Professional Documents
Culture Documents
com
Department of Accounting and Finance, London School of Economics and Political Science, Houghton Street,
London WC2A 2AE, United Kingdom
Abstract
This study examines how comprehensive performance measurement systems (PMS) aect managerial performance.
It is proposed that the eect of comprehensive PMS on managerial performance is indirect through the mediating variables of role clarity and psychological empowerment. Data collected from a survey of 83 strategic business unit managers are used to test the model. Results from a structural model tested using Partial Least Squares regression indicate
that comprehensive PMS is indirectly related to managerial performance through the intervening variables of role clarity and psychological empowerment. This result highlights the role of cognitive and motivational mechanisms in
explaining the eect of management accounting systems on managerial performance. In particular, the results indicate
that comprehensive PMS inuences managers cognition and motivation, which, in turn, inuence managerial
performance.
2007 Elsevier Ltd. All rights reserved.
Introduction
In recent years organizations have sought to
develop more comprehensive performance measurement systems (PMS) to provide managers
and employees with information to assist in managing their rms operations (Fullerton & McWatters, 2002; Ittner, Larcker, & Randall, 2003; Lillis,
2002; Malina & Selto, 2001; Ullrich & Tuttle,
2004). Prior research indicates that more comprehensive PMS include a more diverse set of performance measures, and performance measures that
are linked to the strategy of the rm and provide
information about parts of the value chain (Chenhall, 2005; Malina & Selto, 2001; Nanni, Dixon, &
0361-3682/$ - see front matter 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aos.2007.02.004
142
The remainder of the paper contains four sections: the next section develops the theoretical
model, including presentation of the hypotheses.
The research method, including sample selection
and variable measurement, is then presented. This
is followed by presentation of the results. The nal
section discusses the results and concludes the
paper.
Comprehensive
PMS
H1
H3
H5
Role clarity
H2
Psychological
empowerment
H4
143
Managerial
Performance
144
such, a more comprehensive PMS is one that provides more comprehensive performance information to managers, i.e., measures that fully
describe the SBUs operations and link to strategy
and across the value chain. In contrast, a less comprehensive PMS is one that provides less comprehensive performance information to managers,
i.e., measures that only partially describe the
SBUs operations and contain few (if any) links
to strategy and across the value chain.
The way in which comprehensive PMS provide
enhanced performance information supplies the
basis for linking comprehensive PMS with SBU
managers role clarity and psychological empowerment. Individuals at higher levels in the organisation, such as SBU managers, obtain feedback
about the results of operations and work-related
performance from PMS (Collins, 1982; Luckett
& Eggleton, 1991). A more comprehensive PMS
provides richer and more complete feedback about
operations and results to SBU managers (Chenhall, 2005; Kaplan & Norton, 2001; Malina &
Selto, 2001), which is expected to have positive
eects on managers role clarity and psychological
empowerment.
Comprehensive PMS and role clarity
Role clarity refers to individuals beliefs about
the expectations and behaviours associated with
their work role (Kahn, Wolfe, Quinn, Snoek, &
Rosenthal, 1964).1 In this study I examine whether
comprehensive PMS is related to two aspects of
role clarity; goal clarity (the extent to which the
outcome goals and objectives of the job are clearly
stated and well dened) and process clarity (the
extent to which the individual is certain about
how to perform his or her job) (Sawyer, 1992). It
is expected that more comprehensive performance
information will help to clarify SBU managers
1
Kahn et al. (1964) use the term role ambiguity, which refers
to uncertainty regarding parts of an individuals role. In this
study the term role clarity is used. However, this is conceptually
no dierent from role ambiguity (Sawyer, 1992). Role clarity is
expressed as the extent of certainty, rather than ambiguity, of
role expectations.
