Description
This study integrates transaction cost economics and contingency theory to investigate how differences in the use of
strategic human capital are associated with the design of a firm’s management control system (MCS). The use of strategic
human capital is hypothesized to positively influence the use of personnel and nontraditional controls while
negatively influencing the use of traditional controls. Using survey data from 107 respondents, this study finds that the
data do not support the hypothesized structural equation model. An alternative model is proposed that suggests only
that the use of strategic human capital positively influences the use of personnel and nontraditional controls. This
model is supported by the data.
An empirical investigation of the relation between the use of
strategic human capital and the design of the management
control system
Sally K. Widener*
Rice University, Jones Graduate School of Management, 6100 Main Street/MS 531, Houston, TX 77002, USA
Abstract
This study integrates transaction cost economics and contingency theory to investigate how di?erences in the use of
strategic human capital are associated with the design of a ?rm’s management control system (MCS). The use of stra-
tegic human capital is hypothesized to positively in?uence the use of personnel and nontraditional controls while
negatively in?uencing the use of traditional controls. Using survey data from 107 respondents, this study ?nds that the
data do not support the hypothesized structural equation model. An alternative model is proposed that suggests only
that the use of strategic human capital positively in?uences the use of personnel and nontraditional controls. This
model is supported by the data.
#2003 Elsevier Ltd. All rights reserved.
Introduction
It is widely accepted that a ?rm’s management
control system (MCS) is designed to support its
strategy. Lang?eld-Smith (1997, p. 207) states
that, ‘‘the MCS should be tailored explicitly to
support the strategy of the business’’. Ittner and
Larcker (1997, p. 295) say, ‘‘a key assumption in
the strategic control literature is the need to align
speci?c control practices with the organization’s
chosen strategy’’. Firm-level strategy variables
have conceptualized strategic positioning in terms of
cost leadership versus di?erentiation (e.g., Porter,
1980), strategic typologies in terms of prospectors
versus defenders (e.g., Miles & Snow, 1978;
Simons, 1987) and strategic mission in terms of
build, harvest, or hold (e.g., Gupta & Govindar-
ajan, 1984). Researchers have also investigated the
relation between the MCS and operational strate-
gies such as quality (e.g., see Ittner & Larcker,
1995, 1997) and customer-focus (e.g., Abernethy
& Lillis, 1995; Perera, Harrison, & Poole, 1997).
But, Lang?eld-Smith (1997, p. 207) concludes,
‘‘our knowledge of the relationship between MCS
and strategy is limited, providing considerable
scope for further research’’.
An unexplored dimension of ?rm-level strategy
is the ?rm’s use of strategic resources that enable
the ?rm to sustain its competitive advantage (Amit
& Shoemaker, 1993). Speci?cally, this study
examines the strategic resource of human capital,
which includes the knowledge and skills of
employees in a ?rm (Barney, 1991). In today’s
competitive environment, human capital, which
enhances the knowledge of the ?rm, is a primary
strategic resource of many ?rms (e.g., Quinn,
Anderson, &Finkelstein, 1996). Thus, investigating
0361-3682/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0361-3682(03)00046-1
Accounting, Organizations and Society 29 (2004) 377–399
www.elsevier.com/locate/aos
* Tel.: +1-713-348-3596.
E-mail address: [email protected] (S.K. Widener).
how di?erences in the use of strategic human
capital are associated with the design of the MCS
is important, topical, and novel. In addition to
being the ?rst study to investigate how the use of a
strategic resource a?ects the design of the MCS, this
study makes three contributions to the literature.
First, this study investigates a combination of
controls including personnel controls, nontradi-
tional results controls, and traditional results con-
trols thus responding to criticism that a more
comprehensive MCS should be studied (Fisher,
1995; Milgrom & Roberts, 1995; Otley, 1980).
Second, prior studies have limited themselves to
investigating the a?ect that strategy, con-
ceptualized broadly (e.g., prospector vs. defender),
has on the design of the MCS (Lang?eld-Smith,
1997). This study measures four distinct attributes
of a strategic resource, which include its (1)
importance to the ?rm, (2) ambiguity (e.g., beha-
vioral uncertainty), (3) ?rm-speci?city, and (4)
spread of use throughout the ?rm. Therefore,
more in-depth insights are provided on the rela-
tion between strategy and the design of the MCS.
Finally, investigating a combination of controls
allows this study to provide insights on the trade-
o?s managers make between di?erent types of
controls when designing the MCS.
This study uses structural equation modeling
(SEM) to test the relation between the use of
strategic human capital and the design of the
MCS. In addition to testing individual path coef-
?cients, SEM provides an evaluation of the entire
model thus focusing the analysis at a macro-level
perspective (Kline, 1998, p. 13) and allowing
insights to be drawn about the MCS as a whole,
rather than simply its parts. As illustrated in Fig. 1,
this study hypothesizes that the four distinct attri-
butes of strategic human capital (importance,
behavioral uncertainty, ?rm-speci?city, and
spread of use) positively in?uence the use of per-
sonnel and nontraditional results controls and
negatively in?uence the use of traditional results
controls. Using survey data from 107 respondents,
this study ?nds that the data do not support the
hypothesized model. An alternative model is pro-
posed that suggests only that the use of strategic
human capital positively in?uences the use of per-
sonnel and nontraditional results controls (e.g.,
removes the relation with traditional results con-
trols). This model is supported by the data.
This study is organized as follows. The second
section provides an overview of strategic human
capital and the management control system. Sec-
tion three develops the underlying theory and
hypotheses. The Methods section discusses the
research method and measurement of the vari-
ables. The analyses and results are presented in the
Results section. Finally, conclusions, limitations,
and extensions are discussed in the Conclusion.
Overview of strategic human capital and the
MCS
Strategic human capital
Firms hold and control strategic resources in
order to provide the ?rm with a competitive
advantage (Amit & Shoemaker, 1993; Barney,
1991). Human capital, which includes the tacit
knowledge and training of employees within the
?rm, is one type of resource that a ?rm may use
strategically (Barney, 1991). Firms that use strate-
gic human capital face challenging management
control issues, especially since the individual, not
the ?rm, owns the knowledge (Co?, 1997). As
Roos and Roos (1997, p. 413) state ‘‘the crux is
that it is individuals, not the company, who own
and control the chief source of competitive
advantage—the knowledge of organizational
members’’. Due to this lack of ownership, ?rms
face high uncertainty when predicting employee
behavior, tenure, and performance (Co?, 1997).
To mitigate the lack of ownership of the strategic
resource, ?rms often rely heavily on teams, net-
works, and other information sharing mechanisms
in order to extract the tacit knowledge embedded
in employees, thus making it more valuable to the
?rm (Grant, 1997; Quinn et al., 1996).
Attributes of strategic human capital
A strategic resource is one that is valued and
helps the ?rm sustain its long-term competitive
advantage through its lack of imitability. Human
capital is valuable when it is important to the ?rm
378 S.K. Widener / Accounting, Organizations and Society 29 (2004) 377–399
in terms of creating e?ciencies and enabling the
?rm to be more e?ective (Barney, 1991). Human
capital is di?cult for competitors to imitate when
the tasks and work performed are ambiguous,
when there is a high degree of ?rm-speci?c
knowledge, or when the knowledge and skills of
the human capital are spread through the ?rm
(Barney, 1991; Barney & Wright, 1998). These
attributes are illustrated in column 1, Fig. 1. In
order to gain more insights into the relation
between the use of a strategic resource and the
design of the MCS, I investigate all four attributes
of strategic human capital: (1) importance, (2)
behavioral uncertainty, (3) ?rm-speci?city, and (4)
spread of the resource through the ?rm.
The ?rst attribute is managers’ beliefs regarding
the importance of their strategic human capital to
the ?rm. This includes the degree to which man-
agers believe that strategic human capital helps
‘‘enable a ?rm to conceive of or implement strate-
gies that improve its e?ciency and e?ectiveness’’
(Barney, 1991, p. 106). Although all ?rms rely to
at least some extent on human capital, ?rms vary
in the importance they place on strategic human
capital (Edwards, 1997; Quinn et al., 1996; Snell &
Dean, 1992). Lepak and Snell (1999, p. 31)
emphasize this variability when they state, ‘‘not all
employees possess knowledge and skill that are of
equal strategic importance’’ (see also Barney,
1991, p. 102).
Behavioral uncertainty, the second attribute, is
‘‘when the link between a ?rm’s resources and its
sustained competitive advantage are poorly
understood’’ (Barney, 1991, p. 109). In other
words, ?rms are unable to identify the speci?c
linkages between employees’ e?orts and resulting
output. This lack of transparency hinders compe-
titors from replicating the process.
Fig. 1. Theoretical model and hypotheses (model A).
