging the fact that FWS funding varies across institutions for idiosyncratic reasons. Our results suggest that about half of FWS participants would have worked even in the absence of the subsidy; for these students, FWS reduces hours worked and improves academic outcomes, but has little impact on early post-college employment. For students who would not have worked otherwise, the pattern of effects reverses. Overall, the positive effects are strongest for subgroups who are the least likely to have access to the program, suggesting there may be gains to improved targeting of funds.
Article
Using nationally representative data from the NLSY97 and a simultaneous equations model, this paper analyzes the financial motivations for and the effects of employment on U. S. college students' academic performance. The data confirm the predictions of the theoretical model that lower parental transfers and greater costs of attending college increase the number of hours students work while in school, although students are not very responsive to these financial motivations. They also provide some evidence that greater hours of work lead to lower grade point averages (GPAs).