Description
Based on Harrison’s (Accountfng, Otganfzations and Sod&y, pp. 1-15, 1992; Accounting, Organ&&
tiom and Socfety, pp. 319-339, 1993) 6ndings on the general&ability of research results related to
budgetaty participation and budget emphasis between cultures, two tests on the three-way interaction
between budget emphasis, budgecuy partidpation and task characteristics (task uncertainty and task
diBiculty) atTect.injgo lx&ted tension and managerialp erformance were conducted in Singapore
Acown~ng,Orgnn~aHonrnndSOEIPtY,VoL. 2O,No.5,~p.359-381, 1995
Copyright 0 1995 PJsevier Science Ixd
PrlntcdinGnxt Britain.AUr&htsrcserved
0361~3682/95 $9.50+0.00
‘IWE IMPACT OF RELIANCE ON ACCOUNTING PERFORMANCE MEASURES ON
JOBRELATED TENSION AND MANAGERIAL PERFORMANCE: ADDITIONAL
EVIDENCE*
CHONG M. I&J
Edi th Cowan Uni versi ty
LIANG C. LOW
Si ngapore Pol ytechni c
and
IAN R. C. EGGLETON
The University of Western Austral i a
Based on Harrison’s (Accountfng, Otganfzations and Sod&y, pp. 1-15, 1992; Accounti ng, Organ&&
ti om and Socfety, pp. 319-339, 1993) 6ndings on the general&ability of research results related to
budgetaty participation and budget emphasis between cultures, two tests on the three-way interaction
between budget emphasis, budgecuy partidpation and task characteristics (task uncertainty and task
diBiculty) atTect.ing jolx&ted tension and managerial performance were conducted in Singapore. The
results support a high budget emphasis and high budgctaty participation combination in low task
difticulty skuations; and a high budgetary participation style, regardless of budget emphasis, in high
task dif6cuk.y situations.
The use of accounting data and the manner in
which those data are used by upper level man-
agement (superiors) to evaluate lower level
management (subordinates), and its effects on
subordinates’ job-related attitudes, budget-
related attitudes and performance have been
examined systematically by accounting
researchers commencing w&h the empiricaI
study by Hopwood (1972). while the research
findings of the earlier work of Hopwood (1972)
and OtIey (1978) were contIicting, more recent
work has apparent@ been able to resolve these
conflicting results (Brownell & Hirst, 1986;
BrowneII & Dunk, 1991). There is also evi-
dence to suggest the possibility of generalizing
the results of such research work done in the
AngIoAmerican block to the Asian block
CHwang, 1989; Harrison, 1992, 1993).
The focus of this paper is on the generaliz-
ability of the results of research on budgetary
participation and superiors’ evaluative styles
(specifically the Brownell & Him, 1986, and
Brownell & Dunk, 1991, studies) across
national cultures. We consider Harrison’s
’ The authors gratefully acknowledge the helpful comments of the editor and two anonymous mviewcrs. Special that&s
ucaacn~toAlvlDunLwho?asfsudintheevty~ofthisrrsearch.Anevllerverrsionofthlsplper~prrrnted
at the Account.ing % Finance Wo&shop, The University of Western AustfaLq the Accounting Research Group Seminar
Series, Edith Cowan University; and the 1994 University of New South Wales Management Accounting Research
Codel I Ce.
360 CHONG M. LAU et al.
(1992,1993) suggestion that research results on
budgetary participation are transferable across
the Anglo-American block and the Asian block,
whilst results on reliance on accounting perfor-
mance measures are not transferable. More
importantly, we provide evidence to support
Harrison’s research framework for examining
the Impact of national culture, not only on bud-
getary participation and reliance on accounting
performance measures, but also on other
aspects of the management accounting system.
With many nations in the Asian block experien-
cing fast economic growth, the adoption of
more sophisticated management accounting
systems and practices on a wider basis is
likely. Since a wealth of knowledge has been
accumulated through decades of research with
respect to the Anglo-American block, a systema-
tic approach to segregate those results that
could be generalized across cultures would facil-
itate the rapid adoption of improved manage-
ment accounting systems by Asian businesses.
Additionally, this study provides further
extensions to the research work in the area of
supervisory evaluative styles. Brownell & Dunk
(199 1, p. 703) suggested that the research work
in the area of supervisory evaluative style consti-
tutes an important “. . . organized critical mass
of empirical work in management accounting”.
Lindsay & Ehrenberg (1993, p. 223) considered
this area “one of the relatively few areas in man-
agement accounting research where there has
been any sequence of repeated studies”. From
the work of Hopwood (1972) to Harrison
(1993) management accounting researchers
have sought to study the relations between
superiors’ evaluative styles and subordinates’
attitudes and performance. Comprehensive
reviews of research work in this area have
been undertaken by Kren & Liao (1988) and
Briers & Hlrst (1990). Further research should
contribute to a better comprehension of these
relationships. Such understanding has impor-
tant implications for the design and implementa-
tion of management accounting control systems
as it may lead to changes in superiors’ budgetary
and evaluative styles to bring about desired
subordinates’ attitudes and job performance.
HYPOTHESIS DEVELOPMENT
Brownell & Hirst (1986) attempted to resolve
the conflict between Hopwood (1972) and
Otley (1978) by examining the three-way inter-
action between budget emphasis, budgetary
participation and task uncertainty. They sug-
gested that Brownell’s (1982a) finding that
compatible combinations of high (low) budget
emphasis and high (low) budgetary participa-
tion are associated with lower job-related ten-
sion and improved managerial performance
would hold in low task uncertainty situations,
but not in high task uncertainty situations. A
reward structure based primarily on budget
achievement is likely to motivate subordinates
only if they are heavily involved in the budget
setting process (Cherrington & Cherrlngton,
1973; Brownell, 1982a). If budget emphasis is
high, realistic and attainable budgetary goals
are crucial to subordinates because of the link
between budgetary goal achievement and
rewards. High budgetary participation is thus
necessary to ensure that the budgetary goals
set are attainable. In low task uncertainty envi-
ronments, complete task cause-effect knowl-
edge also facilitates the development of
unambiguous budgetary goals. Prior research
has found goal clarity to be an Important attri-
bute that impacts on performance (Early et al.,
1987). Consequently, high participation in a
high budget emphasis and low task uncertainty
situation is unlikely to be stressful to subordin-
ates as they are likely to agree with the choice
of evaluation criteria. Participation may
improve performance in a number of ways.
Kren & Llao (1988) suggested that participa-
tion may affect performance by acting
“through expectancies thus improving motiva-
tion; . . . (by providing) intrinsic valences for
performance, especially if it fosters group cohe-
siveness; . . . through the setting of higher bud-
get goals; by increasing commitment to budget
goals; and through the cognitive benefits
derived from sharing information during
participation” (p. 289).
In contrast, in low budget emphasis and low
task uncertainty situations, subordinates are
RELIANCE ON ACCOUNTING PERFORMANCE MEASURES 361
unlikely to be concerned with the accuracy of
budgetary goals because of the absence of a
link between budgetary goal achievement and
the reward system. In such situations, imposed
budgets may be more appropriate than partici-
pative budgets (Cherrington & Cherrington,
1973; Brownell, 1982a) because, with com-
plete task knowledge, there is no necessity
for subordinates to seek additional task knowl-
edge through participation. Superiors are also
unlikely to benefit from subordinates’ participa-
tion as information asymmetry is low. Participa-
tion therefore serves no useful purpose and
may be counterproductive, as subordinates
may be confused as to why they must waste
time participating in activities with seemingly
obvious solutions (Kren & Liao, 1988). Low
task uncertainty situations are likely to occur
in industries utilizing routine technologies for
the production of mature products, particularly
in developing Asian nations where labour
intensive industries are still fairly common.
In high task uncertainty environments,
budgetary participation will be beneficial
regardless of budget emphasis because “parti-
cipation may provide the opportunity for man-
agers to gain access to resources which can be
used to buffer task performance from the unan-
ticipated effects of others, and to introduce
new and better means for addressing tasks
which, if highly uncertain, will have character-
istics that change over time” (Brownell & Hirst,
1986, p. 242). It is also plausible that participa-
tion may enable superiors to gain input from
subordinates to improve operating techniques
(Kren & Liao, 1988). High budgetary participa-
tion therefore becomes the primary considera-
tion in high task uncertainty environments.