S.K. Widener / Accounting, Organizations and Society 29 (2004) 377–399 379
The third attribute is ?rm-speci?city. Competi-
tors generally do not value other ?rms’ stock of
?rm-speci?c human capital, thus there is low
transferability of employees that possess ?rm-spe-
ci?c knowledge. Grant (1991, p. 126) states that
‘‘if ?rms can acquire [on similar terms] the
resources required for imitating competitive
advantage of a successful rival, then that rival’s
competitive advantage will be short-lived’’. In
other words, ?rms cannot sustain competitive
advantage if the resource is easily transferable
between ?rms.
The spread of strategic human capital through-
out the ?rm, which is the fourth attribute, is the
degree to which there are small or large numbers
of strategic human capital within the ?rm. This
attribute also makes it di?cult for competitors to
imitate. Barney and Wright (1998) state ‘‘. . .the
synergistic value from a large number of indivi-
duals who work together is quite costly if not
impossible for competitors to imitate.’’
MCS
The purpose of the MCS is to provide informa-
tion that is useful in decision-making, planning,
control, and evaluation (Kaplan, 1983). The MCS
is a system composed of complementary compo-
nents (Milgrom & Roberts, 1995; Otley, 1994).
This study includes personnel controls, traditional
results controls, and nontraditional results con-
trols, illustrated in Fig. 1, column 2. I chose these
types of controls because they represent opposite
ends of the control spectrum (i.e., ex ante controls
and ex post controls). Thus, this study provides a
broader investigation into the design of the MCS.
Personnel controls typically function as ex ante
control mechanisms. Snell (1992, p. 297) de?nes a
personnel control as one that ‘‘regulates the ante-
cedent conditions of performance—the knowl-
edge, skills, abilities, values and motives of
employees’’. These controls are often centered on
human resource policies that help ensure that per-
sonnel will perform at a high level and in con-
gruence with ?rm goals (Merchant, 1982; Peck,
1994). In contrast to personnel controls, results
controls ‘‘regulate results’’ serving as an ex post
control mechanism (Snell, 1992, p. 297). There are
two broad types of results controls. Firms tradi-
tionally relied on ex post controls that provided
?nancial data that, in many respects, mirrored
?nancial data reported for external purposes.
However, recently ?rms have begun incorporating
more non?nancial and operational controls into
their MCS. Ittner and Larcker (1995, pp. 1–2)
state that they
Consider traditional managerial accounting
systems to be those that provide aggregated
?nancial information relatively infrequently,
operational control based on variances from
budgeted standards, and reward systems tied
primarily to ?nancial performance. In con-
trast, nontraditional systems provide more
timely physical measures of operational per-
formance, increased provision of problem-
solving information to the workers actually
performing the job, and reward systems that
focus more on non?nancial measures.
Thus, I explore both traditional and nontradi-
tional ex post controls.
In summary, this study investigates the relation
between four attributes of strategic human capital
(importance, behavioral uncertainty, ?rm-speci?-
city, and spread) and three types of controls
(personnel, nontraditional results, and traditional
results).
Theory development and hypotheses
In developing a theoretical foundation to study
the relationship between the four attributes of a
strategic resource and the three control compo-
nents of the MCS, I draw on contingency theory
1
(e.g, Chapman, 1997; Chenhall, 2003; Fisher,
1995; Otley, 1980) and transaction cost economics
(e.g., Spekle, 2001; Williamson, 1975, 1991). I chose
these theories because they explicitly target the
design of control mechanisms. Since each theory
1
This study uses the term ‘‘contingency theory’’ while
acknowledging that some studies present the perspective that
management control studies do not rely on contingency theory,
but rather employ a contingent framework developed from
underlying organizational theories (e.g., see Chenhall, 2002, p. 2).
380 S.K. Widener / Accounting, Organizations and Society 29 (2004) 377–399
provides only part of the story for understanding
how ?rms design their MCS it is necessary to
incorporate both theories in order to develop
hypotheses for all four attributes of strategic
human capital. Thus, this study provides a more
complete explanation and investigation. As indi-
cated in Fig. 2, each theory o?ers a di?erent
perspective for understanding the design of MCSs.
Contingency theory
Control systems are designed to assist managers
in making progress towards ?rm goals and
achieving desired outcomes (Chenhall, 2003).
Control systems do not exist in a vacuum; rather,
they are in?uenced by the context within which
they operate. The premise of contingency theory is
that control systems di?er across organizations
based on underlying organizational factors (Otley,
1980). Thus accounting researchers have invoked
contingency theory when studying the relationship
between organizational factors and the design of
the MCS. One of the organizational factors stud-
ied has been the organization’s strategy and
Chenhall (2003, p. 2) states that, ‘‘perhaps the
most important new stream of literature has been
that related to the role of strategy.
2
Implicitly
based on contingency theory, the ?rst set of
hypotheses investigates the in?uence that the
importance of strategic human capital has on the
design of the MCS.
Attribute 1: importance of strategic human capital
Firms are more willing to invest in costly ex ante
personnel controls if employees are strategically
important to the ?rm (Quinn et al., 1996; Snell &
Dean, 1992). Snell and Dean (1992) point out that
a ?rm is only willing to invest in personnel con-
trols if the marginal bene?t received from the
control exceeds the marginal cost of implementing
the control. Therefore, as management perceives
its strategic human capital to be increasing in its
importance, the ?rm will be willing to invest more
in personnel controls in order to, ex ante, ?nd and
develop employees whose skills, knowledge, and
goals are congruent with the needs of the organi-
zation (Becker, 1976; Merchant, 1982; Quinn et
al., 1996).
Although there is no direct empirical evidence
supporting the alignment of personnel controls
and the use of strategic human capital, Peck
(1994), in a survey of human resource and corpo-
rate executives, provides support for the positive
association between personnel controls and pro-
spector-like ?rms. She argues that this relationship
holds because prospector-like ?rms are innovative
Fig. 2. Theoretical foundation.
2
For a review of contingency literature see Chenhall (2002),
Chapman (1997), and Fisher (1995). For a review of the asso-
ciation between MCS and strategy see Lang?eld-Smith (1997).
S.K. Widener / Accounting, Organizations and Society 29 (2004) 377–399 381
?rms that desire a long-term relationship with
their employees and instill a work culture based on
trust, cooperation, and interdependent behavior
(Miles & Snow, 1978; Peck, 1994). In addition,
Snell and Dean (1992) ?nd that advanced manu-
facturing strategies (AMS) are associated with the
use of personnel controls. AMS ?rms employ
practices such as just-in-time inventory practices
and total quality management that rely heavily on
skilled knowledge workers working together to
perform interrelated task (Snell & Dean, 1992).
Since ?rms that use strategic human capital wish
to build long-term relationships, and rely on skil-
led knowledge workers in which cooperation,
trust, and knowledge sharing is critical (Co?,
1997; Grant, 1997), it becomes apparent that that
there are some similarities between prospector-like
?rms, those following AMS, and ?rms that use
strategic human capital. Thus the theoretical links,
while somewhat indirect, are suggestive of an
association between the importance of strategic
human capital and personnel controls. This
discussion leads to the following hypothesis:
Hypothesis 1a. Use of personnel controls is posi-
tively associated with the belief that strategic
human capital is important.
It is well-accepted that MCSs should support and
complement organizational strategies (e.g., see
Lang?eld-Smith, 1997). For many ?rms this has
meant a shift from traditional controls to nontradi-
tional controls that are more closely focused on the
?rm’s organizational strategy. The change in control
systems has been supported empirically in di?erent
strategic contexts. For example, Perera et al. (1997)
argue that for ?rms following a customer-focused
strategy ?nancial measures are unable to capture
the strategic intricacies and thus the ?rm will use
non?nancial measures that speci?cally focus on
the multiple dimensions that comprise quality.
Perera et al. (1997) ?nd a positive relationship
between the use of non?nancial measures and the
use of a customer-focused strategy. Although Per-
era et al. (1997) focuses on the narrower de?nition
of ?nancial vs. non?nancial controls, the ?nding
holds when extended to traditional vs. nontradi-
tional controls. Ittner and Larker (1995) investi-
gate the association between nontraditional
controls and ?rms’ reliance on quality initiatives.
They theorize that traditional controls do not
provide management with focused information
necessary to succeed with quality initiatives, thus,
believe that ?rms will change their MCS and rely
more on nontraditional controls. Ittner and
Larcker (1995) investigate quality practices in the
auto and computer industries and provide empiri-
cal support of the association between quality
initiatives and the use of nontraditional controls.
Similar to quality and customer-focused strate-
gies, Lev (2001) suggests that as ?rms rely more on
o?-balance sheet strategic resources such as
human capital, it is likely that they will rely more
on nontraditional controls that provide informa-
tion focused on the strategic resource. An example
that illustrates the importance of shifting from
traditional to nontraditional controls is the
investment ?rms make in their employees. Firms
that rely on human capital as a strategic resource
will invest in that resource through various train-
ing and development costs (Snell & Dean, 1992).