145
146
147
148
exibility to adapt to changing situations and create improved task strategies (Deci & Ryan, 1987;
Thomas & Velthouse, 1990). Work performance
is enhanced when managers believe they have
autonomy over how their work is to be accomplished (Miller & Monge, 1986). In relation to
impact, individuals who believe they can inuence
outcomes at work are more likely to actually have
an impact, and hence be more eective. Impact
results in more eort and greater persistence in
the face of obstacles (Abramson, Seligman, &
Teasdale, 1978; Ashforth, 1989; Spreitzer et al.,
1997; Thomas & Velthouse, 1990). Spreitzer
et al. (1997) and Liden et al. (2000) found that
impact was positively associated with work performance. These arguments and evidence lead to H4:
managers to determine and take actions to complete tasks, and thus should increase self-determination. A lack of role clarity is likely to make
individuals believe they are helpless and thus
reduce the impact they have in their work area
(Spreitzer et al., 1997). In contrast, individuals
who understand their work roles are more likely
to take actions and decisions that inuence results
in their work area (Sawyer, 1992). Prior research
shows that higher levels of role ambiguity are
related to lower levels of psychological empowerment (Smith & Langeld-Smith, 2003; Spreitzer,
1996). This analysis indicates that role clarity will
increase each dimension of psychological empowerment, which leads to H5:
H5: There is a positive relation between role
clarity and the four dimensions of psychological
empowerment.
Research method
Role clarity and psychological empowerment
Sample selection and data collection
Finally, drawing on prior results, I hypothesize
a positive relation between role clarity and psychological empowerment. Unless SBU managers have
a clear sense of their responsibilities and how to
achieve them, it will be dicult for them to know
if they have the necessary skills and abilities to perform their tasks adequately (i.e., feel empowered).
As such, role clarity is expected to increase each
dimension of psychological empowerment; meaning, competence, self-determination and impact.
Spreitzer (1996) argues that it is only when individuals understand their roles that those roles can
take on personal meaning. Clear lines of responsibility and clear task requirements are related to
competence (Conger & Kanungo, 1988; Gist &
Mitchell, 1992; Kahn et al., 1964). SBU managers
with clear work goals, and an understanding of
how to achieve those goals, are likely to believe
they can perform their job with skill and thus feel
more competent. Managers who are uncertain of
their role expectations are likely to hesitate and
not take the initiative due to uncertainty, and thus
experience lower levels of self-determination (Spreitzer et al., 1997). High levels of role clarity enable
I collected data using a questionnaire administered to SBU managers within Australian manufacturing organizations. I obtained a list of 1000
SBU managers of Australian manufacturing rms
from a commercial mailing list provider. Due to
cost constraints, 400 managers were selected to
form the sampling frame for the study. I used a
four-step implementation strategy following the
recommendations of Dillman (2000); telephone
calls to check data accuracy3, a questionnaire
package with cover letter, questionnaire and
reply-paid envelope, a reminder postcard (sent
two weeks after questionnaire package), and a follow-up phone call (made two weeks after the
reminder postcard). To encourage completion of
the questionnaire, participants were promised a
summary of the results and informed that their
3
149
4
16 cases contained missing data: 14 cases with one item
missing, one case with two items missing, and one case with
four items missing. Littles MCAR test revealed that the
missing data were missing completely at random (MCAR)
(v2 = 4.424, degrees of freedom = 516, p > 0.10). As the missing
data is MCAR, any imputation method can be used (Hair,
Anderson, Tatham, & Black, 1998). As such, the data were
replaced using the expectationmaximisation (EM) method in
SPSS. The EM approach is an iterative two-stage process where
the E-stage makes the best estimates of the missing data and the
M-stage makes parameter estimates assuming the missing data
are replaced. This occurs in an iterative process until the
changes in the estimated parameters are negligible and the
missing values are replaced (Hair et al., 1998; Little & Rubin,
1987). This process resulted in a complete data set of 83
responses.
83
Table 1
Descriptive statistics for demographic variables
Variable
Job tenure
(years)
Company tenure
(years)
Age (years)
SBU size (no.
of employees)
Minimum
Maximum
Mean
St Dev
25
5.14
5.95
37
10.64
8.37
26
10
64
4100
46.84
336.13
8.38
497.03
n = 83.
