The job demands-resources (JD-R) model was used to examine the relationship between job
characteristics, burnout, and (other-ratings of) performance (N 146). We hypothesized that
job demands (e.g., work pressure and emotional demands) would be the most important an-
tecedents of the exhaustion component of burnout, which, in turn, would predict in-role per-
formance (hypothesis 1). In contrast, job resources (e.g., autonomy and social support) were hy-
pothesized to be the most important predictors of extra-role performance, through their
relationship with the disengagement component of burnout (hypothesis 2). In addition, we pre-
dicted that job resources would buffer the relationship between job demands and exhaustion
(hypothesis 3), and that exhaustion would be positively related to disengagement (hypothesis 4).
The results of structural equation modeling analyses provided strong support for hypotheses 1,
2, and 4, but rejected hypothesis 3. These findings support the JD-R model’s claim that job de-
mands and job resources initiate two psychological processes, which eventually affect organiza-
tional outcomes. © 2004 Wiley Periodicals, Inc.
Introduction
Although job burnout is known to negatively
affect job satisfaction and organizational
commitment, and creates such undesired be-
haviors as personnel turnover and absen-
teeism (see Lee & Ashforth, 1996, for an
overview), its relationship with an organiza-
tion’s most important outcome—namely, job
performance—has hardly received any re-
search attention. In fact, in Lee and Ash-
forth’s meta-analysis, the relationship be-
tween burnout and performance was not
even mentioned. In addition, the few studies
reported in the literature thus far have shown
inconsistent relationships between burnout
and performance, with some studies showing
the expected negative relationships (e.g.,
Bhagat, Allie, & Ford, 1995; Parker & Kulik,
1995; Wright & Cropanzano, 1998), and
others showing zero or positive relationships
(e.g., Keijsers, Schaufeli, Le Blanc, Zwerts, &
Reis-Miranda, 1995; Lazaro, Shinn, &
Robinson, 1985; Randall & Scott, 1988).
One of the reasons for these mixed findings
may be that several studies only examined
the relationship between performance and
one dimension of burnout—exhaustion (e.g.,
USING THE JOB DEMANDS-RESOURCES
MODEL TO PREDICT BURNOUT AND
PERFORMANCE
Human Resource Management, Spring 2004, Vol. 43, No. 1, Pp. 83–104
© 2004 Wiley Periodicals, Inc. Published online in Wiley InterScience (www.interscience.wiley.com).
DOI: 10.1002/hrm.
Arnold B. Bakker, Evangelia Demerouti, and Willem Verbeke
Correspondence to: Arnold B. Bakker, Utrecht University, Dept. of Social & Organizational Psychology, P.O.
Box 80.14, 3508 TC Utrecht, The Netherlands, Tel: 31 30-253-4794, Fax: 31 30-253-4718, E-mail:
A. Bakker@fss.uu.nl
<<
84 • H
UMAN R
ESOURCE MANAGEMENT
, Spring 2004
Wright & Cropanzano, 1998). In addition,
the majority of studies used self-reports and
did not distinguish between two types of per-
formance: in-role and extra-role performance.
The aim of the current study is to inves-
tigate how burnout may be related to other-
ratings of performance by using a theoretical
model that incorporates the core dimensions
of burnout and by employing adequate mea-
sures to capture in-role and extra-role perfor-
mance. We built upon the job demands-re-
sources model of burnout (Demerouti,
Bakker, Nachreiner, & Schaufeli, 2001),
which delineates how the core burnout di-
mensions, exhaustion and disengagement,
both have different etiologies in organiza-
tional environments, and may subsequently
have different effects on in-role and extra-
role performance.
Defining and Measuring Job Burnout
Burnout is a work-related stress syndrome
that was originally observed among those
who do “people work” (Maslach & Jackson,
1986). However, research of the past decade
has shown that the core dimensions of
burnout—exhaustion and cynicism or disen-
gagement from work—can be observed in
virtually any occupational group (Bakker,
Demerouti, & Schaufeli, 2002; Demerouti et
al., 2001; Leiter & Schaufeli, 1996). Demer-
outi, Bakker, Vardakou, and Kantas (2003)
have defined exhaustion as an extreme form
of fatigue as a consequence of prolonged and
intense physical, affective, and cognitive
strain caused by prolonged exposure to spe-
cific working conditions (or stressors; cf.