Brownell & Hirst’s (1986) results supported a
significant three-way interaction between bud-
get emphasis, budgetary participation and task
uncertainty affecting job-related tension, but
not managerial performance. Brownell &
Dunk (1991) attempted to resolve the failure
of Brownell & Hirst (1986) to support their
hypotheses with respect to managerial perfor-
mance. Task uncertainty comprises the subdi-
mensions of task difficulty (analysability) and
task variability (number of exceptions) (Van
de Ven & Delbecq, 1974). Of these two subdi-
mensions, task difhculty is the more appropri-
ate subdimension to measure to replicate the
findings of Brownell (1982a). When task dif&
culty is high, i.e. when the input and output
relations of the tasks are difficult to specify or
unclear, there is a greater need for budgetary
participation to counter the incomplete task
cause-effect knowledge. Task variability, on
the other hand, refers to the variability of sub
tasks. High task variability may arise because of
many different subtasks, but if each one is
easily analysable in terms of input and output
relations, the task cause-effect knowledge may
be complete (Perrow, 1967, 1970; BrownelI &
Dunk, 1991). With prescribed rules, policies
and procedures for performing the job
(Tushman 8~ Nadler, 1978) there would be
less need for participation. Brownell’s (1982a)
results are therefore expected to apply only to
situations that differ with respect to task diffi-
culty, not task variability. The results of
Brownell & Dunk (1991) supported a three-
way interactive and positive effect between
budget emphasis, participation and task diffi-
culty on performance but, as expected, only
in low task difficulty situations. There were
no significant interactive effects on perfor-
mance when tasks differed along the variability
continuum.
Hwang (1989) and Harrison (1992) extended
this body of research by studying the cross-
cultural generalizability of research results relat-
ing to budgetary participation. Hwang (1989)
examined the moderating effect of power dis-
tance on the relationship between budgetary
participation and motivation. Harrison (1992,
1993) examined the moderating effect of
national culture on the relationship between
budget emphasis and both job-related tension
and job satisfaction.
Harrison’s (1992) research framework was
based on two of Hofstede’s (1980) dimensions
of culture - power distance and individualism
-
which were hypothesized to be related to
budgetary participation. Hofstede (1980) sug-
gested that participation of the lower-placed
362 CHONG M. LAU et al.
groups in important affairs is not encouraged in
high power distance societies because such
societies elaborate, institutionalize and forma-
lize dominance behaviour. Budgetary participa-
tion is therefore likely to be “pseudo” and
unlikely to be beneficial. Low power distance
societies regard humans as equal and attempt
to minimize hierarchies. In such societies, the
participation of all individuals in affairs that
affect everybody is favoured. Budgetary partici-
pation is likely to be genuine and beneficial.
High individualism societies regard each indivi-
dual, and not the group, as the basic unit of
society. Since the focus is on the individual,
decision-making on an individual (superior)
basis is likely to be the norm. Budgetary parti-
cipation is unlikely to be favoured. In contrast,
low individualism societies regard the group to
which the individual belongs as the basic unit
of society. In such collective societies, partici-
pation of all forms, including budgetary partici-
pation, is favoured and likely to be beneficial. It
would seem that budgetary participation
should be encouraged in low power distance
and low individualism societies and discour-
aged in high power distance and high individu-
alism societies. However, Hofstede’s (1983)
study of 50 nations shows that a large majority
(47) have either a low power distance and high
individualism culture (which includes the
Anglo-American block) or a high power dis-
tance and low individualism culture (which
includes the Asian block). Within the Anglo-
American block, the low power distance
which favours budgetary participation is likely
to be counterbalanced by the high individual-
ism which does not favour participation. These
opposing factors could effectively neutralize
the cultural effects on subordinates’ reactions
to budgetary participation. Similarly, the
opposing effects on participation of the two
cultural dimensions within the Asian block cul-
ture could also neutralize their effects on sub
ordinates’ reactions to budgetary participation.
Thus, Harrison (1992) hypothesized that
research results on budgetary participation
obtained in the Anglo-American block could
be generalized to the Asian block. Harrison’s
(1992) results not only reaffirmed Hofstede’s
(1983) classification of Singapore as a nation
of high power distance and low individualism
and Australia as a low power distance and high
individualism nation, they also indicated no sig-
nificant two-way interaction between nation
and participation. This suggests that the effects
of participation on the relation between budget
emphasis and jobrelated tension on one hand,
and job satisfaction on the other, is not mod-
erated by differences in the Australian and
Singaporean cultures.
With regard to budget emphasis, Harrison
(1993) suggested that high power distance
societies are likely to prefer a high budget
emphasis evaluative style because a “prefer-
ence for a non-consultative, decisive leader-
ship style in high PD (power distance)
societies is also likely to produce a preference
for high RAPM (reliance on accounting perfor-
mance measures) in evaluative style. This style
accords with the decisive leadership character-
istic of coming to a decision and sticking to it”
(p. 322). Low individualism societies are also
likely to accept a high budget emphasis evalua-
tive style (which assumes that people can be
compared) because low individualism de-
emphasizes individual differences which
therefore facilitates comparability among indi-
viduals. Thus, high power distance and low
individualism Asian nations are likely to react
favourably to a high budget emphasis evaluative
style.
By contrast, subordinates in low power dis
tance societies will not react favourably to a
high budget emphasis evaluative style because
it “effectively blocks the subordinate from
representation in the evaluation process and
thereby frustrates the realization of equality in
the subordinate’s relations with his/her
superior” (Harrison, 1993, p. 322). Similarly,
subordinates in high individualism societies
are also likely to react unfavourably to a high
budget evaluative style because the assumption
of comparability among individuals with high
budget emphasis is inconsistent with the
emphasis on the uniqueness of individuals in
these societies. Consequently, low power
RELIANCE ON ACCOUNTING PERFORMANCE MEASURES 363
distance and high individualism societies such
as those in the Anglo-American nations are likely
to prefer a low budget emphasis evaluative
style.
Harrison (1992) therefore concluded that
“clear and different directional expectations
could be projected for the effect of culture
on the relation between budget emphasis and
the dependent variables in Singapore and
Australia. . . (but) no such directional expecta-
tion could be projected for the effect of culture
on the relation between participation and the
dependent variables” (p. 12).
Brownell & Hirst (1986) and BrownelI &
Dunk (1991) have provided plausible explana-
tions for the conflict between Hopwood (1972)
and Otley (1978) relating to job-related tension
and managerial performance, respectively.
Hwang (1989) and Harrison (1992, 1993)
have provided the theoretical justification to
extend these results for Anglo-American
nations to Asian nations. However, a number
of issues merit further investigation. First,
results of prior empirical studies on the posi-
tive effect of participation, particularly on man-
agerial performance, are Inconsistent and
therefore lack broad generalizability (Brownell,
1982b; Brownell & Hirst, 1986; Hirst, 1987;
Kren & Liao, 1988; Brlers & Hirst, 1990;
Lindsay & Ehrenberg, 1993). Kren & Liao
(1988) suggested that “existing evidence is still
limited and additional studies in different set-
tings and using different subject samples
would be very beneficial” (p. 298). Lindsay
& Ehrenberg (1993) concluded that “the
results add up to little by way of robust,
generalizable findings” @. 224). Otley et al.,
(1994) suggested that “replication is a much
neglected aspect of accounting research. The
value of repeating previous work is that it
allows a better assessment of the confidence
that should be placed in specific findings . . .”
@. 30). It would also reduce the chance of
accepting false hypotheses and results that are
“a total flash in the pan” (Lindsay & Ehrenberg,
1993, p. 220). Thus there is a need to cor-
roborate and solidify the results of earlier
studies and to establish the range of different
conditions (e.g. different organizations and dif-
ferent cultures) under which these findings will
hold. Second, Hwang’s (1989) study focused
primariIy on participation and motivation.