But the investment is recorded as an expense (e.g.,
employee training) that depresses pro?tability, at
least in the short-run. Thus relying on traditional
?nancial controls will provide managers with
imperfect information for decision-making pur-
poses, and as McNair, Lynch, and Cross (1990)
suggest, may actually be counterproductive.
Instead, ?rms will rely more on nontraditional
controls that provide information and link
rewards to operational and team metrics focused
on important dimensions of strategic human capi-
tal including training, development, knowledge
sharing, turnover, and productivity (Balkcom,
Ittner & Larcker, 1997; Grant, 1997; Lank, 1997;
Lev, 2001; Spicer & Ballew, 1983; Wallman, 1995).
Balkcom et al. (1997, p. 5) state
Measures such as employee satisfaction,
voluntary turnover, employee skill develop-
ment, and employee safety have become
important indicators of the company’s suc-
cess in developing a competitive advantage.
Companies are increasingly integrating some
or all of these non?nancial measures in plan-
ning, control, and compensation decisions.
382 S.K. Widener / Accounting, Organizations and Society 29 (2004) 377–399
This discussion is summarized in the following
hypotheses:
Hypothesis 1b. Use of nontraditional results con-
trols is positively associated with the belief that
strategic human capital is important.
Hypothesis 1c. Use of traditional results controls is
negatively associated with the belief that strategic
human capital is important.
Transaction cost economics theory
Transaction cost economics (TCE) provides a
basis for developing hypotheses about how beha-
vioral uncertainty, ?rm-speci?city, and the spread
of the resource a?ect the design of the MCS. TCE
recognizes these attributes as drivers of transac-
tion costs (Williamson, 1975). A premise of TCE is
that ?rms will organize and develop governance
structures to minimize the sum of production and
transaction costs
3
(Williamson, 1970, 1975, 1991).
Management control systems are one form of
governance structure (Spicer & Ballew, 1983).
Thus MCSs will be designed such that transaction
costs arising from behavioral uncertainty, ?rm-
speci?city, and spread are minimized.
TCE is based on behavioral assumptions of
bounded rationality and opportunism (William-
son, 1975). Bounded rationality implies that
behavior is limited by imperfect cognitive pro-
cesses. Thus, contracts are incomplete. Opportu-
nism implies that individuals behave self-
interestedly, which is enabled by incomplete con-
tracts. Thus, a distinct need arises for measure-
ment and control, especially in the ex post settling
of contracts (Seal, 1993). In other words, because
future states of nature are uncertain and indivi-
duals are boundedly rational, a complete contract
cannot be speci?ed ex ante. Results must be mea-
sured and ex post the parties to the contract must
settle with one another.
Drawing on TCE, the following hypotheses
investigate the in?uence that behavioral uncer-
tainty, ?rm-speci?city, and spread have on the
design of the MCS.
Attribute 2: behavioral uncertainty
Behavioral uncertainty refers to a lack of spe-
ci?ability or programmability regarding the
actions undertaken by employees and how those
actions in?uence subsequent outcomes. Spekle
(2001) de?nes it as ‘‘the degree of speci?ability of
intended performance’’. The result is information
asymmetry since the ?rm cannot directly attribute
results and performance to an individual’s e?ort
(e.g., the individual possesses superior information
regarding his/her e?ort). TCE assumes that indi-
viduals act with self-interest
4
; thus an environment
characterized by behavioral uncertainty may
manifest itself in either adverse selection or moral
hazard (Co?, 1997). Adverse selection arises when
employees possessing information unknown to the
?rm misrepresent themselves and lead the ?rm to
make a costly decision (Baiman, 1982). This can
be seen in the labor market. Firms may o?er low
wages to hedge against the inability to observe
employees’ actual knowledge and skills. In turn,
the labor market may become disproportionately
full of low-quality job candidates. Moral hazard
arises when contracts are based on surrogate
information and the employee is able to take
action that is self-interested (Baiman, 1982). This
often results in shirking or other unwanted beha-
vior in the workplace.
Ex ante personnel controls help mitigate both
adverse selection and moral hazard. To mitigate
adverse selection, ?rms can either avoid the labor
market
5
or implement coping strategies such as
obtaining better information and strengthening
their ability to interpret labor signals (Co?, 1997).
The latter can be facilitated through an increased
emphasis on ex ante personnel controls that focus
on inputs such as recruiting, selection of new
3
Colbert and Spicer (1995, p. 425) state that ‘‘transaction
costs are the costs of engaging in transactions via a particular
arrangement or governance structure’’. Transaction costs also
occur within a ?rm. Examples of such costs are control losses
and planning di?culties (Spicer & Ballew, 1983).
4
Williamson (1985, p. 47) states that opportunism is ‘‘self-
interest seeking with guile’’. In TCE vernacular, it is a very
broad term that subsumes moral hazard and adverse selection.
5
This is an extreme control. This study assumes that this
restrictive of a control cannot be met by the ?rm.
S.K. Widener / Accounting, Organizations and Society 29 (2004) 377–399 383
employees, and developing and using multiple infor-
mation sources in the ex ante decision-making pro-
cess (Snell &Dean, 1992). To mitigate moral hazard,
?rms emphasize personnel controls since this helps
ensure ex ante that future employees will have con-
gruent goals with the organization (Williamson,
1985), and thus are less likely to take advantage of an
environment that might allow for employees to act
with self-interest (Merchant, 1998). In other words,
?rms that rely heavily on human capital will seek to
?nd individuals with characteristics (e.g., values,
goals, skills, etc.) that are congruent to the ?rm’s
culture thus minimizing opportunistic behavior (e.g.,
shirking) and related transaction costs. Abernethy
and Brownell (1997) provide related empirical
evidence and ?nd that ?rms rely more on person-
nel controls in an environment where employees
perform a variety of tasks that do not have well-
programmed routines and structures and where
there is a lack of understanding of how employee
actions in?uence outcomes (see also Chenhall,
2003). Therefore, in environments characterized
by behavioral uncertainty it is likely that ?rms will
rely on ex ante controls (e.g., personnel controls)
since they reduce transaction costs associated with
opportunism (Spicer & Ballew, 1983). This dis-
cussion supports the following hypothesis:
Hypothesis 2a. Use of personnel controls is posi-
tively associated with behavioral uncertainty of
the strategic human capital.
Since there is no speci?ed relationship between
employee inputs or actions and the resulting
?nancial outputs, organizations cannot rely on
traditional results controls. Spekle (2001, p. 432)
states ‘‘the budget [e.g., a results control] itself is a
rather insigni?cant control device in these circum-
stances’’. Instead, Spekle (2001) suggests that the
?rm may use an exploratory control environment
characterized by low reliance on traditional con-
trols such as budgeting practices and monitoring
of ?nancial measures. To illustrate the ine?ective-
ness of traditional controls consider shirking. In
an environment characterized by a low under-
standing or speci?ability of the input/output pro-
cess (e.g., high behavioral uncertainty), a MCS
that regulates results using traditional measures
(e.g., pro?tability) may foster an atmosphere of
shirking since the ?rm will not be able to hold any
one person accountable for performance. Hirst
(1983) empirically shows that when behavioral
uncertainty is high traditional controls are incom-
plete and lead to increased tension among
employees. He notes that this may result in
opportunistic behavior. Abernethy and Brownell
(1997) also provide empirical evidence that tradi-
tional controls are not e?ective in a research and
development setting characterized by lack of
knowledge of the input/output process. Chenhall
(2003) summarizes this literature stream in a pro-
position that associates high uncertainty with less
reliance on traditional accounting measures. This
discussion supports the following hypothesis:
Hypothesis 2b. Use of traditional results controls
is negatively associated with behavioral uncer-
tainty of the strategic human capital.
As just discussed, due to a lack of program-
mability and speci?ability, ?rms cannot use tradi-
tional results controls in environments
characterized by behavioral uncertainty; however,
it is likely that they will rely on nontraditional
results controls. Spekle (2001) suggests that in a
climate characterized by behavioral uncertainty,
?rms will seek to establish an environment of
commitment and congruency to general organiza-
tional goals. In addition, ?rms will seek informa-
tion that they believe may shed insights into the
true nature of the activities being performed by
individuals (Spekle, 2001, p. 431). Thus ?rms will
seek focused information speci?cally related to its
strategic choices to assist managers in assessing
performance outcomes and to use during the
negotiations or the ex post performance eval-
uation process (Seal, 1993). It is likely that non-
traditional results controls, speci?cally focused on
the ?rm’s strategic goals and objectives, may pro-
vide this information since it is often argued that
nontraditional controls provide accurate and
timely information that helps the ?rm assess
employees’ actual performance (Baiman, 1990;
Seal, 1993). For example, Caterpillar Inc. rede-
signed their control system when they realized the
control system was not providing an accurate
384 S.K. Widener / Accounting, Organizations and Society 29 (2004) 377–399
picture of performance (Hendricks, Defreitas, &
Walker, 1996). Caterpillar credits the new control
system, which employs nontraditional controls
focused on its strategy, with cost savings associated
with improved decision making.
