White Collar Workforce Management: An
Wallace J. Hopp, Seyed M. R. Iravani and Fang Liu
Department of Industrial Engineering and Management Sciences
Northwestern University, Evanston, IL 60208, USA
Although white collar work is of vast importance to the economy, the Operations Management
(OM) literature has focused largely on traditional blue collar work. In an eﬀort to stimulate more
OM research into the design, control and management of white collar work systems, this paper
provides a systematic review of disparate streams of research relevant to understanding white collar
work from an operations perspective. Our review classiﬁes research according to its relevance to
white collar work at individual, team and organizational levels. By examining the literature in the
context of this framework, we identify gaps in our understanding of white collar work which suggest
promising research directions.
Keywords: white collar work, operations management, survey
Operations Management (OM) is concerned with the processes involved in delivering goods and services
to customers (Hopp and Spearman 2000, Shim and Siegel 1999). At the core of many of these processes
is the workforce. Indeed, the ﬁeld of OM has its roots in the labor eﬃciency studies of Frederick W.
Taylor and other champions of the Scientiﬁc Management movement of the early twentieth century.
Because these early studies focused on manufacturing and other physical tasks, the OM ﬁeld developed
a tradition of studying “blue collar” systems. The dramatic improvements in direct labor productivity
over the past several decades suggest that this line of research has been highly eﬀective.
However, in recent years, the U.S. economy has steadily shifted toward service and professional
“white collar” work, with such workers now constituting 34 percent of the workforce according to the
Bureau of Labor Statistics (BLS) (Davenport et al. 2002). Furthermore, according to the BLS, workers
in “management, business, and ﬁnancial occupations” and in “professional and related occupations”
will increase by 14.4% and 21.2%, respectively, from 2004 to 2014, which ranks them as the 3rd and 1st
fastest growing occupation categories 1. This trend suggests that future economic growth will depend
much more on improving productivity of workers in white collar work settings than on achieving
further improvements in blue collar productivity.
Despite the obvious importance of white collar work to the economy, it is much less understood
in an operations sense than is blue collar work. Well-known principles of bottleneck behavior, task
sequencing, line balancing, variability buﬀering and many others (Askin and Goldberg 2002, Hopp
and Spearman 2000) help us evaluate, improve and design blue collar work work systems. But in
white collar work systems, where tasks are less precisely deﬁned and controlled than in blue collar
systems, we do not yet have principles for guiding operations decisions. Fundamental questions remain
unanswered. For example: What is the bottleneck of a white collar work system? What are appropriate
measures of productivity? How does collaboration aﬀect performance? To answer these and many
other questions, we need a science of white collar workforce operations.
A variety of ﬁelds, including Operations Management, Economics, Sociology, Marketing, and Or-
ganizational Behavior have produced streams of research relevant to white collar work. While these
have yet to coalesce into a coherent science, research in these ﬁelds has yielded useful insights. In
this paper, we survey a wide range of research that oﬀers promise for understanding the operations of
white collar work. Our objectives are to bring together these disparate threads, provide a framework
for organizing them, and identify needs and opportunities for developing a science of white collar work.
Deﬁnition of White Collar Work
To achieve these objectives we must ﬁrst deﬁne what we mean by white collar work. Historically, the
term “white collar” has been used loosely to refer to salaried oﬃce workers, in contrast with hourly
“blue collar” manual laborers (Shirai 1983).2 Sometimes ”white collar” refers to the rank or social
status of the worker. For example, answer.com deﬁnes white collar worker as “oﬃce worker in profes-
sional, managerial, or administrative position. Such workers typically wear shirts with white collars.3”
Other deﬁnitions of white and blue collar work are based on whether the worker performs manual
work. For example, Prandy et al. (1982) used the term “white-collar” to refer to non-manual labor,
e.g., supervisors, clerks, professionals, and senior managers. Still other deﬁnition of white collar work
focused on job categories. For example, Coates (1986) divided white collar work into three categories:
clerical, professional, and managerial. Because of the nature of the work, some scholars have equated
2The root of these terms is the color of the shirts worn by the workers; oﬃce workers traditionally wore white shirts,
while laborers wore work shirts that were often blue. Relaxation of professional dress codes and colorful trends in fashion
have rendered these terms somewhat anachronistic.
white collar workers with knowledge workers (McNamar 1973, Ramirez and Nembhard 2004). In this
vein, Stamp (1995) summarized eight important aspects of white collar work: “Surfacing and aligning
values and vision,” “Thinking strategically,” “Focusing key resources, at the same time maintaining
ﬂexibility,” “Managing priorities,” “Measuring performance,” “Accepting ownership, responsibility and
accountability,” “Inﬂuencing, while maintaining interpersonal awareness,” and “Continually improving
people, products and processes.”