Harrison’s (1992, 1993) emphasis was on job
related tension and job satisfaction. Examining
managerial performance systematically may
support or reject his suggestion (Harrison,
1992) that the results of BrownelI (1982a)
could be generalized to high power distance
and low individualism nations. Furthermore, a
systematic examination of a three-way interac-
tion between budget emphasis, participation
and task uncertainty consistent with Brownell
& Hirst (1986) has not been undertaken in a
high power distance and low individualism
nation such as Singapore. Although there was
matching of task in Harrison’s (1992) study, the
effects of task uncertainty were not systematic-
aliy examined. Finally, the effects of budgetary
participation on the relation between budget
emphasis and managerial performance in high
task difficulty situations is also untested in high
power distance/low individualism nations.
Given these gaps in the current literature and
the significant results pertaining to job-related
tension and task uncertakry in BrownelI &
Hirst (1986); and managerial performance and
tusk di fJ i cul ty i n BrownelI & Dunk (1991), this
study examines the moderating effect of(i) task
uncertai nty and participation on the relation
between budget emphasis and j ob-rel ated ten-
si on and (ii) task d@i cuZty and participation on
the relation between budget emphasis and
manageri al performance using Singapore as
the test site.
The following two sets of null and related
alternative hypotheses are tested:
Set 1: Task uncertai nty and j ob-rel ated
tensi on
I & There i s no stati sti cal l y sigoificant three-way
interaction between budget emphasis, budget-
ary partidpation and task uncwkzinry affecting
j ob-Mated tensi on.
HA,, Compatible combinations of high sow) budget
emphasis and bigb (low) budgw participation
364 CHONG M. LAU et al.
are associated with lowerjob-rekated tension in
low task uncertainly situations.
Regardless of the types of supervisory evaluative
style, high budgetary participation is associated
with lower j ob-rel ated tensi on i n hi gh task
uncertainty situations.
Set 2: Task difficulty and managerial
performance
H
**a
HAab
There is no statistically sign&ant three-way
interaction between budget emphasis, budget-
ary participation and task dtQ@zr&y affecting
manageri al perfownance.
Compatible combinations of high (low) budget
emphasis and high (low) budgetary participation
are associated with improved rnunage&zZ per-
formance i n l ow task di ffi cul ty situations.
Regardless of the types of supervisory evaluative
style, high budgetary participation is associated
with improved manageri al performance i n hi gh
task d@i cul ty situations.
As the cultural differences between Singapore
and Australia are expected to affect only sub
ordinates’ reactions to budget emphasis, but
not subordinates’ reactions to participation
(Harrison, 1992, 1993) the extent to which
the results of Brownell & Hirst (1986) and
Brownell & Dunk (1991) will be supported in
Singapore will depend mainly on the influence
of culture on Singaporean managers’ reactions
to budget emphasis. With a high power
distance and low individualism culture in
Singapore (Hofstede, 1980, 1983; Harrison,
1992, 1993), Singaporean managers are
expected to react favourably to a high budget
emphasis evaluative style. As these reactions
are opposite to those of Australian managers,
our Singapore results are therefore expected to
be different from those of Brownell & Hirst
(1986) and Brownell & Dunk (1991). Although
the extent of these differences are indetermin-
able, directional predictions are possible.
With regard to low task difficulty (uncer-
tainty) situations, both Brownell & Hirst
(1986) and Brownell & Dunk (1991) have
found the high budget emphasis and high parti-
cipation combination to be superior (in terms
of lower job-related tension and improved man-
agerial performance, respectively). This combi-
nation is expected to be a superior combination
in Singapore as well since the high budget
emphasis component is compatible with the
Singaporean culture while the high participa-
tion component is unaffected by the culture.
Thus a high budget emphasis and high partici-
pation combination in low task difficulty
(uncertainty) situations is expected to be asso-
ciated with improved managerial performance
(lower jobrelated tension) in Singapore.
Apart from the high budget emphasis and
high participation combination, both Brownell
& Hirst (1986) and Brownell 81 Dunk (1991)
have also found the low budget emphasis and
low participation combination to be superior in
low task difficulty (uncertainty) situations.
Since a low budget emphasis strategy is incom-
patible with the Singaporean culture, this com-
bination may not be a desirable combination in
Singapore. Hence, unlike in Australia, a low
budget emphasis and low participation combi-
nation in low task difficulty (uncertainty) situa-
tions is not expected to be associated with
improved managerial performance (lower job
related tension) in Singapore.
With regard to high task difficulty (uncer-
tainty) situations, both Brownell & Hirst
(1986) and Brownell & Dunk (1991) have
found a high budgetary participation strategy,
regardless of budget emphasis, to be superior.
Since culture has no effect on participation, a
high budgetary participation strategy, regard-
less of budget emphasis, is also expected to
be a superior strategy in Singapore. In addi-
tion, since the Singaporean culture favours a
high budget emphasis evaluative style, a high
budgetary participation and high budget
emphasis combination is also likely to be a
desirable combination in Singapore.
METHOD
A questionnaire was administered to a sample
of 240 functional heads from 80 Singapore
manufacturing companies with more than 100
RELIANCE ON ACCOUNTING PERFORMANCE MEASURES 365
employees each.’ The companies were
sampled randomly from the list of manufactur-
ing firms published in Kompass: Register of
Industry and Commerce of Singapore
(1992). Singapore was chosen as a surrogate
for the Asian block of nations with high power
distance and low individualism cultural dimen-
sions (Hofstede, 1980, 1983, 1984; Harrison,
1992, 1993). As it is also English speaking, the
difficulties associated with the translation of
the questionnaire were avoided (Chow et al.,
1991; Harrison, 1992). Manufacturing compa-
nies were selected to determine if Harrison’s
(1992) results (on the cross-cultural generaliza-
tion of results on budgetary participation),
based only on departmental and retail stores,
could be extended to manufacturing compa-
nies where the use of budgets is common.
Eighty companies, representing about 15% of
all manufacturing companies in Singapore
with more than 100 employees each, were con-
sidered adequate to ensure external validity.
Only companies with more than 100 employ-
ees each were chosen because smaller compa-
nies may not be organized on a work-unit
(departmental) basis (Brownell & Dunk,
1991). The requirement for companies to be
organized on a work-unit basis is important
because the measurement instruments for
task uncertainty (Withey et al., 1983) and task
difficulty (Van de Ven & Delbecq, 1974) used in
this study are designed to be used on a work-
1 Copies of the questionnaire are available from the authors
unit basis (Van de Ven & Delbecq, 1974;
Withey et al, 1983; Brownell & Dunk, 1991).
Three respondents from three different func-
tional areas including the heads of produc-
tion, sales or marketing and a staff functional
area (e.g. personnel, administrative or account-
ing) were selected from each company allow-
ing the research results to be generalized
across functional areas.* The names of the func-
tional heads were obtained by telephone calls
to the companies to ensure that the question-
naire would reach, and be answered, only by
the selected respondents. Only functional
heads were selected as they were regarded as
surrogates for work-units. Moreover, one res-
pondent from each functional area was selec-
ted to ensure independence of observations.
A questionnaire with a covering letter
explaining the objectives of the study and
assuring anonymity of response was mailed to
each respondent. A self-paid addressed envel-
ope was also enclosed for the completed ques
tionnaire to be returned directly to one of the
researchers in Singapore. Follow-up of the
questionnaire was by telephone calls to those
who had not responded after 3 weeks.
A total of 129 responses were returned yield-
ing a response rate of 53.8%. Of the 129
responses, 3 were excluded from the study
for incomplete responses and 14 for unreliable
responses.3 This leaves the study with 112
usable responses.4 Of these, 49 were from the
* The staff functional heads (non-production and non-marketing or sales) were not selected randomly as most companies
contacted were reluctant to supply the researchers with a complete list of their staIf functional heads. Consequently, the
questionnaire was mailed to the staff functional heads suggested by the companies and these were genetally from
administration, personnel or accounting.
3 The questionnaire includes a seven-item measure for task dilIiculty. Of the seven items, two items are polar reverse. The
mean score of the two polar reverse items was compared with the mean score of the other five items for each respondent
to isolate any oudiers. Fourteen responses have a difference of 4 or more out of a maximum of 6. The coefficient of
correlation (r) of the mean score of the two poIar reverse questions and the mean score of the other five items of these 14
responses was a very high negative 0.83. This justihes the exclusion of these 14 outIiers from the study. There are no
signiticant differences between the regression results based on 112 responses (after the exclusion of the 14 responses) and
those based on 126 responses (before the exclusion of the 14 responses).