In addition, nontraditional controls focused
speci?cally on dimensions of strategic human
capital (e.g., employee satisfaction, knowledge
sharing, teamwork, etc.) may reduce future trans-
action costs. For example, by collecting data on
employee satisfaction, ?rms may better commu-
nicate their objectives and create alignment within
the ?rm focused on organizational goals and
objectives. This may help in establishing a corpo-
rate culture conducive to high employee e?ort
thus minimizing costs associated with moral
hazard. Furthermore, turnover may be minimized
which reduces di?culties associated with adverse
selection (Seal, 1993; Co?, 1997). This discussion
suggests that in an environment characterized by
behavioral uncertainty, ?rms will rely on ex post
controls that are nontraditional in nature.
Hypothesis 2c. Use of nontraditional results con-
trols is positively associated with behavioral
uncertainty of the strategic human capital.
Attribute 3: ?rm-speci?city
Firm-speci?city is the skills and knowledge pos-
sessed by employees that are speci?c to a unique
?rm (Co?, 1997). Often, ?rm-speci?city is referred
to as the extent to which specialized or non-
redeployable investments are needed to support an
exchange (Williamson, 1975). Similar to behavioral
uncertainty, ?rm-speci?city may facilitate opportu-
nistic behavior since ?rm-speci?c human capital
possesses idiosyncratic skills and knowledge that
others are often unable to observe (Co?, 1997).
Therefore, Co? (1997, p. 378) concludes that ?rm-
speci?city results in the same agency problems as
does behavioral uncertainty e.g., adverse selection
and moral hazard. Furthermore, the extent to
which opportunistic behavior is tolerated increases
since the market does not provide protection to
parties that invest in relational speci?c assets
(Baiman, 1982). Thus, there is an increased need
for a well-designed MCS that is heavily focused on
ex ante personnel controls and ex post nontradi-
tional results control. The former control ensures
that employees with similar goals, ethics, and mor-
als are brought into the organization. The latter
control ensures that there is a monitoring system in
place that focuses on information targeted speci?-
cally to the strategic resource. Traditional results
controls are not relied on as they are not e?ective in
this environment. These latter statements are
developed more fully in the preceding section and
thus not repeated here (see the previous section).
This discussion suggests the following hypotheses:
Hypothesis 3a. Use of personnel controls is
positively associated with ?rm-speci?city of the
strategic human capital.
Hypothesis 3b. Use of nontraditional results con-
trols is positively associated with ?rm-speci?city of
the strategic human capital.
Hypothesis 3c. Use of traditional results controls is
negatively associated with ?rm-speci?city of the
strategic human capital.
Attribute 4: spread of strategic human capital
A primary tenet of TCE is cost minimization
(Williamson, 1991). TCE argues that ?rms need to
align governance structures with the drivers of
transaction costs in order to minimize costs and
thus economize. One such driver of transaction
costs is ‘‘frequency’’, which this paper considers to
be the spread of the strategic human capital
through the ?rm.
As discussed earlier, a traditional MCS focuses
on aggregated ?nancial accounting information
and has strong links to the budgetary system (Itt-
ner & Larcker, 1995). A more sophisticated man-
agement control system is one that is tailored
solely to the needs of the organization and aligned
closely with its strategy. But changing from a tra-
ditional MCS to a more sophisticated MCS can be
costly since ?rms incur costs when designing and
implementing a governance structure, e.g., the
MCS (Williamson, 1970, 1975, 1991). Therefore,
the ?rm must consider whether a specialized MCS
can be used to capacity, or whether there will be
much unused capacity that may make a sophisti-
cated MCS unduly costly (Williamson, 1985).
S.K. Widener / Accounting, Organizations and Society 29 (2004) 377–399 385
The size of the spread of strategic human capital
throughout the ?rm helps determine whether the
bene?ts from the MCS will outweigh the design
and implementation costs of the MCS (William-
son, 1975). If there are large numbers of strategic
human capital it is likely that the ?rm will invest in
a more complex governance structure. If strategic
human capital is concentrated (e.g., at the
extreme, the human capital is localized within one
top employee), then the ?rm is not likely to invest
in a complex governance structure since they may
not recoup the costs of designing and implement-
ing a more complex MCS. Therefore, as the num-
bers of strategic human capital increase, the ?rm
will be more willing to invest in personnel controls
designed to ?nd skilled, knowledgeable employees
(e.g., Becker, 1976; Snell & Dean, 1992; William-
son, 1991). In addition, ?rms will invest in non-
traditional controls more closely aligned with its
strategy (e.g., Lang?eld-Smith, 1997). Further-
more, it is likely that ?rms will decrease their
reliance on traditional results controls since
there are limits on cognitive capacity (William-
son, 1975). Simply increasing the information
provided to managers through the implemen-
tation of additional controls can lead to costs
associated with information overload. Therefore,
to remain in equilibrium and satisfy a cost
minimizing objective (Williamson, 1985) ?rms
will likely trade-o? the cost of traditional con-
trols (that are not speci?cally focused on its
strategic choice) for nontraditional and person-
nel controls (that are speci?cally focused on its
strategic choice). This discussion supports the
following hypotheses:
Hypothesis 4a. Use of personnel controls is posi-
tively associated with the spread of strategic
human capital throughout a ?rm.
Hypothesis 4b. Use of nontraditional results con-
trols is positively associated with the spread of
strategic human capital throughout the ?rm.
Hypothesis 4c. Use of traditional results controls is
negatively associated with the spread of strategic
human capital throughout the ?rm.
Methods
Research design and sample
Both Compustat and survey data are used as
sources of data. Firms that report sales for
between one and ?ve four-digit SIC codes in the
same division code are included. This ensures that
?rms conduct business in only one major SIC code
division.
6
To enable analysis of non-response bias
and to validate the variable measures, selected
?rms are required to have sales from 1993 to 1996
and depreciation expense for 1996 on the Compu-
stat ?le. Firms are also required to have at least
250 employees. Eliminating small ?rms is neces-
sary since they are less likely to have a formal
control system.
After deleting ?rms that are either used in the
pretest or foreign-owned, the remaining popula-
tion is 1, 662 ?rms, of which 800 were randomly
surveyed. The largest concentration of ?rms
(43%) is classi?ed as manufacturing (SIC codes
2000-3999). Other concentrated segments include
?nancial services ?rms (17%), other service ?rms
(13%), and transportation, communication and
utilities (11%).
Research method
The design and administration of the survey
accords with the ‘‘total design approach’’ advo-
cated by Dillman (1978).
7
As a pretest, a survey
and cover letter were sent to 250 randomly selec-
ted ?rms. In addition to asking for responses, both
the survey and cover letter asked respondents to
comment on clarity, understanding, ambiguity,
etc. Four in-depth visits to ?rms were also con-
ducted in order to gain a better understanding of
the design of the MCS and use of human capital
as a strategic resource, and to help in designing the
6
This screening technique attempts to remove highly diver-
si?ed companies from the sample. Highly diversi?ed companies
may follow multiple strategies thus adding considerable noise
to the measures.
7
The Total Design Method involves an ‘‘identi?cation of
each aspect of the survey process that may a?ect response
quality or quantity’’ (Dillman, 1978, p. 2). It is a very detailed,
step-by-step approach to conducting a quality survey.
386 S.K. Widener / Accounting, Organizations and Society 29 (2004) 377–399
?nal instrument. Based on the results of the pret-
est (30 respondents) several changes were made to
the ?nal survey. Upon revision, the survey along
with a personalized cover letter and a stamped
return envelope was sent to the chief ?nancial
o?cer of 800 ?rms. As an incentive to respond,
the participants were promised a summary of the
results. Three follow-up mailings, along with a
postcard reminder, were sent to the sample selection
of ?rms.
The mailing process resulted in 118 responses
(15% response rate), which is below the average of
20% that Young (1996) reports for comparable
surveys to CFOs. One reason for the lower
response rate is the asking of multiple questions
for each construct. This study chose to increase
construct validity at the potential risk of incurring
a lower response rate. As a result, non-response
bias is rigorously investigated. The pattern of SIC
classi?cations for respondents mirrors both the
sample and the Compustat population. The two
largest groups of respondent ?rms are either man-
ufacturers (48 ?rms) or ?nancial services ?rms (19
?rms). Non-response bias is investigated by com-
paring respondents to non-respondents. The
respondent ?rms are further sub-divided into
early, middle and late respondents based on their
return date. There are no statistically signi?cant
di?erences between any classi?cations of respon-
dent ?rms for 1997 sales, total assets, number of
employees and depreciation expense. Early
respondents are also compared to late respondents
for all constructs of interest. The results are
reported in Table 1. Early respondents are sig-
ni?cantly higher (P
This study integrates transaction cost economics and contingency theory to investigate how differences in the use of
strategic human capital are associated with the design of a firm’s management control system (MCS). The use of strategic
human capital is hypothesized to positively influence the use of personnel and nontraditional controls while
negatively influencing the use of traditional controls. Using survey data from 107 respondents, this study finds that the
data do not support the hypothesized structural equation model. An alternative model is proposed that suggests only
that the use of strategic human capital positively influences the use of personnel and nontraditional controls. This
model is supported by the data.