Although these deﬁnitions give a general sense of what constitutes white collar work and how it
diﬀers from blue collar work, they do not provide a precise or consistent statement that we can use to
focus research into the operations of white collar work. For example, Coates (1986) classiﬁes clerical
work, such as typing, as white collar work. However, typing does not have any of the eight features of
white collar work as deﬁned in Stamp (1995). Moreover, from an operations perspective, typing has
much more in common with machining (commonly thought of as “blue collar”) than with management
(commonly thought of as “white collar”). To study the operations aspects of white collar work, we
need a deﬁnition that distinguishes white and blue collar work in operationally meaningful ways.
Some researchers have argued that the old white-blue work dichotomy is obsolete (Barley and
Kunda 2001, Zuboﬀ 1988). While we agree that management practices, such as empowerment and
self-directed teams may indeed blur the distinction between white and blue collar work, we believe
there remains a fundamental distinction between the two types of work at the task level. That is,
we focus on the tasks involved in the work, (e.g., ﬁnancial consulting, operating machine tool) rather
than on the workers (e.g., ﬁnancial advisors, machine tool operators).
Viewed in this way, someone we customarily think of as blue collar worker may perform white
collar tasks (e.g., a machinist brainstorms methods for improving the yield of his operation). Con-
versely, some we normally think of as a white collar worker may perform blue collar tasks (e.g., a
professor makes her own photocopies). Hopp and Van Oyen (2004) deﬁned a task as a process that
brings together labor, entities and resources to accomplish a speciﬁed objective. In this highly general
deﬁnition, labor refers to workers (e.g, machinist, doctor, cashier, banker). An entity represents the
job being worked on (e.g., part, patient, customer, ﬁnancial transaction). Resources include anything
used by labor to carry out the activity of the task, such as equipment (e.g., machines, computers),
technology (e.g., algorithms, infrastructure systems), and intellectual property (e.g., books, reports,
A task is deﬁned by the three element - labor, entities and resources - as well as the processes that
describe how they are brought together. For our purposes, whether a task is classiﬁed as blue or white
collar depends on how it is characterized along two dimensions:
1. Intellectual vs. Physical: White collar tasks mainly involve using knowledge as a dominant
element in generating ideas, processes or solutions (Davenport and Prusak 2002), while blue
collar tasks mainly involve physical labor to perform a mechanical transformation of a material
object. For example, data analysis requires the worker to select and/or develop appropriate
models speciﬁc to each diﬀerent case by drawing on his/her expertise, statistical knowledge,
and prior experiences. In contrast, moving a batch from one machine to another in shop ﬂoor
requires physical eﬀort but demand a low level of knowledge.
2. Creative vs. Routine: White collar tasks mainly involve generation of novel solutions or com-
bination of previously unrelated ideas (Davenport and Prusak 2002, Perry-Smith and Shalley
2003, Shalley 1995), while blue collar tasks consist primarily of repetitive application of known
methods to familiar situations. For example, to formulate a new drug, researchers must design
new experiments based on their domain knowledge and creative thinking. Upon completion of
each experiment, a new set of data is collected, analyzed, and used to direct new experiments. In
contrast, sewing involves repetition of the same actions on each garment. Because the required
actions are repetitive in nature, clear procedures, which govern the work, can be speciﬁed in
advance of the arrival of the work.
To provide a reasonable correspondence with the colloquial use of the terms “blue collar” and
“white collar,” we deﬁne a blue collar task to be one that is both physical and routine. Any task
that is either intellectual or creative, we deﬁne as white collar. We illustrate this deﬁnition in Figure
1, with some examples of types of work characterized by diﬀerent positions in this two dimensional
It is important to point out that, under this deﬁnition, there is no such thing as a pure blue collar
or pure white collar job (Ramirez and Nembhard 2004). For example, driving a lift truck to move
heavy parts from one part of the factory to another is generally considered to be blue collar work.