Aronson, Pines, & Kafry, 1983; Lee & Ash-
forth, 1993; Shirom, 1989). Disengagement
refers to distancing oneself from one’s work,
work objects (e.g., computers, recipients), or
work content (e.g., software programming,
providing services). It represents an extensive
and intensive reaction in terms of an emo-
tional, cognitive, and behavioral rejection of
the job and it delineates an occupational dis-
illusionment (cf. Freudenberger, 1974).
The third classical burnout component,
reduced personal accomplishment (Maslach
& Jackson, 1986), is excluded from our defi-
nition of burnout for several reasons. First,
there is accumulating empirical evidence
that exhaustion and cynicism (disengage-
ment) constitute the core of burnout,
whereas reduced personal accomplishment
plays a far less prominent role (Maslach,
Schaufeli, & Leiter, 2001; Shirom, 2002).
Personal accomplishment has weak relation-
ships with the two other components of
burnout as well as with hypothesized an-
tecedents and outcomes (for a meta-analysis,
see Lee & Ashforth, 1996). This supports the
notion that emotional exhaustion and disen-
gagement constitute a syndrome that is only
loosely related to personal accomplishment
(see also, Schaufeli, Bakker, Hoogduin,
Schaap, & Kladler, 2001). Second, Leiter
(1993) has argued and shown that emotional
exhaustion leads to cynicism/disengagement,
whereas feelings of reduced personal accom-
plishment develop independently. This is an-
other indication of the exceptional status of
this particular burnout dimension. Finally, it
has been suggested that personal accom-
plishment reflects an individual difference
characteristic similar to self-efficacy (Cordes
& Dougherty, 1993).
Although the Maslach Burnout Inven-
tory (MBI; Maslach, Jackson, & Leiter,
1996) is the most often used instrument to
assess burnout, in the present study, we ap-
plied an alternative measure of burnout that
can be used among occupations within and
outside human service professions. This in-
strument—called the Oldenburg Burnout
Inventory (OLBI; Demerouti et al., 2001,
2003)—conceives burnout as a syndrome of
work-related negative experiences, including
feelings of exhaustion and disengagement
from work. The reason why we chose to em-
ploy this instrument is the distinctive feature
of the OLBI compared to the MBI to include
both negatively and positively framed items.
Such a procedure is recommended by con-
ventional psychometric standards and has a
higher probability to avoid artifacts due to
acquiescence tendencies.
Burnout and Performance
Employees normally engage in two sorts of
performances: they accomplish in-role and
extra-role performance. In-role performance
... research of
the past decade
has shown
that the core
dimensions of
burnout—
exhaustion and
cynicism or
disengagement
from work—
can be observed
in virtually any
occupational
group.
Using the Job Demands-Resources Model To Predict Burnout and Performance • 85
... burnout
entraps
employees in a
negative,
vicious spiral in
which they do
not seek help or
are not prone
to strive for
changes in their
situation, and,
as a result, they
continue to
perform
ineffectively.
is defined as those officially required out-
comes and behaviors that directly serve the
goals of the organization (Motowidlo & Van
Scotter, 1994). Among other things, in-role
performance includes meeting organiza-
tional objectives and effective functioning
(Behrman & Perreault, 1984). In addition,
employees display extra-role activities (Mor-
rison, 1994). Extra-role performance is de-
fined as discretionary behaviors on the part
of an employee that are believed to directly
promote the effective functioning of an or-
ganization, without necessarily directly influ-
encing a person’s target productivity (Pod-
sakoff & MacKenzie, 1994). Examples are
the willingness to help colleagues who have
heavy workloads or the avoidance of prob-
lems with colleagues (this is also known as a
specific form of organizational citizenship
behavior; Organ & Paine, 1999).
A generally accepted notion in work psy-
chology is that job stressors tend to reduce the
individual’s capacity to exert control over their
work environment, which, in turn, is supposed
to adversely affect an individual’s ability to
function in an efficient way (Fried, Ben-David,
Tiegs, Avital, & Yeverechyahu, 1998; McGrath,
1976). While plausible, this notion has re-
ceived little empirical support. The apparent
inconsistencies in the obtained associations
between stressors and job performance have
led researchers to search for moderators of
these relationships (Fisher & Gitelson, 1983;
Fried et al., 1998). For instance, several stud-
ies have examined the potential impact of con-
textual and personal variables as moderators of
the role stressor-job performance relationship
(Fisher & Gitelson, 1983). Another possibility
is to use mediators in the aforementioned rela-
tionship that directly indicate the remained in-
dividual capacity. Burnout may be such a me-
diator since it represents an outcome of the
combined effect of several work characteristics
(Demerouti, Bakker, Nachreiner, & Schaufeli,
2000, 2001) and indicates the depletion of in-
dividual coping and energy resources (Hobfoll
& Freedy, 1993; Shirom, 1989).