* The presence of non-response bias was investigated by the test suggested by Oppenheim (1!966, p. 34). No statistIcaIIy
significant differences in the mean scores between the earIy and Iate responses were found.
366 CHONG M. LAU et al.
production function, 19 from the sales or mar- styles separately and found significant differ-
keting function and 44 from staff functions. ences in job-related tension and negative atti-
The mean age of the respondents was 38 years tudes between the extent (Budgetconstrained
and 90% of them had tertiary qualifications.
and Budget-profit styles) and the manner
They had held their current positions for an
(Frofit-conscious style) of the reliance on
average of 4.5 years and had been employed
accounting measures in performance evalm-
by their respective companies for 7.7 years.
tion, there is also a theoretical justification to
The average length of experience in the areas
evaluate budget emphasis on the scores of the
they managed was 9 years and the average
“Meeting the budget” item alone instead of
number of employees in their areas of
summing the two accounting items. We tested
responsibility was 109.
our hypotheses using both approaches and
found no significant differences in the results.
Measurement of variables
This could be due to the relatively high correla-
Budget emphasis. Consistent with Brownell
tion between the two accounting items in our
& Hirst (1986), budget emphasis was measured
study.5 Given the absence of significant differ-
with a ten-item instrument which is a modified
ences in our results and the high correlation
version of that developed by Hopwood (1972).
between the two accounting items, the results
Since the ten-item instrument also incorporates
based on the combined scores were used in
all the eight items included in the instrument
this study so that any differences in the results
used in Brownell & Dunk (1991), it could also
between this study and those of Brownell &
be regarded as being consistent with Brownell
Hirst (1986) and Brownell & Dunk (1991)
& Dunk (1991). The rating form, rather than
could be attributed to cultural differences
the ranking form, was used as the former per-
and not to a change in the measurement
mits the evaluative style to be measured as a
instrument.
continuum (Otley, 1978; Govindarajan, 1984;
Budgetaly participation. The Milani (1975)
Brownell & Dunk, 1991; Harrison, 1992) and
six-item, seven-point Like&scaled instrument
avoids the problem (reported by BrowneIl,
was used to measure budgetary participation.
1985) encountered by respondents with the
This instrument was used by both Brownell &
ranking instrument. Consistent with Brownell
Hirst (1986) and Brownell & Dunk (1991). It
& Dunk (1991), the scores of the two account-
has also been tested and used extensively in
ing items (“Meeting the budget” and “My con-
other studies, and has provided high internal
cern with costs”) were combined to provide a
reliability (Brownell, 1982a, 1985; Chenhall h
score for budget emphasis. The objective of
Brownell, 1988; Dunk, 1989; Mia, 1989; Frucot
combining the two items is to operationalize
& Shearon, 1991; Harrison, 1992). The
budget emphasis “along a continuum from
Cronbach alpha (Cronbach, 1951) of 0.91
accounting to non-accounting style of evalua-
obtained for this study indicates high internal
tion” (Browneli & Dunk, 1991, p. 702). How-
reliability.
ever, since Hopwood (1972) had categorized Task uncertainty. The nine-item, seven-point
the Budget-constrained and Profit-conscious Likert-scaled instrument developed by Withey
’ The two accounting items have a coef5cient of correlation of 0.56 (p 4 0.01). As Eu as the “Meeting the budget” item is
concerned, this represents the highest correlation it has with any bf the other nine items. As far as the “My concern with
costs” item is concerned, this is also the highest correlation it has with any of the other items except for the item,
“Concern with quality” which also has a correlation coelkient of 0.56 with the “My concern with costs” item. Brownell
(1985) has argued that the legitimacy of w the two account@ items is dependent on a signi5cant correlation
between the responses to the two items. Hopwood’s (1972) Budget-profit styie was also made up of respondents who had
ranked both “Meeting the budgets” and “My concern with costs” in the top three at c gor i es and he found no s@Wicant
differences in his results between the Budget-pro5t style and the Budget~onstmined style.
et al. (1983) was employed to evaluate task despite the low Cronbach alpha scores. The
uncertainty. This instrument was employed by Cronbach alpha obtained in this study was
Brownell & Hirst (1984) and represents an addi- 0.41. Deleting the two polar reverse items
tive scale for the composite measure of task increases the Cronbach alpha to 0.52 which
difficulty and task variability. The scale was is close to that obtained by Brownell & Dunk
used by Brownell & Hirst (1986) because it (1991). However, as there were no significant
was consistent with Thompson’s (1967) defti- differences between the regression results
tion of task uncertainty and because the scale based on the scores of the seven-item instru-
had been tested and found to be reliable by ment and those based on the scores of the
Withey et al. (1983). The Cronbach alpha revised five-item instrument, the results based
(Cronbach, 1951) of 0.85 obtained in this on the score of the seven-item instrument were
study indicates satisfactory internal reliability reported to ensure consistency with Brownell
and is close to the 0.82 obtained by the & Dunk (1991). Given the low alpha in our
Brownell & I-I&t (1986). study, the reliability of task difficulty is ques-
Tusk dificulty. Task difficulty was evaluated
tionable. Hence, caution should be exercised
with the Van de Ven & Delbecq (1974) seven-
in interpreting the results of tests involving
item instrument. A seven-point rating scale con-
this variable. Remedial strategies are discussed
sistent with that used by Brownell & Dunk
in the final section of this report.
(1991) was used.6 This subscale was tested by J ob-related tension. Consistent with Brownell
Withey et al. (1983) along with five other simi- & Hirst (1986) job-related tension was mea-
lar scales developed by Hage & Aiken (1969) sured with a seven-point Likert-scaled, fifteen-
Lynch (1974) Glisson (1978) Van de Ven & item, job-related tension index (Kahn et al,
Ferry (1980) and Daft & Macintosh (1981). 1964) developed by the Institute for Social
The Van de Ven & Delbecq (1974) scale and Research at the University of Michigan. The
the Daft & Macintosh (1981) scale were scores on this index were found to be corre-
regarded by Withey et al. (1983) as the two lated positively with role conflict, role ambi-
most appropriate scales for research as they guity and anxiety (Kahn et aZ., 1964, pp. 424-
both have reasonable measurement properties 427). The Cronbach alpha obtained in this
in terms of the extent to which the two instru- study was 0.87.
ments measure the same construct as other Managerial perjomance. The ninedimen-
instruments (from separate studies of work- sional seven-point Likert-scaled Mahoney et al.
unit technology), and in terms of their relative (1963, 1%5) self-rating measure was employed
convergent and discrlminant validity compared to evaluate this dependent variable. The scale
with other similar instruments. In their study, comprises eight performance dimensions and
Van de Ven & Delbecq (1974) reported a reli- one overall effectiveness dimension. This self-
ability alpha of 0.86 for the task difficulty rating scale has been criticized for the ten-
instrument. However, the Cronbach alpha dency of respondents to be too lenient on
(Cronbach, 1951) obtained by Brownell & themselves, thereby resulting in a small range
Dunk (1991) with this scale was a low 0.57. in the scores being observed (Prien & Liske,
Another study by Mia (1987) with this scale 1%2; Thornton, 1968; Mia, 1989). However,
achieved a similarly low alpha value of 0.55. according to Brownell (1982a), it has the
Since task difficulty was an important variable advantages of overcoming the problem of
in Brownell & Dunk (I991), the same additive superiors’ tendency “to evaluate on only one
scale instrument was used in this study to facil- cognitive dimension” and to “clearly capture
itate comparability of studies between cultures, the multidimensional structure of performance
RELIANCE ON ACCOUNTING PERFORMANCE MEASURES 367
o In their e work, Van de Ven % Delbecq (1974) used a ten-item interval scale.
368 CHONG M.
without introducing the problem of excessive
dimensional&y” (pp. 17-18). Heneman (1974)
also found a very high correlation between
superior-rating and self-rating scales. Consis-
tent with both Brownell & Hirst (1986) and
Brownell & Dunk (1991), only the score of
the overall effectiveness dimension was used
for managerial performance in this study.