However, while driving a lift truck is mainly physical and routine, the driver must sometimes use his
creativity to ﬁgure out how to eﬃciently load and unload large items with irregular shapes. So we
classify the task of driving parts from point A to point B as a blue collar task, but classify the task
of ﬁnding a way to transport new or unusual parts as a white collar task. Under our deﬁnition, all
workers, whether they are conventionally thought of as white or blue collar, do both white and blue
collar work (Drucker 1999). Since, as OM scholars, we are interested in the eﬃciency of operations,
we are more concerned with classifying and analyzing tasks than with classifying people. Models of
white collar tasks are the foundation for a science of white collar work.
WHITE COLLAR WORK
BLUE COLLAR WORK
consulting, legal services
Figure 1: White Collar Work vs. Blue Collar Work
The above deﬁnition raises the question of how white collar work is related to service work. One
might be tempted to classify all service work as white collar work because it does not involve heavy
physical activity. For example, the tasks carried out by a bank teller do not involve signiﬁcant work in
the physics sense. But, since these tasks are highly routine, they are neither intellectual nor creative.
Hence, in our framework, tasks such as counting money, entering transactions in a bank book, cashing
checks, etc., are predominantly physical and routine and therefore qualify as blue collar work. From
an operations standpoint, the work of a bank teller has far more in common with that of an assembly
line worker than it does with that of a lawyer or consultant.
A second distinction that is worth making is that between white collar work and knowledge work
(Davenport et al. 2002). Roughly speaking, knowledge work corresponds to the right half of Figure 1,
while production work corresponds to the left half. Any task with a high intellectual content qualiﬁes
as knowledge work. Under our deﬁnition, this also makes it white collar work. But there are also
white collar tasks that are physical and not intellectual in nature. For instance, but they require a
high level of creativity and so qualify as white collar work in our framework. Again, the work of a
surgeon has more in common with that of a lawyer than that of a janitor, so it makes sense to include
surgical tasks in the white collar category.
To build toward a science of white collar work, we follow the standard OM approach used to
model blue collar systems by starting with a simple structures, such as single-class job, single-server
(e.g., simple produce-to-order system) and extending the analysis to more complex structures, such
as multi-class, multiple-server systems. To do this, we divide our taxonomy of white collar research
into work at the individual, group, and organization levels. This allows us to compare and contrast
issues in white and blue collar work systems. In Section 3, 4, and 5, we propose generic models for
representing white collar work at individual, group, and organization level and then discuss research
relevant to elements of the models. By noting which aspects of the generic models have not been well
studied in the literature, we are able to suggest promising avenues of future research in Section 6. We
summarize our overall conclusions in Section 7.
Covering all aspects of white collar work systems, which could include issues as diverse as public
policy, education, urban development, etc., is impossible. So we restrict our goals to: (1) identifying
key streams of research that are relevant to an operations understanding of white collar work, and
(2) highlighting important papers within each stream that will help direct OM researchers to useful
sources of literature for understanding white collar work.
White Collar Work at the Individual Level
The simplest context in which to study white collar work is that of a single person carrying out tasks
independently. Examples include a doctor treating a patient, a scientist writing a research paper and
a lawyer preparing a case. Although many studies in the OM literature have addressed systems that
involve individual work (Buzacott and Shanthikumar 1993, Hopp and Spearman 2000), these often
implicitly combine workers with equipment by assuming “workers are not a major factor”, “people (i.e.,
workers) are deterministic and predictable,” “workers are stationary,” and “workers are emotionless”
(Boudreau et al. 2003). While such assumptions may be oversimpliﬁcations in blue collar settings,
they are completely unrealistic in white collar systems because white collar tasks involve knowledge
and creativity, as well as human characteristics like learning, emotion and judgment. So representing
these is a key step in modeling white collar work.