Singh, Goolsby, and Rhoads (1994) pro-
vide some explanations why burnout should
affect behavioral outcomes such as job per-
formance. According to them, exhaustion di-
minishes the available energy of employees
and leads to an impairment of the efforts put
into work. Moreover, burnout entraps em-
ployees in a negative, vicious spiral in which
they do not seek help or are not prone to
strive for changes in their situation, and, as a
result, they continue to perform ineffectively.
Finally, the experience of burnout reduces
employees’ self-confidence in solving work-
related problems (e.g., Bakker, Demerouti,
Taris, Schaufeli, & Schreurs, 2003), and
therefore their performance diminishes.
While these explanations are plausible, re-
sults are somewhat less than convincing.
More specifically, Schaufeli and Enz-
mann (1998) traced five studies in which the
relationship between burnout components,
as measured by the Maslach Burnout Inven-
tory–Human Services Survey (MBI-HSS),
and self-reported performance were exam-
ined. They reanalyzed the data and found
that, on average, self-reported performance
shared 5% of its variance with emotional ex-
haustion, 4% with depersonalization, and 6%
with reduced personal accomplishment (p.
92). The studies that examined the relation-
ship between exhaustion and others-rated or
objectively assessed performance resulted in
even lower explained variance (only a meager
1%). In addition, several studies on burnout
and objective performance that included not
only exhaustion, but also depersonalization
and (reduced) personal accomplishment
(Parker & Kulik, 1995; Wright & Bonett,
1997) failed to find relationships between the
latter two MBI dimensions and performance.
Even more surprising is the finding that while
some studies showed the expected negative
relationships between burnout dimensions
and performance (e.g., Bhagat et al., 1995;
Parker & Kulik, 1995; Wright & Cropanzano,
1998), others have shown zero or positive re-
lationships (e.g., Keijsers et al., 1995; Lazaro
et al., 1985; Randall & Scott, 1988).
Should we conclude that burnout is not
systematically related to objective perfor-
mance? We believe this conclusion is prema-
ture, since there are at least two reasons why
it is difficult to reveal a relationship between
burnout and objective performance. First,
the fact that researchers can explain only a
limited amount of variance in objective per-
formance may partly be attributed to the use
86 • H
UMAN R
ESOURCE MANAGEMENT
, Spring 2004
of different sources of information, which
leads to an underestimation of the strength
of relationships (cf. Zapf, Dormann, & Frese,
1996). It may be expected that two sources
of information (e.g., a supervisor who as-
sesses performance and an employee who in-
dicates his/her burnout) have their own
unique causes of (statistically independent)
error variance. If one uses only one method,
the sources of error variance are the same.
The consequence of this is inflated correla-
tions. Second, and particularly relevant to
the present study, most researchers have not
made a distinction between in-role and extra-
role performance. For example, Wright and
Bonett (1997) used a one-item measure of
global performance as assessed by the super-
visors of 44 human service workers, and
Wright and Cropanzano (1998) used the
same measure of global performance in their
study among 52 social workers. Supervisors’
global performance ratings may be based on
both in-role and extra-role behaviors shown
by their subordinates and unrelated to any of
the burnout dimensions if in-role and extra-
role performance have different predictors.
Regarding extra-role performance, we
could only locate the study of Klein and Ver-
beke (1999), who reported that depersonaliza-
tion and reduced personal accomplishment
but not emotional exhaustion had substantial
negative correlations with extra-role perfor-
mance. It can be hypothesized that it might be
difficult for employees who experience
burnout to reduce their output or quality of
performance because of organizational sanc-
tions and reward systems. Instead, they might
choose to withhold behaviors that are discre-
tionary, such as organizational citizenship be-
havior, since this would not result in direct
consequences for themselves (Schnake,
1991). Indeed, as Schaufeli and Enzmann
(1998, p. 26) note, burned-out professionals
lose their concern for the organization and be-
come hypercritical, distrusting management,
peers, and colleagues, which corresponds, in
other words, to low extra-role performance.