Table 2
Manufacturing industry classication
ANZSICa manufacturing industry
classication
Frequency
8
3
9.64
3.61
6
3
7.23
3.61
12
14.46
4
11
25
11
4.82
13.25
30.12
13.25
100
150
Variable measurement
The questionnaire obtained information on
comprehensive PMS, psychological empowerment,
role clarity and managerial performance. Established scales were used for each variable, except
comprehensive PMS. The development of the
questionnaire involved a review by three senior
management accounting academics with experience in survey design. I also pilot tested the questionnaire with four SBU managers (not part of
the sample), who completed the questionnaire
and participated in a brief interview. The review
process and the pilot test resulted in minor changes
to the wording of some items and to the layout of
the questionnaire.
Ittner, Larcker, and Randall (2003) argue that
improved methods are needed for determining
what rms mean by contemporary PMS, such as
the balanced scorecard. Prior research relating to
comprehensive PMS has used scales that examine
the extent to which a PMS contains a series of
specic performance measures (for example,
Hoque & James, 2000). A limitation of this type
of instrument is that it assumes that the performance measures contained in the instrument are
representative of the specic types of performance
measures used by the rms in the sample. Firms
may use similar nancial performance measures;
however, non-nancial and/or strategic measures
are likely to be unique to each rm (Lipe & Salterio, 2000). In addition, this type of scale may not
capture the strategic linkages of more comprehensive PMS (Hoque & James, 2000). As such, I
developed a new scale to capture the comprehensive PMS construct. The scale consists of nine
items. Five items relate to the extent to which the
PMS provides a variety of performance information about the important parts of the SBUs operations. The remaining four items were drawn from
Chenhall (2005), and relate to the extent of integration of measures with strategy and across the
value chain. The explanatory statement indicated
that we were interested in the extent to which the
PMS provides information about the operations
of the respondents business unit. This was done
to help ensure that when SBU managers were
responding to the statements, they were focused
Factor loading
0.915
0.782
0.843
0.817
0.896
0.864
0.852
0.739
0.836
6.350
70.559%
151
152
Results
Measurement model
Statistics from the PLS measurement model are
used to examine the psychometric properties of the
variables. First I examine the factor loadings for
each variable. All items load on their respective
constructs; however, two items from the managerial performance scale have factor loadings below
0.5 (Hulland, 1999) (item MP7 = 0.461 and item
MP8 = 0.246). Low item loadings add very little
to the explanatory power of the model while
potentially biasing the estimates of the parameters
linking the constructs (Chin, 1998; Hulland, 1999).
Further tests show that the reason the two items
have low factor loadings is because they do not
form part of a unidimensional managerial performance scale.8 As such, items MP7 and MP8 are
removed from the scale and not used in further
analysis. The factor loadings from the nal PLS
measurement model are reported in Table 4.
I assess the reliability of each variable using Fornell and Larckers (1981) measure of composite
reliability and Cronbachs (1951) alpha. As shown
in Table 5, the composite reliability and alpha
7
scores for each variable are above 0.80, which demonstrates acceptable reliability (Nunally, 1978).
Convergent validity of the variables is assessed
by examining the average variance extracted
(AVE) statistics. Table 5 shows that the AVE for
each variable is 0.50 and above, which demonstrates adequate convergent validity (Chin, 1998;
Hair et al., 1998).
The AVE statistic is also used to assess discriminant validity by comparing the square root of the
AVE statistics to the correlations among the latent
variables (Chin, 1998). This tests whether a construct shares more variance with its measures than
it shares with other constructs (Fornell & Larcker,
1981). Table 5 shows that the square roots of the
AVEs (diagonal) are all greater than the respective
correlations between constructs. In addition, Table
4 shows that each item loads higher on the construct it intends to measure than on any other construct (Barclay et al., 1995; Chin, 1998). The results
of these two tests demonstrate adequate discriminant validity. Overall, the results from the PLS
measurement model indicate that each construct
exhibits satisfactory reliability and validity.
Tests of hypotheses
I estimate a structural model in PLS to test the
hypotheses. In addition to the hypothesized paths,
I also include job tenure in the structural model to
control for the endogeneity concern that more tenured employees have access to more information
and also feel more psychologically empowered
(Chenhall & Moers, in press). The objective of
PLS is to maximise variance explained rather than
t, therefore prediction-orientated measures, such
as R2, are used to evaluate PLS models (Chin,
1998). The R2 for each endogenous variable is
shown in Table 6. PLS produces standardised bs
for each path coecient, which are interpreted in
the same way as in OLS regression. As PLS makes
no distributional assumptions, bootstrapping (500
samples with replacement) is used to evaluate the
statistical signicance of each path coecient
(Chin, 1998).9
9
Statistical signicance is determined using the reported
original PLS estimates and bootstrapped standard errors.