A Basic Model
To provide a conceptual framework for representing individual work, we return to the basic represen-
tation of a task in Hopp and Van Oyen (2004), which depicts tasks in terms of labor, entities and
resources. Since we are talking about individual work, the labor in these systems consists of a single
worker. The entities are the logical triggers of tasks. These could be outside requests (e.g., demands
from the boss, customer calls for service) or internally generated items (e.g., an idea for a research pa-
per, a plan for improving a system). The resources could include a broad range of physical (e.g., pen,
paper, computer) and informational (e.g., books, web sites, personal knowledge, outside expertise)
elements. Finally, a fourth element that describes an individual work system is the set of processes
that govern how the labor, entities and resources are brought together to complete tasks. These could
include sequencing/scheduling rules, incentive policies and a variety of management directives. We
illustrate this individual work system schematically in Figure 2.
outside expertise, etc.
Figure 2: White Collar Work at the Individual Level
Note that this model highlights both some similarities and some key diﬀerences between white and
blue collar work. Similarities stem from the fact that both systems exhibit queueing behavior, in which
entities pile up awaiting attention from a worker with ﬁnite capacity. This means that variability and
high utilization will cause congestion (see Hopp and Spearman (2000) for a discussion). But there are
important diﬀerences, including:
1. By our deﬁnition of white collar work, the tasks themselves are of an intellectual and/or creative
nature. Workers must accumulate suﬃcient domain knowledge before they can carry out tasks.
For example, a risk analyst must master a body of knowledge in order to understand, formulate,
and analyze risk problems. Moreover, white collar tasks rarely repeat themselves, which implies
that creativity is often important in white collar work. For example, in addition to assessing
risks in familiar settings, a risk analyst must evaluate new risk scenarios, which requires a certain
amount of creativity.
2. White collar work systems rely more heavily on knowledge-based resources. While blue collar
tasks may require informational inputs (e.g., an instruction sheet showing how parts should
be assembled), the standardized nature of the work implies that these inputs will be relatively
simple. In contrast, white collar tasks, which involve a higher level of intellectual complexity,
may rely on general information that must be processed and synthesized by the worker. For
instance, a lawyer preparing a case may have to cull through a vast backlog of precedents and
select those relevant to the case at hand.
3. Learning is slower and more central in white collar systems. The complexity of the resources and
the novelty of the tasks mean that white collar workers often have more to learn than blue collar
workers. While some models of blue collar work systems involve learning (e.g., by representing
workers as growing more productive over time), such learning dynamics are even more important
in white collar work systems. Moreover, since the skills involved may be diverse, this learning
may be correlated with other things beyond time in the position.
4. Measurement of output is more diﬃcult in white collar work systems. In blue collar systems the
outputs are primarily physical (e.g., completed assemblies, cleaned hotel rooms, painted houses).
As such, their value can be measured immediately upon completion of a task. For example, a
machining operation could go directly to a test station where it is checked for quality, so that the
value created by the machinist could be measured as the rate of acceptable parts produced per
day. But in white collar systems, the outputs often have a knowledge component. For example,
a consultant writes up an analysis of a management problem for a client. The value of such
outputs is more diﬃcult to measure. Even if client satisfaction (measured via a survey) could
be used as a quality measure for the direct deliverables (i.e., the reports), there may be indirect
value of the studies. For instance, a consulting job may produce new knowledge that will be
valuable to the consulting ﬁrm in performing future jobs. These intangible knowledge outputs
of white collar work are particularly diﬃcult to value economically until long after the task has
5. White collar work systems are much more likely to involve self-generated work. Blue collar tasks
(e.g., assembling parts, sweeping a ﬂoor, ringing up an order on a cash register) generally address
requests from the outside. But, because white collar tasks involve a higher degree of creativity,
they are not so easily standardized. Hence, it is common for creative and intellectual workers
to deﬁne at least some of their own workload. Examples include a poet turning an idea into a
poem and a consultant adding a task to a consulting job to address an issue that was revealed
by previous work.
6. Workers tend to have more discretion over processing times in white collar systems. In blue
collar systems, tasks are well-deﬁned and so come with concrete completion criteria. A casting
must be machined to speciﬁed tolerances, a room must be cleaned to stipulated standards, etc.
But in white collar systems, where work is intellectually complex and/or nonstandard, detailed
speciﬁcations are diﬃcult to provide. An engineer tasked with solving a design problem has a
general idea of what constitutes an acceptable solution. But he/she must use personal judgment
to determine when the task is complete; this decision may depend on customer needs, as well as
the engineer’s backlog of other work. Since the amount of time spent on a task is discretionary,
system utilization is not exogenously determined in white collar systems as it is in blue collar
systems. Hopp et al. (2007a) showed that this implies important diﬀerences in the operating
behavior of blue and white collar work systems.