An empirical investigation of the relation between the use of
strategic human capital and the design of the management
control system
Sally K. Widener*
Rice University, Jones Graduate School of Management, 6100 Main Street/MS 531, Houston, TX 77002, USA
Abstract
This study integrates transaction cost economics and contingency theory to investigate how di?erences in the use of
strategic human capital are associated with the design of a ?rm’s management control system (MCS). The use of stra-
tegic human capital is hypothesized to positively in?uence the use of personnel and nontraditional controls while
negatively in?uencing the use of traditional controls. Using survey data from 107 respondents, this study ?nds that the
data do not support the hypothesized structural equation model. An alternative model is proposed that suggests only
that the use of strategic human capital positively in?uences the use of personnel and nontraditional controls. This
model is supported by the data.
#2003 Elsevier Ltd. All rights reserved.
Introduction
It is widely accepted that a ?rm’s management
control system (MCS) is designed to support its
strategy. Lang?eld-Smith (1997, p. 207) states
that, ‘‘the MCS should be tailored explicitly to
support the strategy of the business’’. Ittner and
Larcker (1997, p. 295) say, ‘‘a key assumption in
the strategic control literature is the need to align
speci?c control practices with the organization’s
chosen strategy’’. Firm-level strategy variables
have conceptualized strategic positioning in terms of
cost leadership versus di?erentiation (e.g., Porter,
1980), strategic typologies in terms of prospectors
versus defenders (e.g., Miles & Snow, 1978;
Simons, 1987) and strategic mission in terms of
build, harvest, or hold (e.g., Gupta & Govindar-
ajan, 1984). Researchers have also investigated the
relation between the MCS and operational strate-
gies such as quality (e.g., see Ittner & Larcker,
1995, 1997) and customer-focus (e.g., Abernethy
& Lillis, 1995; Perera, Harrison, & Poole, 1997).
But, Lang?eld-Smith (1997, p. 207) concludes,
‘‘our knowledge of the relationship between MCS
and strategy is limited, providing considerable
scope for further research’’.
An unexplored dimension of ?rm-level strategy
is the ?rm’s use of strategic resources that enable
the ?rm to sustain its competitive advantage (Amit
& Shoemaker, 1993). Speci?cally, this study
examines the strategic resource of human capital,
which includes the knowledge and skills of
employees in a ?rm (Barney, 1991). In today’s
competitive environment, human capital, which
enhances the knowledge of the ?rm, is a primary
strategic resource of many ?rms (e.g., Quinn,
Anderson, &Finkelstein, 1996). Thus, investigating
0361-3682/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0361-3682(03)00046-1
Accounting, Organizations and Society 29 (2004) 377–399
www.elsevier.com/locate/aos
* Tel.: +1-713-348-3596.
E-mail address: [email protected] (S.K. Widener).
how di?erences in the use of strategic human
capital are associated with the design of the MCS
is important, topical, and novel. In addition to
being the ?rst study to investigate how the use of a
strategic resource a?ects the design of the MCS, this
study makes three contributions to the literature.
First, this study investigates a combination of
controls including personnel controls, nontradi-
tional results controls, and traditional results con-
trols thus responding to criticism that a more
comprehensive MCS should be studied (Fisher,
1995; Milgrom & Roberts, 1995; Otley, 1980).
Second, prior studies have limited themselves to
investigating the a?ect that strategy, con-
ceptualized broadly (e.g., prospector vs. defender),
has on the design of the MCS (Lang?eld-Smith,
1997). This study measures four distinct attributes
of a strategic resource, which include its (1)
importance to the ?rm, (2) ambiguity (e.g., beha-
vioral uncertainty), (3) ?rm-speci?city, and (4)
spread of use throughout the ?rm. Therefore,
more in-depth insights are provided on the rela-
tion between strategy and the design of the MCS.
Finally, investigating a combination of controls
allows this study to provide insights on the trade-
o?s managers make between di?erent types of
controls when designing the MCS.
This study uses structural equation modeling
(SEM) to test the relation between the use of
strategic human capital and the design of the
MCS. In addition to testing individual path coef-
?cients, SEM provides an evaluation of the entire
model thus focusing the analysis at a macro-level
perspective (Kline, 1998, p. 13) and allowing
insights to be drawn about the MCS as a whole,
rather than simply its parts. As illustrated in Fig. 1,
this study hypothesizes that the four distinct attri-
butes of strategic human capital (importance,
behavioral uncertainty, ?rm-speci?city, and
spread of use) positively in?uence the use of per-
sonnel and nontraditional results controls and
negatively in?uence the use of traditional results
controls. Using survey data from 107 respondents,
this study ?nds that the data do not support the
hypothesized model. An alternative model is pro-
posed that suggests only that the use of strategic
human capital positively in?uences the use of per-
sonnel and nontraditional results controls (e.g.,
removes the relation with traditional results con-
trols). This model is supported by the data.
This study is organized as follows. The second
section provides an overview of strategic human
capital and the management control system. Sec-
tion three develops the underlying theory and
hypotheses. The Methods section discusses the
research method and measurement of the vari-
ables. The analyses and results are presented in the
Results section. Finally, conclusions, limitations,
and extensions are discussed in the Conclusion.
Overview of strategic human capital and the
MCS
Strategic human capital
Firms hold and control strategic resources in
order to provide the ?rm with a competitive
advantage (Amit & Shoemaker, 1993; Barney,
1991). Human capital, which includes the tacit
knowledge and training of employees within the
?rm, is one type of resource that a ?rm may use
strategically (Barney, 1991). Firms that use strate-
gic human capital face challenging management
control issues, especially since the individual, not
the ?rm, owns the knowledge (Co?, 1997). As
Roos and Roos (1997, p. 413) state ‘‘the crux is
that it is individuals, not the company, who own
and control the chief source of competitive
advantage—the knowledge of organizational
members’’. Due to this lack of ownership, ?rms
face high uncertainty when predicting employee
behavior, tenure, and performance (Co?, 1997).
To mitigate the lack of ownership of the strategic
resource, ?rms often rely heavily on teams, net-
works, and other information sharing mechanisms
in order to extract the tacit knowledge embedded
in employees, thus making it more valuable to the
?rm (Grant, 1997; Quinn et al., 1996).
Attributes of strategic human capital
A strategic resource is one that is valued and
helps the ?rm sustain its long-term competitive
advantage through its lack of imitability. Human
capital is valuable when it is important to the ?rm
378 S.K. Widener / Accounting, Organizations and Society 29 (2004) 377–399
in terms of creating e?ciencies and enabling the
?rm to be more e?ective (Barney, 1991). Human
capital is di?cult for competitors to imitate when
the tasks and work performed are ambiguous,
when there is a high degree of ?rm-speci?c
knowledge, or when the knowledge and skills of
the human capital are spread through the ?rm
(Barney, 1991; Barney & Wright, 1998). These
attributes are illustrated in column 1, Fig. 1. In
order to gain more insights into the relation
between the use of a strategic resource and the
design of the MCS, I investigate all four attributes
of strategic human capital: (1) importance, (2)
behavioral uncertainty, (3) ?rm-speci?city, and (4)
spread of the resource through the ?rm.
The ?rst attribute is managers’ beliefs regarding
the importance of their strategic human capital to
the ?rm. This includes the degree to which man-
agers believe that strategic human capital helps
‘‘enable a ?rm to conceive of or implement strate-
gies that improve its e?ciency and e?ectiveness’’
(Barney, 1991, p. 106). Although all ?rms rely to
at least some extent on human capital, ?rms vary
in the importance they place on strategic human
capital (Edwards, 1997; Quinn et al., 1996; Snell &
Dean, 1992). Lepak and Snell (1999, p. 31)
emphasize this variability when they state, ‘‘not all
employees possess knowledge and skill that are of
equal strategic importance’’ (see also Barney,
1991, p. 102).
Behavioral uncertainty, the second attribute, is
‘‘when the link between a ?rm’s resources and its
sustained competitive advantage are poorly
understood’’ (Barney, 1991, p. 109). In other
words, ?rms are unable to identify the speci?c
linkages between employees’ e?orts and resulting
output. This lack of transparency hinders compe-
titors from replicating the process.
Fig. 1. Theoretical model and hypotheses (model A).