153
Table 4
Factor loadings from nal PLS measurement model
Item
CPMS
GC
PC
MEAN
COMP
IMP
SD
MP
CPMS1
PMS2
CPMS3
CPMS4
CPMS5
CPMS6
CPMS7
CPMS8
CPMS9
0.920
0.786
0.837
0.810
0.896
0.860
0.859
0.733
0.841
0.431
0.316
0.280
0.362
0.365
0.396
0.380
0.381
0.361
0.196
0.188
0.135
0.098
0.190
0.174
0.134
0.096
0.142
0.497
0.368
0.404
0.360
0.352
0.430
0.435
0.340
0.398
0.407
0.318
0.300
0.158
0.308
0.261
0.315
0.159
0.365
0.475
0.419
0.308
0.382
0.426
0.379
0.491
0.307
0.457
0.269
0.226
0.159
0.201
0.249
0.281
0.268
0.175
0.268
0.319
0.236
0.305
0.201
0.238
0.316
0.251
0.251
0.254
GC1
GC2
GC3
GC4
GC5
0.349
0.363
0.265
0.364
0.399
0.787
0.795
0.786
0.801
0.875
0.532
0.615
0.530
0.507
0.566
0.450
0.454
0.426
0.535
0.535
0.431
0.498
0.335
0.457
0.488
0.564
0.538
0.378
0.517
0.483
0.478
0.473
0.357
0.244
0.326
0.481
0.559
0.430
0.580
0.524
PC1
PC2
PC3
PC4
PC5
0.203
0.122
0.115
0.100
0.185
0.586
0.535
0.590
0.453
0.596
0.795
0.813
0.847
0.811
0.817
0.295
0.247
0.363
0.268
0.450
0.446
0.488
0.426
0.357
0.533
0.464
0.408
0.394
0.339
0.467
0.448
0.501
0.427
0.455
0.481
0.448
0.336
0.414
0.364
0.462
MEAN1
MEAN2
MEAN3
0.457
0.486
0.405
0.559
0.590
0.534
0.366
0.401
0.375
0.949
0.961
0.918
0.593
0.632
0.625
0.556
0.648
0.525
0.361
0.407
0.305
0.583
0.597
0.554
COMP1
COMP2
COMP3
0.288
0.233
0.370
0.428
0.440
0.548
0.432
0.483
0.525
0.514
0.522
0.642
0.868
0.777
0.932
0.453
0.566
0.560
0.336
0.403
0.473
0.471
0.330
0.559
IMP1
IMP2
IMP3
0.502
0.346
0.413
0.538
0.545
0.535
0.353
0.540
0.461
0.619
0.490
0.481
0.618
0.480
0.478
0.850
0.881
0.886
0.495
0.644
0.683
0.519
0.332
0.366
SD1
SD2
SD3
0.264
0.319
0.183
0.400
0.424
0.430
0.455
0.489
0.572
0.334
0.395
0.302
0.318
0.474
0.461
0.634
0.657
0.580
0.891
0.938
0.871
0.248
0.382
0.418
MP1
MP2
MP3
MP4
MP5
MP6
MP9
0.262
0.220
0.244
0.187
0.223
0.265
0.170
0.412
0.370
0.385
0.402
0.509
0.532
0.511
0.302
0.308
0.342
0.213
0.394
0.413
0.449
0.529
0.312
0.258
0.333
0.540
0.456
0.500
0.431
0.294
0.246
0.280
0.411
0.460
0.450
0.337
0.276
0.237
0.289
0.357
0.372
0.415
0.228
0.236
0.291
0.138
0.350
0.298
0.372
0.664
0.568
0.701
0.689
0.768
0.743
0.788
CPMS, comprehensive performance measurement system; GC, goal clarity; PC, process clarity; MEAN, meaning; COMP, competence; IMP, impact; SD, self-determination; MP, managerial performance.
n = 83.