Regressing the score of the overall effective-
ness item on the scores of the other eight
items provided a coefficient of determination
of 0.61 (p < 0.001) which is higher than the
0.55 (desirable lower limit) suggested by
Mahoney et al. (1963). The remaining 39% of
unexplained variance is probably job specific
or related to individual differences and sug-
gests that factors other than those covered by
the questionnaire also influence overall man-
agerial performance. Given the cross-sectional
nature of our sample and the fact that we have
used the instrument in a different culture from
that within which it was developed, this level
of R2 is not surprising.
Descriptive statistics for the variables meas-
ured in our study are shown in Table 1 and the
intercorrelations among the variables are
LAU et al.
shown in Table 2. The Pearson pairwise corre-
lations among the independent variables are
not high (less than 0.5). In addition, because
interaction regression models were used in
this study, the approach of centring the inde-
pendent variables (by using deviation scores) is
adopted to reduce the correlations between
the product terms and the component parts
of the independent variables (Southwood,
1978; Cohen & Cohen, 1983; Cronbach, 1987;
Jaccard et cd., 1990). Tolerance of greater than
0.1 was achieved for all independent variables
in the regression models used. Multicollinearity
was therefore not considered to be a problem.
RESULTS AND DISCUSSION
Hierarchical regression models were used to
analyse our results. This approach was selected
to facilitate the partitioning of the proportion
of the total variance accounted for by the addi-
tive model, the two-way interactive model and
three-way interactive model (Cohen & Cohen,
1983). Since the main effects of the three-way
interaction model are arbitrary and have no
TABLE 1. Descriptive statistics for the variables in the study (n = 112)
Theoretical range Actual range
Variable Mean S.D. Min. MaX. Min. MaX.
Budgetary participation 30.6518 7.6320 6 42 6 42
Budget emphasis 11.66% 2.1619 2 14 2 14
Task uncertlinty 37.2320 9.2440 9 63 18 57
Task difficulty 24.4911 4.6557 7 49 14 39
Job-related tension 47.6964 12.0720 15 105 15 80
Managerial performance 5.79% 0.8398 1 7 4 7
TABLE 2. Correhtion matrix among dependent and independent variables
Jobrelated
Variable Performance tension Participation Budget emphasis
Job-related tension -0.31”
Participation -0.39” 0.15
Budget emphasis -0.42” 0.13 0.45”
Task uocertainty 0.09 0.03 0.09 0.01
Task difficuky 0.21. -0.22. -0.01 -0.07
* p c 0.01.
l p s 0.05.
RELIANCE ON ACCOUNTING PERFORMANCE MEASURES 369
theoretical meaning as we have used non-ratio
scales (Allison, 1977; Southwood, 1978;
Schoonhoven, 1981) the hierarchical
approach also permits the main effects and
the interactive effects of the independent vari-
ables to be analysed separately. The following
models were used:
Yi = bo +blPi +bzB* +b$i +et
(1)
Yr = bO +b,P, +b2B, +b3T, +b4pI Bi +
bP& +b6 BA +ei (2)
Yi = bO +b,P, +b2B, +b& +bflJ .3, +
bP& +b&J , +b7P&Wi +ei, (3)
where Yi is job-related tension (HOI) or manage-
rial performance @I,&; Pi is budgetary partici-
pation; Bi is budget emphasis; T, is task
uncertainty ml> or task difficulty @I&; and
ef is the error term.
Equations (1) (2) and (3) were used to ana-
lyse the main effects, the two-way interactive
effects and the three-way interactive effect,
respectively. Deviation scores based on the
overall mean score minus the raw scores
were used for all independent variables in the
regression models. Hence, low raw scores have
positive deviation scores and high raw scores
have negative deviation scores. This enables
the coefficient of the two-way interaction
between budget emphasis and participation in
low task uncertainty situations to be predicted
(we elborate upon this later). Tests on the ade-
quacy of the multiple linear regression models
indicate that the inherent assumptions of the
models were satisfied by the data7
Hypothesfs HOI
Hypothesis HoI states that there is no statis-
tically significant three-way interaction between
budget emphasis, budgetary participation and
task uncertafnty affecting job-related tensfon.
To test this hypothesis, job-related tension was
regressed on budget emphasis, participation
and task uncertainty. The results in Table 3
(equation (3)) indicate that the three-way inter-
action between budget emphasis, participation
and task uncertainty affecting job-related
tension is not significant as reflected by the
insignificant b, coefficient (p d 0.791). Accord-
TABLE 3. Results of regression of jobrelated tension on budget participation, budget emphasis and task uncertainty
Equation (I) Equation (2)
Bquation (3)
(additive)
(tJvov=y) (three-way)
Variable Coeff. Est.
P
Est.
P
Est.
P
constant
bo
47.6964
Budget participation (P) bl 0.1889
Budget emphasis (B)
bz 0.4390
Task uncen.dnty (2’)
bs
0.0269
PXB
b4
PXT
b5
BXT
b6
PXBXT
b7
R2 0.030
F value 1.110
PG
0.348
J? explained by twoway interaction terms = 4.5%
If explained by three-way interaction term = 0%
0.000 48.1%2
0.267 0.2405
0.463 0.6208
0.829 0.0733
-0.0409
-0.0308
0.0309
0.075 0.075
1.414 1.211
0.216 0.303
0.000 48.21% 0.000
0.161 0.2305 0.192
0.329 0.6303 0.325
0.566 0.0593 0.669
0.429 -0.0500 0.422
0.052 -0.0279 0.152
0.619 0.0206 0.778
0.0024 0.791
‘The tests to check for the three important assumptions of the selected regression models (normality of residuals,
homogeneity of variance of residuals and the appropriateness of the linear models) indicate that the assumptions inherent
in the regression analysis were satisfied in all equations.
370 CHONG M IAU et al .
ingly, hypothesis HoI cannot be rejected. This
result is inconsistent with Brownell & Hirst’s
(1986) finding. No significant results were
found in the two-way interaction model nor
in the additive model.
our failure to control for the effect of uncer-
tainty avoidance on job-related tension.
Hypothesis Ho2
To determine if Brownell & Dunk’s (1991)
observation that the subdimension of task diff?-
culty, rather than task uncertainty, may be a
more relevant variable to measure task chamc-
teristics, job-related tension was regressed on
budget emphasis, participation and task dfffi-
c&y. This three-way interaction was also
insignificant (p s 0.49).
A possible explanation for our failure to
replicate the results of Brownell & Hirst
(1986) could be due to differences in the
sampling methods between the two studies.
Brownell & Hi&s (1986) sample was selected
from a single manufacturing company, whereas
our sample was selected from 80 different man-
ufacturing companies. Alternatively, cultural
differences could have infhtenced the results.
Our study has considered the influence of the
two underlying cultural dimensions of power
distance and individualism on employees’ reac-
tions to budgetary participation; but has not
considered the influence of the cultural dimen-
sion of uncertainty avoidance. Uncertainty
avoidance refers to the ways societies adapt
to uncertainty in life. Low uncertainty avoid-
ance countries have lower anxiety levels in
the population and lower job stress (Hofstede,
1980, p. 176). High uncertainty avoidance
nations have higher anxiety in the population
and higher job stress. Since the dependent vari-
able in HoI is job-related tension, it is feasible
that cultural differences in respect to uncer-
tainty avoidance could influence it. The range
of uncertainty avoidance index for the 40
nations studied by Hofstede (1980) is 8-112.
Significantly, Singapore has a score of 8 while
Australia has a score of 51. Given this substan-
tial difference in the score between the two
nations, it is feasible that our failure to repli-
cate Brownell & Hirst (1986) could be due to
Q, states that there is no statistically sign&
cant three-way interaction between budget
emphasis, budgetary participation and task dff-
@ulty affecting managerial momrance. To
test this hypothesis, managerial performance
was regressed on budget emphasis, participa-
tion and the subdlmension of task difficulty.
The results in Table 4 (equation (3)) show a
significant three-way interaction between bud-
get emphasis, participation and task difficulty
affecting managerial performance (p 6 0.032).
Hypothesis H,,, can therefore be rejected. This
implies that Brownell’s (1982a) results cannot
be generalized across all levels of task difficulty.