S.K. Widener / Accounting, Organizations and Society 29 (2004) 377–399 379
The third attribute is ?rm-speci?city. Competi-
tors generally do not value other ?rms’ stock of
?rm-speci?c human capital, thus there is low
transferability of employees that possess ?rm-spe-
ci?c knowledge. Grant (1991, p. 126) states that
‘‘if ?rms can acquire [on similar terms] the
resources required for imitating competitive
advantage of a successful rival, then that rival’s
competitive advantage will be short-lived’’. In
other words, ?rms cannot sustain competitive
advantage if the resource is easily transferable
between ?rms.
The spread of strategic human capital through-
out the ?rm, which is the fourth attribute, is the
degree to which there are small or large numbers
of strategic human capital within the ?rm. This
attribute also makes it di?cult for competitors to
imitate. Barney and Wright (1998) state ‘‘. . .the
synergistic value from a large number of indivi-
duals who work together is quite costly if not
impossible for competitors to imitate.’’
MCS
The purpose of the MCS is to provide informa-
tion that is useful in decision-making, planning,
control, and evaluation (Kaplan, 1983). The MCS
is a system composed of complementary compo-
nents (Milgrom & Roberts, 1995; Otley, 1994).
This study includes personnel controls, traditional
results controls, and nontraditional results con-
trols, illustrated in Fig. 1, column 2. I chose these
types of controls because they represent opposite
ends of the control spectrum (i.e., ex ante controls
and ex post controls). Thus, this study provides a
broader investigation into the design of the MCS.
Personnel controls typically function as ex ante
control mechanisms. Snell (1992, p. 297) de?nes a
personnel control as one that ‘‘regulates the ante-
cedent conditions of performance—the knowl-
edge, skills, abilities, values and motives of
employees’’. These controls are often centered on
human resource policies that help ensure that per-
sonnel will perform at a high level and in con-
gruence with ?rm goals (Merchant, 1982; Peck,
1994). In contrast to personnel controls, results
controls ‘‘regulate results’’ serving as an ex post
control mechanism (Snell, 1992, p. 297). There are
two broad types of results controls. Firms tradi-
tionally relied on ex post controls that provided
?nancial data that, in many respects, mirrored
?nancial data reported for external purposes.
However, recently ?rms have begun incorporating
more non?nancial and operational controls into
their MCS. Ittner and Larcker (1995, pp. 1–2)
state that they
Consider traditional managerial accounting
systems to be those that provide aggregated
?nancial information relatively infrequently,
operational control based on variances from
budgeted standards, and reward systems tied
primarily to ?nancial performance. In con-
trast, nontraditional systems provide more
timely physical measures of operational per-
formance, increased provision of problem-
solving information to the workers actually
performing the job, and reward systems that
focus more on non?nancial measures.
Thus, I explore both traditional and nontradi-
tional ex post controls.
In summary, this study investigates the relation
between four attributes of strategic human capital
(importance, behavioral uncertainty, ?rm-speci?-
city, and spread) and three types of controls
(personnel, nontraditional results, and traditional
results).
Theory development and hypotheses
In developing a theoretical foundation to study
the relationship between the four attributes of a
strategic resource and the three control compo-
nents of the MCS, I draw on contingency theory
1
(e.g, Chapman, 1997; Chenhall, 2003; Fisher,
1995; Otley, 1980) and transaction cost economics
(e.g., Spekle, 2001; Williamson, 1975, 1991). I chose
these theories because they explicitly target the
design of control mechanisms. Since each theory
1
This study uses the term ‘‘contingency theory’’ while
acknowledging that some studies present the perspective that
management control studies do not rely on contingency theory,
but rather employ a contingent framework developed from
underlying organizational theories (e.g., see Chenhall, 2002, p. 2).
380 S.K. Widener / Accounting, Organizations and Society 29 (2004) 377–399
provides only part of the story for understanding
how ?rms design their MCS it is necessary to
incorporate both theories in order to develop
hypotheses for all four attributes of strategic
human capital. Thus, this study provides a more
complete explanation and investigation. As indi-
cated in Fig. 2, each theory o?ers a di?erent
perspective for understanding the design of MCSs.
Contingency theory
Control systems are designed to assist managers
in making progress towards ?rm goals and
achieving desired outcomes (Chenhall, 2003).
Control systems do not exist in a vacuum; rather,
they are in?uenced by the context within which
they operate. The premise of contingency theory is
that control systems di?er across organizations
based on underlying organizational factors (Otley,
1980). Thus accounting researchers have invoked
contingency theory when studying the relationship
between organizational factors and the design of
the MCS. One of the organizational factors stud-
ied has been the organization’s strategy and
Chenhall (2003, p. 2) states that, ‘‘perhaps the
most important new stream of literature has been
that related to the role of strategy.
2
Implicitly
based on contingency theory, the ?rst set of
hypotheses investigates the in?uence that the
importance of strategic human capital has on the
design of the MCS.
Attribute 1: importance of strategic human capital
Firms are more willing to invest in costly ex ante
personnel controls if employees are strategically
important to the ?rm (Quinn et al., 1996; Snell &
Dean, 1992). Snell and Dean (1992) point out that
a ?rm is only willing to invest in personnel con-
trols if the marginal bene?t received from the
control exceeds the marginal cost of implementing
the control. Therefore, as management perceives
its strategic human capital to be increasing in its
importance, the ?rm will be willing to invest more
in personnel controls in order to, ex ante, ?nd and
develop employees whose skills, knowledge, and
goals are congruent with the needs of the organi-
zation (Becker, 1976; Merchant, 1982; Quinn et
al., 1996).
Although there is no direct empirical evidence
supporting the alignment of personnel controls
and the use of strategic human capital, Peck
(1994), in a survey of human resource and corpo-
rate executives, provides support for the positive
association between personnel controls and pro-
spector-like ?rms. She argues that this relationship
holds because prospector-like ?rms are innovative
Fig. 2. Theoretical foundation.
2
For a review of contingency literature see Chenhall (2002),
Chapman (1997), and Fisher (1995). For a review of the asso-
ciation between MCS and strategy see Lang?eld-Smith (1997).
S.K. Widener / Accounting, Organizations and Society 29 (2004) 377–399 381
?rms that desire a long-term relationship with
their employees and instill a work culture based on
trust, cooperation, and interdependent behavior
(Miles & Snow, 1978; Peck, 1994). In addition,
Snell and Dean (1992) ?nd that advanced manu-
facturing strategies (AMS) are associated with the
use of personnel controls. AMS ?rms employ
practices such as just-in-time inventory practices
and total quality management that rely heavily on
skilled knowledge workers working together to
perform interrelated task (Snell & Dean, 1992).
Since ?rms that use strategic human capital wish
to build long-term relationships, and rely on skil-
led knowledge workers in which cooperation,
trust, and knowledge sharing is critical (Co?,
1997; Grant, 1997), it becomes apparent that that
there are some similarities between prospector-like
?rms, those following AMS, and ?rms that use
strategic human capital. Thus the theoretical links,
while somewhat indirect, are suggestive of an
association between the importance of strategic
human capital and personnel controls. This
discussion leads to the following hypothesis:
Hypothesis 1a. Use of personnel controls is posi-
tively associated with the belief that strategic
human capital is important.
It is well-accepted that MCSs should support and
complement organizational strategies (e.g., see
Lang?eld-Smith, 1997). For many ?rms this has
meant a shift from traditional controls to nontradi-
tional controls that are more closely focused on the
?rm’s organizational strategy. The change in control
systems has been supported empirically in di?erent
strategic contexts. For example, Perera et al. (1997)
argue that for ?rms following a customer-focused
strategy ?nancial measures are unable to capture
the strategic intricacies and thus the ?rm will use
non?nancial measures that speci?cally focus on
the multiple dimensions that comprise quality.
Perera et al. (1997) ?nd a positive relationship
between the use of non?nancial measures and the
use of a customer-focused strategy. Although Per-
era et al. (1997) focuses on the narrower de?nition
of ?nancial vs. non?nancial controls, the ?nding
holds when extended to traditional vs. nontradi-
tional controls. Ittner and Larker (1995) investi-
gate the association between nontraditional
controls and ?rms’ reliance on quality initiatives.
They theorize that traditional controls do not
provide management with focused information
necessary to succeed with quality initiatives, thus,
believe that ?rms will change their MCS and rely
more on nontraditional controls. Ittner and
Larcker (1995) investigate quality practices in the
auto and computer industries and provide empiri-
cal support of the association between quality
initiatives and the use of nontraditional controls.
Similar to quality and customer-focused strate-
gies, Lev (2001) suggests that as ?rms rely more on
o?-balance sheet strategic resources such as
human capital, it is likely that they will rely more
on nontraditional controls that provide informa-
tion focused on the strategic resource. An example
that illustrates the importance of shifting from
traditional to nontraditional controls is the
investment ?rms make in their employees. Firms
that rely on human capital as a strategic resource
will invest in that resource through various train-
ing and development costs (Snell & Dean, 1992).