CPMS, comprehensive performance measurement system; GC, goal clarity; PC, process clarity; MEAN, meaning; COMP, competence; IMP, impact; SD, selfdetermination; MP, managerial performance.
Diagonal elements are the square roots of the AVE statistics. O-diagonal elements are the correlations between the latent variables calculated in PLS. All correlations
above 0.20 are statistically signicant (p < 0.01, two-tailed).
n = 83.
0.902
0.397
0.872
0.691
0.472
0.861
0.608
0.472
0.536
0.943
0.654
0.613
0.381
0.613
0.817
0.404
0.557
0.513
0.567
0.500
0.809
0.681
0.595
0.552
0.619
0.466
0.641
0.840
0.434
0.182
0.478
0.351
0.487
0.281
0.315
0.705
0.655
0.667
0.889
0.742
0.761
0.811
0.500
0.946
0.868
0.875
0.935
0.804
0.842
0.880
0.824
4.657
5.963
5.191
5.916
5.891
6.121
5.971
5.405
CPMS
GC
PC
MEAN
COMP
IMP
SD
MP
1.289
0.829
0.871
1.019
0.729
0.778
0.938
0.610
0.955
0.905
0.909
0.960
0.895
0.905
0.929
0.874
SD
IMP
COMP
MEAN
PC
GC
CPMS
Correlations
AVE
Composite reliability
Cronbach alpha
Standard deviation
Mean
Variable
Table 5
Descriptive statistics, reliability and average variance extracted (AVE) statistics, and correlations from PLS model
0.707
MP
154
10
0.513
0.480
0.375
0.027 (0.379)
0.129
(0.591)
n = 83. Each cell reports the path coecient (t-value). Blank cells indicate that the path was not hypothesized within the model.
*
p < 0.10 (one-tailed).
**
p < 0.05.
0.113
(0.826)
0.351
(2.350)***
Managerial performance
Impact
Self-determination
Discussion
0.107
(0.833)
0.063 (0.835)
0.143 (1.647)
0.417
Competence
Meaning
Process clarity
155
**
0.151 (1.857)**
0.442
0.143 (2.100)
0.052
(0.445)
0.374
(3.369)***
0.495
(4.171)***
0.233
(2.328)***
0.047
(0.311)
0.448
(3.153)***
0.224
(1.594)*
0.064
(0.492)
0.334
(2.607)***
0.380
(2.328)***
0.034
**
0.020 (0.237)
0.016
0.200
0.121 (1.365)
0.050 (0.600)
0.440
(4.668)***
0.184
(1.497)*
0.256
(2.845)***
0.167
(1.474)*
0.147
1.305)*
0.292
(2.717)***
0.030
(0.298)
Comprehensive PMS
Goal clarity
Goal clarity
Comprehensive PMS
Independent variables
Dependent variables
Table 6
PLS structural model: path coecients, t-statistics and R2
Process clarity
Meaning
Competence
Self-determination
Job tenure
Impact
R2
156
Comprehensive
PMS
0.147*
Selfdetermination
0.184*
0.495***
0.167*
Process
Clarity
0.374***
Competence
0.440***
0.233***
0.292***
0.224*
Goal
Clarity
0.334***
Impact
0.448***
0.256***
0.380***
Meaning
0.351***
Managerial
Performance
Fig. 2. PLS structural model with signicant path coecients. *** p < 0.01, ** p < 0.05, * p < 0.10 (one-tailed). n = 83 (signicant paths
for job tenure not shown).
157
this study show that comprehensive PMS positively aect managerial work performance by
enhancing role clarity and psychological empowerment. In contrast, prior research indicates that
the use of comprehensive PMS for performance
evaluation and compensation purposes can be
problematic (Ittner, Larcker, & Meyer, 2003; Lipe
& Salterio, 2000). As such, an important avenue
for future research is to investigate the simultaneous use of comprehensive PMS for providing
feedback and evaluating/compensating managers
(Sprinkle, 2000, 2003).