To test Brownell & Dunk’s (1991) assertion that
task difficulty and not task uncertainty is the
task characteristic associated with perfor-
mance, the three-way interaction between bud-
get emphasis, budgetary participation and task
uncertainty affecting managerial pe$or-
mance was also tested. As expected, no signif-
icant three-way interaction was found.
Table 4 indicates that the two-way interac-
tions (equation (2)) and the three-way interac-
tion (equation (3)) only enhance the predictive
ability of the model by 4.6% and an additional
3%, respectively. This suggests that the two-
way interactions and the three-way interaction
(although statistically significant) may have less
practical significance, relative to the main
effects (26.8%). However, we are unable to
compare the main effects of our study with
those of Brownell & Dunk (1991) as they
have not disclosed the main effect results in
their article.
To determine if compatible combinations of
high (low) budget emphasis and high (low)
participation is associated with improved per-
formance only in low task difficulty situations,
task difficulty was dichotomized at its mean.’
’ There arc no si gni f i c ant di f f er enc es i n t he r egr essi on r esul t s based on dichotomizing task diiliculty at the mean and those
obtained when task difiiculty is dichotomized at the median.
RELIANCE ON ACCOUNTING PERFORMANCE MEASURES 371
TABLE 4. Results of regression of e performance on budget participation, budget emphasis and task diGculty
Variable Coeff.
Equation (1)
(additive)
Est.
P
Equation co Equation (3)
(*~~y~ (thrrcway)
Est.
P
Est.
P
constant
Budget participation (9
Budget emphasis (s>
Task difficulty (T,
PXB
PXT
BXT
PXBXT
b0
5.7946 0.000 5.7699 0.000 5.7593 0.000
bl -0.0284 0.006 -0.0321 0.002 -0.0365 0.001
bz
-0.1138 0.002 -0.1234 0.002 -0.1078 0.006
b,
0.0350 0.021 0.0307 0.039 0.0162 0.311
b4
0.0031 0.306 0.0044 0.155
bs
0.0054 0.032 0.0055 0.026
b6
-0.0042 0.630 -0.0113 0.223
b7
0.0019 0.032
R2 0.268 0.314 0.344
F value 13.152 8.010 7.781
P S 0.001 0.001 0.001
J? explained by two-way interaction terms = 4.6%
I? explained by three-way interaction term = 3.0%
The following regression models were used to
test the main effects and the twoway interac-
tion effect between budget emphasis and parti-
cipation for each of the low and high task
difGculty situations:
Yi = co + clPi + czBr + el
(4)
Yi = co + clPi +c& +c$&l, +ei,
(5)
where Y, is managerial performance; Pi is bud-
getary participation; & is budget emphasis; and
ef is the error term.
Equations (4) and (5) were used to analyse
the main effects and the two-way interaction
effect, respectively. Since deviation scores
based on the overall mean less raw scores
were used for all independent variables, high
participation and high budget emphasis have
negative deviation scores and low participa-
tion and low budget emphasis have positive
scores. The product of the deviation scores of
low budget emphasis and low participation
would be the same as the product of the devia-
tion scores of high budget emphasis and high
participation. This enables the direction of c3,
the coefficient for the twc~~3y interaction
between budget emphasis and participation
to be predicted. Since compatible combina-
tions of high (‘low) budget emphasis and high
(low) participation are expected to be asso-
ciated with improved performance only in
low task difficulty situations, c3 was expected
to be positive in low task difficulty situations.
The results of the regression are shown in
Table 5 for the low task difhculty condition
and in Table 6 for the high task ticulty
condition.
As expected, the interaction between parti-
cipation and budget emphasis on managerial
performance was found to be sign&ant and
the coefficient positive for the low task diffr-
culty subsample (p G 0.016 in Table 5, equa-
tion (5)), but not for the high task diBiculty
subsample (p G 0.944 in Table 6, equation
(5)). Note that with the low task difficulty sub-
sample (Table 5), the J? of the regression
model with the interactive term (0.271 in equa-
tion (5)) is considerably higher than the Ra of
the additive model (0.191 in equation (4)).
However, in the case of the high task difficulty
subsample (Table 6), the value of I? of the
model witb the interactive term (0.295 in equa-
tion (5)) is the same as that of the additive
model (0.295 in equation (4)). These results
support hypothesis HAzp that budget emphasis
interacts with participation onty in low task
difficulty situations. Thus the null hypothesis
372 CHONG M. LAU et al.
TABLE 5. Results of regression of managerial performance on budget participation and budget emphasis - low task
difficulty subsample
Equation (4) Equation (5)
(additive)
@weway)
Variable Coeff. Est.
P
Est.
P
constant CO 3.%15 0.000 5.8070 0.000
Budget participation (P) cl 0.0089 0.468 -0.0172 0.164
Budget emphasis (B) cz 0.1408 0.003 -0.1450 0.002
PXE
c3
0.0117 0.016
ti 0.191 0.271
F value 6.729 6.929
PS 0.002 0.001
ti explained by twc+way interaction terms = 8%
TABLE 6. Results of regression of managerial performance on budget participation and budget emphasis - high task
difliculty subsample
Equation (4) Equation (T)
(additive)
@weway)
Variable Caeff. Est.
P
Est.
P
constant CO 3.2606 0.000 5.7024 0.000
Budget participation (P) Cl 0.0528 0.004 -0.0527 0.004
Budget emphasis (E) c2 0.0704 0.231 -0.0677 0.344
PXB
c3
-0.0003 0.944
Rz 0.295 0.295
F value 10.24 6.692
PC 0.001 0.001
J? explained by nvoway interaction terms = 0%
Hc,a may be rejected in favour of the two alter-
native hypotheses HMl and Hub. _
The results in Table 6 also indicate that in
high task difficulty situations, only the coeffi-
cient for participation was positive and signifi-
cant (P
=Z 0.004, equation (4)). Budget
emphasis, however, had no significant main
effect (p d 0.231 in equation (4)) nor interac-
tive effect (p S 0.944 in equation (5)). These
results support hypothesis HAa,-, that participa-
tion is important in high task difficulty
situations regardless of budget emphasis.
There are, however, important differences
between our results and those of Brownell &
Dunk (1991), particularly in low task difficulty
situations. First, the interaction between parti-
cipation and budget emphasis resulted in a 20%
increase (0.276-0.073) in Ra in the Brownell &
Dunk study. The corresponding increase in R*
in our study was only 8% (0.271-0.191) (Table
5). Second, Brownell & Dunk (p. 701) found no
sign&ant main effect (p s 0.406) with budget
emphasis in the low task difficulty situation. By
contrast, our results indicate a significant main
effect for budget emphasis (p G 0.003 in equa-
tion (4), Table 5). The Ra of the additive model
(equation (4)) was a substantial 19% (more
than twice that of the interaction effect), most
of which was accounted for by budget empha-
sis since participation was not significant (p S
RELIANCE ON ACCOUNTING PERFORMANCE MEASURES
373
TABLE 7. CeII frequencies, mean and standard deviation for
managerial performance across high and low budget
participation and budget emphasis - low task difficulty
subsample
Budget emphasis
Participation
Low
High
LOW
=eh
n = 12 n = 10
7 = 5.58 7 = 6.00
o,, = 0.67 a, = 0.67
n=9 n = 29
p = 5.44 9 = 6.17
0, = 0.73 0, = 0.66
TABLE 8. CeII frequencies, mean and standard deviation for
managerhI performance across high and low budget
participation and budget emphasis - high task diflicuky
subsample
Budget emphasis
Participation
Low
High
LOW
ma
n = 13 n=9
9 = 4.85 7 = 5.56
a, = 0.90 0, = 0.53
n=9 12 = 21
ji = 5.78 7 = 6.14
o, = 0.83 o,, = 0.85
0.468). The combination of (i) a signiticant
interaction effect between budget emphasis
and participation on performance in low task
difficulty situations (Brownell & Dunk, 1991);
and (ii) a positive cultural effect on subordin-
ates’ reaction to budget emphasis in Singapore
(Harrison, 1992, 1993) may have augmented
the effect of budget emphasis on performance
in Singapore, resulting in a significant main
effect not found by Brownell 81 Dunk (1991)
in Australia. These results support Harrison’s
(1992, 1993) theory and findings that culture
has a significant interaction effect with budget
emphasis. Overall, our results in low task dif&
cuhy situations therefore suggest that manage-
rial performance in Singapore was driven by
both budget emphasis and the interaction of
budget emphasis and participation. By con-
trast, Brownell & Dunk (1991) found that
performance in Australia was driven mainly
by the interaction of budget emphasis and
participation.