7. Incentives are more critical. As we mentioned earlier, since tasks are intellectual and creative in
nature, workers are given more control over task processing. This greater ﬂexibility allows for
a large variation in work performance, which suggests that incentives are extremely important
in motivating worker behavior. Furthermore, a substantial amount of job satisfaction from
white collar work largely is gained through non-pecuniary means, such as peer recognition, task
complexity, exposure to smart colleagues, opportunity for self advancement, etc.. Hence, the
focus of incentives in white collar work settings should diﬀer from that in blue collar settings.
Moreover, due to the diﬃculty of measuring performance objectively, white collar incentive plans
must often be based on subjective measures of performance (e.g., staﬀ evaluations).
By describing the operations of white collar tasks in a manner that highlights the above distinctions
from blue collar work, the model in Figure 2 provides a framework for classifying research on white
collar work at the individual level. Based on our deﬁnition of white collar tasks and the above
discussion, some critical aspects of white collar tasks that are distinctive from blue collar tasks are:
creativity, discretion, learning, performance measures, incentives, and technology. In the following
subsections, we summarize streams of research that have addressed these elements.
Creativity generally refers to the ability to generate novel ideas or solutions that are appropriate to
the context (Amabile 1983a, 1996, Amabile et al. 1996, Barron and Harrington 1981). Early studies of
creativity revealed the importance of individual characteristics, such as intelligence, broad interests,
intuition, self-conﬁdence, attraction to complexity, etc., to creativity (Amabile 1983b, Barron and Har-
rington 1981, Woodman and Schoenfeldt 1989, Gough 1979). More recent studies have emphasized
the impact of task processes and organizational and social environments on creativity. One school
of thought has argued that work contexts, such as task complexity, deadlines, goal orientations, per-
ceived evaluations, and supervisory styles aﬀect worker motivation and therefore creative performance
(Oldham and Cummings 1996, Shalley 1991, 1995, Shalley et al. 2000, Chesbrough 2003). Work from
this stream of research suggests that increasing job complexity and enhancing supportive supervisory
style can improve worker creativity (Oldham and Cummings 1996). Another school of researchers
have focused on the process of creativity. Fleming and Marx (2006) argued that creativity is a process
of combining existing ideas with new ones. For example, research is a creative process implemented
by combining existing disparate knowledge streams. MacCrimmon and Wagner (1994) examined cre-
ative process through computer simulation. They proposed a creativity model in which the process of
creativity can be further divided into “problem structuring, idea generation, and evaluation”. A more
prevailing view of creativity is to treat creativity as a consequence of social exchange behaviors. Since
this view often is examined in the context of organizations, we will extensively discuss it in Section 5.
Another core diﬀerence between white and blue collar work lies in discretion, i.e., a worker’s power
to make decisions regarding processing time, task quality, task sequences, etc. Lack of prescribed
detailed operational rules requires workers to handle tasks with high degree of discretion. For example,
a consultant may determine how much time to spend writing a report based on his/her judgement of
quality; a doctor may determine when to release a patient based on the patient’s health condition.
These discretionary decisions are important because spending extra time and eﬀorts may add value to
the output by either improving the quality (e.g., spending longer time may produce a better consulting
report (Hopp et al. 2007a)), increasing the quantity (e.g., a doctor may charge more money for extra
service (Debo et al. 2004)), or both. Such discretion is less common in blue collar tasks than in white
collar tasks because blue collar work is generally straightforward and well deﬁned. Spending extra time
beyond a threshold required to complete the task does not signiﬁcantly change the output. In contrast,
in the more complex setting of white collar tasks, discretion is frequently reﬂected in task selection,
prioritization and scheduling, processing time and output quality. The prevalence of discretion in
white collar work makes it diﬃcult to apply many results from blue collar research to white collar
work systems because most of research on blue collar work systems is built on the assumption that
workers are inﬂexible or have very limited ﬂexibility (Boudreau et al. 2003, Hopp et al. 2007a).
Because task completion criteria in white collar work settings cannot be speciﬁed precisely in most