But the investment is recorded as an expense (e.g.,
employee training) that depresses pro?tability, at
least in the short-run. Thus relying on traditional
?nancial controls will provide managers with
imperfect information for decision-making pur-
poses, and as McNair, Lynch, and Cross (1990)
suggest, may actually be counterproductive.
Instead, ?rms will rely more on nontraditional
controls that provide information and link
rewards to operational and team metrics focused
on important dimensions of strategic human capi-
tal including training, development, knowledge
sharing, turnover, and productivity (Balkcom,
Ittner & Larcker, 1997; Grant, 1997; Lank, 1997;
Lev, 2001; Spicer & Ballew, 1983; Wallman, 1995).
Balkcom et al. (1997, p. 5) state
Measures such as employee satisfaction,
voluntary turnover, employee skill develop-
ment, and employee safety have become
important indicators of the company’s suc-
cess in developing a competitive advantage.
Companies are increasingly integrating some
or all of these non?nancial measures in plan-
ning, control, and compensation decisions.
382 S.K. Widener / Accounting, Organizations and Society 29 (2004) 377–399
This discussion is summarized in the following
hypotheses:
Hypothesis 1b. Use of nontraditional results con-
trols is positively associated with the belief that
strategic human capital is important.
Hypothesis 1c. Use of traditional results controls is
negatively associated with the belief that strategic
human capital is important.
Transaction cost economics theory
Transaction cost economics (TCE) provides a
basis for developing hypotheses about how beha-
vioral uncertainty, ?rm-speci?city, and the spread
of the resource a?ect the design of the MCS. TCE
recognizes these attributes as drivers of transac-
tion costs (Williamson, 1975). A premise of TCE is
that ?rms will organize and develop governance
structures to minimize the sum of production and
transaction costs
3
(Williamson, 1970, 1975, 1991).
Management control systems are one form of
governance structure (Spicer & Ballew, 1983).
Thus MCSs will be designed such that transaction
costs arising from behavioral uncertainty, ?rm-
speci?city, and spread are minimized.
TCE is based on behavioral assumptions of
bounded rationality and opportunism (William-
son, 1975). Bounded rationality implies that
behavior is limited by imperfect cognitive pro-
cesses. Thus, contracts are incomplete. Opportu-
nism implies that individuals behave self-
interestedly, which is enabled by incomplete con-
tracts. Thus, a distinct need arises for measure-
ment and control, especially in the ex post settling
of contracts (Seal, 1993). In other words, because
future states of nature are uncertain and indivi-
duals are boundedly rational, a complete contract
cannot be speci?ed ex ante. Results must be mea-
sured and ex post the parties to the contract must
settle with one another.
Drawing on TCE, the following hypotheses
investigate the in?uence that behavioral uncer-
tainty, ?rm-speci?city, and spread have on the
design of the MCS.
Attribute 2: behavioral uncertainty
Behavioral uncertainty refers to a lack of spe-
ci?ability or programmability regarding the
actions undertaken by employees and how those
actions in?uence subsequent outcomes. Spekle
(2001) de?nes it as ‘‘the degree of speci?ability of
intended performance’’. The result is information
asymmetry since the ?rm cannot directly attribute
results and performance to an individual’s e?ort
(e.g., the individual possesses superior information
regarding his/her e?ort). TCE assumes that indi-
viduals act with self-interest
4
; thus an environment
characterized by behavioral uncertainty may
manifest itself in either adverse selection or moral
hazard (Co?, 1997). Adverse selection arises when
employees possessing information unknown to the
?rm misrepresent themselves and lead the ?rm to
make a costly decision (Baiman, 1982). This can
be seen in the labor market. Firms may o?er low
wages to hedge against the inability to observe
employees’ actual knowledge and skills. In turn,
the labor market may become disproportionately
full of low-quality job candidates. Moral hazard
arises when contracts are based on surrogate
information and the employee is able to take
action that is self-interested (Baiman, 1982). This
often results in shirking or other unwanted beha-
vior in the workplace.
Ex ante personnel controls help mitigate both
adverse selection and moral hazard. To mitigate
adverse selection, ?rms can either avoid the labor
market
5
or implement coping strategies such as
obtaining better information and strengthening
their ability to interpret labor signals (Co?, 1997).
The latter can be facilitated through an increased
emphasis on ex ante personnel controls that focus
on inputs such as recruiting, selection of new
3
Colbert and Spicer (1995, p. 425) state that ‘‘transaction
costs are the costs of engaging in transactions via a particular
arrangement or governance structure’’. Transaction costs also
occur within a ?rm. Examples of such costs are control losses
and planning di?culties (Spicer & Ballew, 1983).
4
Williamson (1985, p. 47) states that opportunism is ‘‘self-
interest seeking with guile’’. In TCE vernacular, it is a very
broad term that subsumes moral hazard and adverse selection.
5
This is an extreme control. This study assumes that this
restrictive of a control cannot be met by the ?rm.
S.K. Widener / Accounting, Organizations and Society 29 (2004) 377–399 383
employees, and developing and using multiple infor-
mation sources in the ex ante decision-making pro-
cess (Snell &Dean, 1992). To mitigate moral hazard,
?rms emphasize personnel controls since this helps
ensure ex ante that future employees will have con-
gruent goals with the organization (Williamson,
1985), and thus are less likely to take advantage of an
environment that might allow for employees to act
with self-interest (Merchant, 1998). In other words,
?rms that rely heavily on human capital will seek to
?nd individuals with characteristics (e.g., values,
goals, skills, etc.) that are congruent to the ?rm’s
culture thus minimizing opportunistic behavior (e.g.,
shirking) and related transaction costs. Abernethy
and Brownell (1997) provide related empirical
evidence and ?nd that ?rms rely more on person-
nel controls in an environment where employees
perform a variety of tasks that do not have well-
programmed routines and structures and where
there is a lack of understanding of how employee
actions in?uence outcomes (see also Chenhall,
2003). Therefore, in environments characterized
by behavioral uncertainty it is likely that ?rms will
rely on ex ante controls (e.g., personnel controls)
since they reduce transaction costs associated with
opportunism (Spicer & Ballew, 1983). This dis-
cussion supports the following hypothesis:
Hypothesis 2a. Use of personnel controls is posi-
tively associated with behavioral uncertainty of
the strategic human capital.
Since there is no speci?ed relationship between
employee inputs or actions and the resulting
?nancial outputs, organizations cannot rely on
traditional results controls. Spekle (2001, p. 432)
states ‘‘the budget [e.g., a results control] itself is a
rather insigni?cant control device in these circum-
stances’’. Instead, Spekle (2001) suggests that the
?rm may use an exploratory control environment
characterized by low reliance on traditional con-
trols such as budgeting practices and monitoring
of ?nancial measures. To illustrate the ine?ective-
ness of traditional controls consider shirking. In
an environment characterized by a low under-
standing or speci?ability of the input/output pro-
cess (e.g., high behavioral uncertainty), a MCS
that regulates results using traditional measures
(e.g., pro?tability) may foster an atmosphere of
shirking since the ?rm will not be able to hold any
one person accountable for performance. Hirst
(1983) empirically shows that when behavioral
uncertainty is high traditional controls are incom-
plete and lead to increased tension among
employees. He notes that this may result in
opportunistic behavior. Abernethy and Brownell
(1997) also provide empirical evidence that tradi-
tional controls are not e?ective in a research and
development setting characterized by lack of
knowledge of the input/output process. Chenhall
(2003) summarizes this literature stream in a pro-
position that associates high uncertainty with less
reliance on traditional accounting measures. This
discussion supports the following hypothesis:
Hypothesis 2b. Use of traditional results controls
is negatively associated with behavioral uncer-
tainty of the strategic human capital.
As just discussed, due to a lack of program-
mability and speci?ability, ?rms cannot use tradi-
tional results controls in environments
characterized by behavioral uncertainty; however,
it is likely that they will rely on nontraditional
results controls. Spekle (2001) suggests that in a
climate characterized by behavioral uncertainty,
?rms will seek to establish an environment of
commitment and congruency to general organiza-
tional goals. In addition, ?rms will seek informa-
tion that they believe may shed insights into the
true nature of the activities being performed by
individuals (Spekle, 2001, p. 431). Thus ?rms will
seek focused information speci?cally related to its
strategic choices to assist managers in assessing
performance outcomes and to use during the
negotiations or the ex post performance eval-
uation process (Seal, 1993). It is likely that non-
traditional results controls, speci?cally focused on
the ?rm’s strategic goals and objectives, may pro-
vide this information since it is often argued that
nontraditional controls provide accurate and
timely information that helps the ?rm assess
employees’ actual performance (Baiman, 1990;
Seal, 1993). For example, Caterpillar Inc. rede-
signed their control system when they realized the
control system was not providing an accurate
384 S.K. Widener / Accounting, Organizations and Society 29 (2004) 377–399
picture of performance (Hendricks, Defreitas, &
Walker, 1996). Caterpillar credits the new control
system, which employs nontraditional controls
focused on its strategy, with cost savings associated
with improved decision making.