To assist the interpretation of the results of
equation (5) in the low and high task difficulty
subsamples presented in Tables 5 and 6,
respectively, the levels of both participation
Graph (a) Low task difficulty Graph cb) High task difficulty
Per
6.0
5.5
5.0
f
5
high budget
6.17
5.58 -
low -
5.44
budget emphasis
I I
I I
P&
6.0
5.5
5.0
6.14
I I
LOW
=sh
Participation
LOW
High
Participation
Fig. 1. Three-way interaction graphs
374 CHONG M
and budget emphasis were dichotomized at the
mean. The results (based on the raw scores) are
reported in Tables 7 and 8, and illustrated
graphically in Fig. 1.
Note that for the low task difficulty sub-
sample in Table 7, the mean performance
score is higher in the low budget emphasis
and low participation cell (5.58) than in the
low budget emphasis and high participation
cell (5.44). Similarly, the mean performance
score is higher in the high budget emphasis
and high participation cell (6.17) than in the
high budget emphasis and low participation
cell (6.00). These results are consistent with
hypothesis HAza that, in low task difliculty situa-
tions, compatible combinations of high (low)
budget emphasis and high (low) participation
are associated with improved managerial
performance.
However, the results in Table 7 and Fig. l(a)
also indicate that the mean performance of a
high budget emphasis and low participation
combination (6.00) is higher than that of a
low budget emphasis and low participation
combination (5.58). These results are different
from those of Brownell & Dunk (1991) as they
had observed reverse results. Taken together
with the result that high budget emphasis and
high participation (6.17) is superior to low bud-
get emphasis and high participation (5.44), the
results suggest that, regardless of participation,
a high budget emphasis is always superior to a
low budget emphasis in terms of its impact on
managerial performance in low task difficulty
environments. To test for this possibility, the
point of intersection of the regression lines
for (i) participation and performance for high
budget emphasis and (ii) participation and per-
formance for the low budget emphasis sub-
sample was computed (Kerlinger & Pedhazur,
1973, p. 247). This point represents the level of
participation at which the level of performance
for high budget emphasis is equal to that of low
budget emphasis. The point of intersection for
IAU et al.
the participation score is 19 which is within
the actual observed range of 6 and 42, the
research range of interest. This interaction is
regarded as disordina19 indicating that high
budget emphasis is superior to low budget
emphasis (in terms of performance) for partici-
pation scores of more than 19, and low budget
emphasis is superior to high budget emphasis
(in terms of performance) for participation
scores of less than 19. Only 4 respondents
(7%) in the low task difficulty subsample had
participation scores of 19 or less. Conse-
quently, even though we cannot conclude
that a high budget emphasis is always superior
to low budget emphasis for all levels of partici-
pation, our results indicate that a high budget
emphasis is superior to a low budget emphasis
for a large majority of the Singaporean
managers.
To explore further the levels of participation
at which performance can be significantly
attributed to the level of budget emphasis,
the range of participation within which no sig-
nificant differences in the level of performance
could be attributed to the level of budget
emphasis was computed using the Johnson-
Neyman technique (Kerlinger & Pedhazur,
1973, p. 256). The range of non-significance
(alpha = 0.05) was between 16.6 and 32.6.
This means that for participation scores in
excess of 32.6, high budget emphasis is statis-
tically superior to low budget emphasis in
terms of performance. For participation scores
of less than 16.6, low budget emphasis is statis-
tically superior to high budget emphasis.
Within the range between 16.6 and 32.6,
neither level of budget emphasis is superior.
Since the mean level of participation for the
low task difficulty subsample (31) is within
this range, we can conclude that a high budget
emphasis is not always superior to low budget
emphasis at all level of participation.
However, as the mean of 31 is near the
upper @vel of the non-significance zone, our
’ The results of the test suggested by Schoonhoveu (1981> also indicate tbat budget emphash has a nonmonotonic effect
on managerial pexfomance as the point of infIexion of 19 for tbe raw pattic@ation score lies within the observed range of
participation for the low task dit%culty subsample.
RELIANCE ON ACCOUNTING PERPORMANCE MEASURES 375
Fig. 2.
6.5
6.0 Y6.77994.CXM2K
Y=4.8629+0.0379K
6
19 31 42
mean
Kuticipaticm
Results of Johnson-Neyman test - low task di!iiculty subsamples. The shaded area represents the non-signikancc
7.one, Pi i s the raw score.
result also suggests that for many respondents,
performance is influenced mainly by high bud-
get emphasis, rather than by the intemction of
high budget emphasis and participation. A
further analysis of our low task difficulty sub
sample indicates that there were 4 respondents
with participation scores of 16 or less; 24 with
participation scores of between 17 and 32; and
32 with parddpation score in excess of 33.
This also suggests that for a large majority of
our respondents ln our low task difficulty sub
sample, a high budget emphasis is either (a)
associated with improved performance (53%);
or (b) not detrimental to performance (40%).
Only for a small minority (7%) is a low budget
emphasis associated with improved perfor-
mance. This may explain our results in Table
7 which indicate that a high budget emphasis
and a low participation combination (6.0) is
better than a low budget emphasis and a low
participation combination (5.58). In this
regard, our results are di&rent from those of
Brownell & Dunk (1991, p. 700) who found
that a low budget emphasis and low partidpa-
tion combination (5.53) is superior to a
high budget emphasis and low participation
combination (5.0).
Table 7 also indicates that the number of
respondents in the high budget emphasis cate-
gories (39) is also substantially more than those
in the low budget emphasis categories (21).
Together with our finding that participation
does have a moderating effect on the relation-
ship between budget emphasis and perfor-
mance, our results suggest that in high power
distance and low individualism societies, the
best strategy in low task d@iculty envhm-
men& is a bigb budget emphasis and bigb
particiftfon combfnation. This is substan-
tiated by the large numbef of respondents in
this category (29) which is nearly three times
that of each of the other three categories.
These results are consistent with Browneli h
Dunk’s (1991) finding that a high budget
emphasis and a high participation is a pre-
ferred combination. Although Brownell &
Dunk also found a low budget emphasis and
low participation combination to be desirable,
376 CHONG M
this combination is not popular in Singapore
because the Singaporean culture favours high
budget emphasis (Harrison, 1992, 1993). The
incompatibility of low budget emphasis with
the Singaporean culture is probably the reason
for the low number of respondents in the low
budget emphasis and low participation cate-
gory (12), even though this combination may
in some cases be associated with improved
performance.
For high task difficulty, an inspection of the
cells in Table 8 and Fig. l(b) reveals that parti-
cipation is important regardless of the level of
budget emphasis. With low budget emphasis,
the high participation and low budget empha-
sis cell mean (5.78) is higher than that of the
low participation and low budget emphasis celI
(4.85). With high budget emphasis, the high
participation and high budget emphasis cell
mean (6.14) is higher than that of the low par-
ticipation and high budget emphasis cell
(5.56). The argument for hypothesis HAZb is
that when task difficulty is high, participation
is associated with improved performance
regardless of whether it is matched with bud-
get emphasis or not. These results are consis-
tent with the theory that participation exerts
most of the effect on performance in high
task difficulty (Brownell & Dunk, 1991) and
culture does not inlluence subordinates’ reac-
tions to participation (Harrison, 1992, 1993).”
Our results in Table 8 and Fig. l(b) show that
for both low and high levels of participation,
high budget emphasis is superior to low budget
emphasis (5.56 vs 4.85, and 6.14 vs 5.78).