Acknowledgements
This paper is based on my PhD thesis completed at Monash University. I am grateful
for the support and advice of my PhD supervisor
Professor Robert Chenhall. Thanks to the two
anonymous reviewers, David Smith, Steve
Salterio, Margaret Abernethy, Sally Widener,
Shannon Anderson, Teemu Malmi, Peter Luckett,
Graeme Harrison, Laurence van Lent, workshop
participants at the University of Melbourne,
University of New South Wales, Macquarie University, University of Technology Sydney, University of Tilburg, and participants at the Global
Management Accounting Research Symposium
2004, Accounting and Finance Association of
Australia and New Zealand Annual Conference
2004 and New Directions in Management
Accounting: Innovations in Practice and Research
Conference 2004 for helpful comments.
158
Label
CPMS1
CPMS2
CPMS3
CPMS4
CPMS5
CPMS6
CPMS7
CPMS8
CPMS9
CPMS2, CPMS4, CPMS6 and CPMS8 were taken from Chenhall (2005). The other items were developed for this study.
Description
Comprehensive The performance measurement system consists of a diverse set of measures which
capture the key performance areas of the business unit. It presents information
about dierent aspects of the business units operations, which provides a
comprehensive and complete view of the business units performance
Partial
Label
Meaning
The work I do is very important to me
My job activities are personally meaningful to me
The work I do is meaningful to me
MEAN1
MEAN2
MEAN3
159
Appendix (continued)
Item
Label
Competence
I am condent about my ability to do my job
I am self-assured about my capabilities to perform my work activities
I have mastered the skills necessary for my job
COMP1
COMP2
COMP3
Self-determination
I have signicant autonomy in determining how I do my job
I can decide on my own how to go about doing my work
I have considerable opportunity for independence and freedom in how I do my job
SD1
SD2
SD3
Impact
My impact on what happens in my work area is large
I have a great deal of control over what happens in my work area
I have signicant inuence over what happens in my work area
IMP1
IMP2
IMP3
Label
Goal clarity
My duties and responsibilities
The goals and objectives for my job
How my work relates to the overall objectives of my work unit
The expected results of my work
What aspects of my work will lead to positive evaluations
GC1
GC2
GC3
GC4
GC5
Process clarity
How to divide my time among the tasks required of my job
How to schedule my work day
How to determine the appropriate procedures for each work task
The procedures I use to do my job are correct and proper
Considering all your work tasks, how certain are you that you
know the best way to do these tasks?
PC1
PC2
PC3
PC4
PC5
Label
MP1
MP2
MP3
160
Appendix (continued)
Item
Label
MP4
References
Abramis, D. J. (1994). Work role ambiguity, job satisfaction,
and job performance: meta-analyses and review. Psychological Reports, 75, 14111433.
Abramson, L. Y., Seligman, M. E. P., & Teasdale, J. D. (1978).
Learned helplessness in humans: critique and reformulation.
Journal of Abnormal Psychology, 87, 1974.
Anderson, S. W., Hesford, J. W., & Young, S. M. (2002).
Factors inuencing the performance of activity based
costing teams: a eld study of ABC model development
time in the automobile industry. Accounting, Organizations
and Society, 27, 195211.
Ashforth, B. (1989). The experience of powerlessness in
organizations. Organizational Behavior and Human Decision
Processes, 43, 207242.
Atkinson, A., & Epstein, M. (2000). Measure for measure:
realizing the power of the balanced scorecard. CMA
Management(September), 2228.
Baines, A., & Langeld-Smith, K. (2003). Antecedents to
management accounting change: a structural equation
approach. Accounting, Organizations and Society, 28,
675698.
Bandura, A. (1977). Self-ecacy: toward a unifying theory
of behavioral change. Psychological Review, 84, 191
215.
Banker, R. D., Chang, H., & Pizzini, M. (2004). The balanced
scorecard: judgmental eects of performance measures
linked to strategy. The Accounting Review, 79, 123.
Barclay, D., Thompson, R., & Higgins, C. (1995). The partial
least squares (PLS) approach to causal modeling: personal
computer adoption and use as an illustration. Technology
Studies, 2, 285309.
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator
variable distinction in social psychological research: conceptual, strategic and statistical considerations. Journal of
Personality and Social Psychology, 51, 11731182.
MP5
MP6
MP7
MP8
MP9
161
162
163