The Job Demands-Resources Model
In the present study, we use the job demands-
resources (JD-R) model of burnout (Bakker,
Demerouti, De Boer, & Schaufeli, 2003; De-
merouti et al., 2000, 2001) to examine how
job characteristics and burnout contribute to
explaining variance in in-role and extra-role
performance. One central assumption of the
JD-R model is that although every occupation
may have its own specific work characteris-
tics associated with burnout, it is still possible
to model these characteristics in two broad
categories—namely, job demands and job re-
sources. Job demands refer to those physical,
psychological, social, or organizational as-
pects of the job that require sustained physi-
cal and/or psychological (cognitive and emo-
tional) effort and are therefore associated
with certain physiological and/or psychologi-
cal costs. Examples are a high work pressure,
role overload, emotional demands, and poor
environmental conditions.
Job resources refer to those physical, psy-
chological, social, or organizational aspects
of the job that are (1) functional in achieving
work goals; (2) reduce job demands and the
associated physiological and psychological
costs; or (3) stimulate personal growth and
development. Resources may be located at
the level of the organization (e.g., salary, ca-
reer opportunities, job security), interper-
sonal and social relations (e.g., supervisor
and coworker support, team climate), the or-
ganization of work (e.g., role clarity, partici-
pation in decision making), and the level of
the task (e.g., performance feedback, skill va-
riety, task significance, task identity, auton-
omy). In fact, these latter working character-
istics are the classical job characteristics in
Hackman and Oldham’s (1976) model. In
general, job demands and resources are neg-
atively related, since job demands such as a
high work pressure and emotionally demand-
ing interactions with clients may preclude
the mobilization of job resources (see
Bakker, Demerouti, De Boer et al., 2003,
Bakker, Demerouti, & Euwema, 2003b; De-
merouti et al., 2000, 2001). In a similar vein,
high job resources such as social support and
feedback may reduce job demands.
A second assumption in the JD-R model
is that working characteristics may evoke
two psychologically different processes (see
Figure 1; the numbers in the figure corre-
spond with the hypotheses). In the first
… burned-out
professionals
lose their
concern for the
organization
and become
hypercritical,
distrusting
management,
peers, and
colleagues,
which
corresponds, in
other words, to
low extra-role
performance.
Using the Job Demands-Resources Model To Predict Burnout and Performance • 87
process, demanding aspects of work (i.e.,
work overload) lead to constant overtaxing
and, in the long run, exhaustion (e.g., Lee &
Ashforth, 1996; Leiter, 1993; Wright &
Cropanzano, 1998). The literature on men-
tal fatigue may be used to explain this
process. Mental fatigue is a response of the
mind and body to the reduction in re-
sources due to mental task execution. It
warns of the increasing risk of performance
failure (Veldhuizen, Gaillard, & de Vries,
2003). Under normal circumstances, peo-
ple become tired by their everyday work ac-
tivities, but their energetical resources are
sufficient to meet the task demands. How-
ever, when a person is working under high
levels of (mental) workload and is already
fatigued (e.g., at the end of a workday),
extra energy to compensate fatigue has to
be mobilized through mental effort in order
to maintain task performance (Gaillard,
2001; Hockey, 1997; Hockey, Coles, &
Gaillard, 1986). The mobilization of extra
energy may result in (feelings of) acute fa-
tigue. A subsequent return to physiological
and emotional baseline levels is crucial. In-
complete recovery from workload demands
disrupts the energetic homeostasis, which
in turn may lead to chronic effects on
health and well-being (Frankenhaeuser,
1979; Frankenhaeuser & Johansson, 1986).
When incomplete recovery takes place, the
effects of high workload demands can accu-
mulate gradually, carrying over from one
day to the next (Craig & Cooper, 1992;
Frankenhaeuser, 1980; Frankenhaeuser &
Johansson, 1986; Gaillard, 2001; Ursin,
1980). Veldhuizen et al. (2003), using of-
fice tasks in order to simulate a working day,
found that exhausted participants (assessed
with the emotional exhaustion subscale of
the Maslach Burnout Inventory; Maslach &
Jackson, 1986) had problems investing suf-
ficient energy in their tasks. Moreover, their
performance results decreased since they
reacted more slowly and produced a smaller
number of correct responses. Exhausted
subjects seemed to be unable to perform
particularly well in the evening, although
they tried to invest more effort than their
nonexhausted counterparts.
This implies that when people become
exhausted under the influence of environ-
mental demands, they will not be able to
perform well because their energetical re-
sources are diminished. Thus, the impact
of job demands on job performance should
be mediated by the feelings of (enhanced)
Figure 1. The Job Demands-Resources Model Applied to Burnout and Performance.