In addition, nontraditional controls focused
speci?cally on dimensions of strategic human
capital (e.g., employee satisfaction, knowledge
sharing, teamwork, etc.) may reduce future trans-
action costs. For example, by collecting data on
employee satisfaction, ?rms may better commu-
nicate their objectives and create alignment within
the ?rm focused on organizational goals and
objectives. This may help in establishing a corpo-
rate culture conducive to high employee e?ort
thus minimizing costs associated with moral
hazard. Furthermore, turnover may be minimized
which reduces di?culties associated with adverse
selection (Seal, 1993; Co?, 1997). This discussion
suggests that in an environment characterized by
behavioral uncertainty, ?rms will rely on ex post
controls that are nontraditional in nature.
Hypothesis 2c. Use of nontraditional results con-
trols is positively associated with behavioral
uncertainty of the strategic human capital.
Attribute 3: ?rm-speci?city
Firm-speci?city is the skills and knowledge pos-
sessed by employees that are speci?c to a unique
?rm (Co?, 1997). Often, ?rm-speci?city is referred
to as the extent to which specialized or non-
redeployable investments are needed to support an
exchange (Williamson, 1975). Similar to behavioral
uncertainty, ?rm-speci?city may facilitate opportu-
nistic behavior since ?rm-speci?c human capital
possesses idiosyncratic skills and knowledge that
others are often unable to observe (Co?, 1997).
Therefore, Co? (1997, p. 378) concludes that ?rm-
speci?city results in the same agency problems as
does behavioral uncertainty e.g., adverse selection
and moral hazard. Furthermore, the extent to
which opportunistic behavior is tolerated increases
since the market does not provide protection to
parties that invest in relational speci?c assets
(Baiman, 1982). Thus, there is an increased need
for a well-designed MCS that is heavily focused on
ex ante personnel controls and ex post nontradi-
tional results control. The former control ensures
that employees with similar goals, ethics, and mor-
als are brought into the organization. The latter
control ensures that there is a monitoring system in
place that focuses on information targeted speci?-
cally to the strategic resource. Traditional results
controls are not relied on as they are not e?ective in
this environment. These latter statements are
developed more fully in the preceding section and
thus not repeated here (see the previous section).
This discussion suggests the following hypotheses:
Hypothesis 3a. Use of personnel controls is
positively associated with ?rm-speci?city of the
strategic human capital.
Hypothesis 3b. Use of nontraditional results con-
trols is positively associated with ?rm-speci?city of
the strategic human capital.
Hypothesis 3c. Use of traditional results controls is
negatively associated with ?rm-speci?city of the
strategic human capital.
Attribute 4: spread of strategic human capital
A primary tenet of TCE is cost minimization
(Williamson, 1991). TCE argues that ?rms need to
align governance structures with the drivers of
transaction costs in order to minimize costs and
thus economize. One such driver of transaction
costs is ‘‘frequency’’, which this paper considers to
be the spread of the strategic human capital
through the ?rm.
As discussed earlier, a traditional MCS focuses
on aggregated ?nancial accounting information
and has strong links to the budgetary system (Itt-
ner & Larcker, 1995). A more sophisticated man-
agement control system is one that is tailored
solely to the needs of the organization and aligned
closely with its strategy. But changing from a tra-
ditional MCS to a more sophisticated MCS can be
costly since ?rms incur costs when designing and
implementing a governance structure, e.g., the
MCS (Williamson, 1970, 1975, 1991). Therefore,
the ?rm must consider whether a specialized MCS
can be used to capacity, or whether there will be
much unused capacity that may make a sophisti-
cated MCS unduly costly (Williamson, 1985).
S.K. Widener / Accounting, Organizations and Society 29 (2004) 377–399 385
The size of the spread of strategic human capital
throughout the ?rm helps determine whether the
bene?ts from the MCS will outweigh the design
and implementation costs of the MCS (William-
son, 1975). If there are large numbers of strategic
human capital it is likely that the ?rm will invest in
a more complex governance structure. If strategic
human capital is concentrated (e.g., at the
extreme, the human capital is localized within one
top employee), then the ?rm is not likely to invest
in a complex governance structure since they may
not recoup the costs of designing and implement-
ing a more complex MCS. Therefore, as the num-
bers of strategic human capital increase, the ?rm
will be more willing to invest in personnel controls
designed to ?nd skilled, knowledgeable employees
(e.g., Becker, 1976; Snell & Dean, 1992; William-
son, 1991). In addition, ?rms will invest in non-
traditional controls more closely aligned with its
strategy (e.g., Lang?eld-Smith, 1997). Further-
more, it is likely that ?rms will decrease their
reliance on traditional results controls since
there are limits on cognitive capacity (William-
son, 1975). Simply increasing the information
provided to managers through the implemen-
tation of additional controls can lead to costs
associated with information overload. Therefore,
to remain in equilibrium and satisfy a cost
minimizing objective (Williamson, 1985) ?rms
will likely trade-o? the cost of traditional con-
trols (that are not speci?cally focused on its
strategic choice) for nontraditional and person-
nel controls (that are speci?cally focused on its
strategic choice). This discussion supports the
following hypotheses:
Hypothesis 4a. Use of personnel controls is posi-
tively associated with the spread of strategic
human capital throughout a ?rm.
Hypothesis 4b. Use of nontraditional results con-
trols is positively associated with the spread of
strategic human capital throughout the ?rm.
Hypothesis 4c. Use of traditional results controls is
negatively associated with the spread of strategic
human capital throughout the ?rm.
Methods
Research design and sample
Both Compustat and survey data are used as
sources of data. Firms that report sales for
between one and ?ve four-digit SIC codes in the
same division code are included. This ensures that
?rms conduct business in only one major SIC code
division.
6
To enable analysis of non-response bias
and to validate the variable measures, selected
?rms are required to have sales from 1993 to 1996
and depreciation expense for 1996 on the Compu-
stat ?le. Firms are also required to have at least
250 employees. Eliminating small ?rms is neces-
sary since they are less likely to have a formal
control system.
After deleting ?rms that are either used in the
pretest or foreign-owned, the remaining popula-
tion is 1, 662 ?rms, of which 800 were randomly
surveyed. The largest concentration of ?rms
(43%) is classi?ed as manufacturing (SIC codes
2000-3999). Other concentrated segments include
?nancial services ?rms (17%), other service ?rms
(13%), and transportation, communication and
utilities (11%).
Research method
The design and administration of the survey
accords with the ‘‘total design approach’’ advo-
cated by Dillman (1978).
7
As a pretest, a survey
and cover letter were sent to 250 randomly selec-
ted ?rms. In addition to asking for responses, both
the survey and cover letter asked respondents to
comment on clarity, understanding, ambiguity,
etc. Four in-depth visits to ?rms were also con-
ducted in order to gain a better understanding of
the design of the MCS and use of human capital
as a strategic resource, and to help in designing the
6
This screening technique attempts to remove highly diver-
si?ed companies from the sample. Highly diversi?ed companies
may follow multiple strategies thus adding considerable noise
to the measures.
7
The Total Design Method involves an ‘‘identi?cation of
each aspect of the survey process that may a?ect response
quality or quantity’’ (Dillman, 1978, p. 2). It is a very detailed,
step-by-step approach to conducting a quality survey.
386 S.K. Widener / Accounting, Organizations and Society 29 (2004) 377–399
?nal instrument. Based on the results of the pret-
est (30 respondents) several changes were made to
the ?nal survey. Upon revision, the survey along
with a personalized cover letter and a stamped
return envelope was sent to the chief ?nancial
o?cer of 800 ?rms. As an incentive to respond,
the participants were promised a summary of the
results. Three follow-up mailings, along with a
postcard reminder, were sent to the sample selection
of ?rms.
The mailing process resulted in 118 responses
(15% response rate), which is below the average of
20% that Young (1996) reports for comparable
surveys to CFOs. One reason for the lower
response rate is the asking of multiple questions
for each construct. This study chose to increase
construct validity at the potential risk of incurring
a lower response rate. As a result, non-response
bias is rigorously investigated. The pattern of SIC
classi?cations for respondents mirrors both the
sample and the Compustat population. The two
largest groups of respondent ?rms are either man-
ufacturers (48 ?rms) or ?nancial services ?rms (19
?rms). Non-response bias is investigated by com-
paring respondents to non-respondents. The
respondent ?rms are further sub-divided into
early, middle and late respondents based on their
return date. There are no statistically signi?cant
di?erences between any classi?cations of respon-
dent ?rms for 1997 sales, total assets, number of
employees and depreciation expense. Early
respondents are also compared to late respondents
for all constructs of interest. The results are
reported in Table 1. Early respondents are sig-
ni?cantly higher (P