These results are consistent with Harrison’s
(1992, 1993) theory that Singaporean man-
agers react more favourably to a high budget
emphasis than a low budget emphasis. How-
ever, since Table 6 shows that budget empha-
sis has no significant main effect (p G 0.231;
equation (4)), nor significant interactive effect
(p d 0.944; equation (5)), the positive effect of
budget emphasis on performance in a high task
LAU et al.
difficulty situation is therefore not statistically
significant. Moreover, from the results in Table
8 and Fig. l(b), it is not possible to conclude
that a high budget emphasis always results in
superior performance than does a low budget
emphasis, as a low budget and high participa-
tion combination (5.78) is still superior to a
high budget emphasis and low participation
combination (5.56). These results are consis-
tent with hypothesis HAZb that regardless of
the types of supervisory evaluative style, high
budgetary participation is associated with
improved managerial per$ormance in high
task difficulty situations. These results suggest
that the positive impact of culture on budget
emphasis in Singapore (Harrison, 1992, 1993)
is not sufficient to influence the results of
Brownell & Dunk (1991) in high task difficulty
situations.
CONCLUSIONS AND LIMITATIONS
The replication in another culture of
research performed in one culture is important
for an understanding of the impact of culture
(Alder, 1984; Frucot & Shearon, 1991; Harrison,
1992, 1993; Otley et al., 1994). With regard to
managerial performance, our results confirm
that (i) high budget emphasis and high budget-
ary participation in low task df@x.&y situa-
tions and (ii) high budgetary participation in
high task dificulty situations, regardless of
budget emphasis, are both associated with
improved managerial performance. These
results are consistent with, and hence rein-
force, those of Brownell & Dunk (1991). The
sign&ant results on managerial performance
associated with task drfficulty and not with
task uncertainty also provide evidence to
substantiate BrownelI & Dunk’s (1991) sugges
tion that task dfffi cul ty is the more appro-
priate choice of instrument to measure task
characteristics.
lo To determine if our results were tienced by the moderate level of task difiiculty in our sample, we also dichotomized
our sample based on the task dif6culty theoretical mean of 28. This resulted in 80 cases for low task difticulty and 32 cases
for high task difiiculty. There were no differences in the general patterns of our results.
RELIANCE ON ACCOUNTING PERFORMANCE MEASURES 377
However, our results also differ from those
of Brownell & Dunk (1991). First, our results
indicate that a low budget emphasis and low
participation combination is not superior to a
high budget emphasis and low participation
combination in low task difficulty situations.
Second, in a low task difficulty environment,
budget emphasis has a highly signiticant main
effect, which is substantially more than that of
the interaction between budget emphasis and
participation. Differences in our results and
those of Brownell & Dunk (1991) appear to
be caused not by subordinates’ reactions to
budgetary participation, but by subordinates’
positive reactions to budget emphasis. The
extent and the direction of the differences in
the results are consistent with Harrison’s
theory on the influence of national culture on
budget emphasis and participation. Hence our
results provide additional evidence to substanti-
ate Harrison’s (1992, 1993) findings on the
impact of national culture. Since our results
are based on a sample of manufacturing com-
panies, incorporating managers from various
functional areas, they also suggest the possi-
bility that Harrison’s (1992, 1993) conclusions
can be generalized to a broader range of indus-
tries and functional areas. More importantly,
these confirmations provide additional evi-
dence in support of Harrison’s research frame-
work for examining the impact of culture on
other aspects of management accounting sys-
tems. Further research work along this line,
not only in the budgetary area, but also on
other aspects of managerial accounting, would
enable Asian businesses to adopt Western man-
agerial accounting systems with greater confi-
dence and, when and if appropriate, in a more
discriminatory manner.
Further research using more systematic tests
of the impact of culture on the relation
between budgetary participation and manage-
rial performance would also be worthwhile.
For example, the use of a pooled sample
drawn from different cultural blocks, together
with tests on the effects of budget emphasis,
participation, task difficulty and culture affect-
ing managerial performance, may provide
further evidence to evaluate Harrison’s conclu-
sions. Other dependent variables, such as job
satisfaction and other job-related attitudes and
budget-related attitudes, could also be studied.
Moreover, both our and Harrison’s (1992,
1993) studies have been conducted in low
power distance and high individualism, and/
or high power distance and low individualism
blocks. Similar studies on nations in the other
two cultural blocks characterized by high
power distance and high individualism, and
low power distance and low individualism
would provide a more rigorous test of
Harrison’s (1992, 1993) framework. Finally,
our failure to replicate the results of Brownelf
& Hirst (1986) in Singapore suggests that the
relationship between budget emphasis,
national culture and job-related tension may
be a complex one. Further research which con-
siders the effect of the cuhural dimension of
uncertainty avoidance may be worthwhile.
Although the results of this study provide
some support for those of Brownell & Dunk
(1991) and Harrison (1992, 1993) generalizing
the effect of culture on the research work on
budgetary participation and budget emphasis
should be viewed with caution, First, as with
most studies on supervisory evaluative styles
based on questionnaire surveys, our study has
the usual limitations associated with that
methodology. Hidden biases and random
errors can also inflate associations (see Bagozzi
et d, 1991, p. 424). Second, comparisons of
research results based on research work done
in different time periods may be distorted by
inter-temporal changes. Third, the responses
we observed for the task difficulty measure
were limited to the moderate range, restricting
the generalizability of our results to organiza-
tions involving similar task difficulty situa-
tions. Fourth, as our sample was selected
from manufacturing companies, caution
should be exercised in generalizing our results
to other industries. Further research involving
the service industries and the public sector
would be worthwhile. Budget emphasis is
likely to remain high in the public sector
because of the appropriation process and the
378 CHONG M. JAU et al.
resultant accountability requirements imposed
on managers.
As the focus of our paper is to generalize the
results of BrowneII & Hirst (1986) and BrownelI
& Dunk (1991) to Singapore, the choice of
measurement instruments was problematic.
We have tried to use the same measurement
instruments as those in BrowneIl & Hirst
(1986) and BrowneIl and Dunk (1991) so that
the differences between our results and those
of the two prior studies could be attributed to
cuhurai difference. This means that some mea-
sures selected may have inherent Limitations.
With regard to budget emphasis, Kren & Liao
(1988) pointed out several methodological
problems including the iimitations associated
with the use of self-reported values; the use
of different instruments in different studies;
and the use of different operational defini-
tions. Some researchers (e.g. BrowneIl, 1985;
Briers h Hirst, 1990; Manzoni, 1992) have
also expressed a concern on the need to cap
ture not only the etient but also the munner in
which accounting data are used by superiors in
performance evaluation to overcome the pro-
blem of construct vaiidity. With regard to man-
agerial performance, it should be noted that
the use of overall job performance has been
regarded as problematic (Briers & Hirst, 1990;
Hirst & Lowy, 1990). Concerns have included
(i) whether budgetary performance, rather
than overall job performance, may be the
more appropriate dependent variable @Iirst &
Lowy, 1990); (ii) whether the relationship
between supervisory styles and job perfor-
mance may only be a spurious one (Otley,
1978); (iii) whether job performance should
be treated as an independent variable or a
dependent variable (BrownelI, 1982a); and
(iv) whether the use of the &f-rating scale is
appropriate (Govindarajan, 1984).
With regard to task diflicuhy, our study and
some of the other accounting studies (e.g. Mia,
1987; Browneli & Dunk, 1991) using the Van
de Ven & Delbecq (1974) instrument have
reported low alpha Cronbach scores. It is per-
haps worth noting that Van de Ven & Delbecq
(1974) in their original study on the develop
ment and testing of their task difIicuity instru-
ment report an alpha coefficient of 0.86. It
seems that the use of this scale in a different
situation has signi6cantIy lowered its reliability.
As the task difficulty instrument has a history of
low Cronbach alpha values in cross-sectional
studies, further use of this instrument in future
research should proceed with caution. Further
research with other existing task difiicuhy
instruments (e.g. Van de Ven 8r Ferry, 1980;
Daft & Macintosh, 1981) or the development
of a new instrument with improved construct
validity (Churchih, 1979; Peter, 1979) is
needed. The incorporation of alternative mea-
sures of task uncertainty and task difficulty
would also assist in determining the validity
of Browneil & Dunk’s (1991) explanation on
the failure of Brownell 8r Hirst (1986) to con-
firm their hypotheses relating to manageriai
performance.
Nevertheless, despite the limitations listed
above, this study has provided additional evi-
dence on the area of superior’s evaluative
styles. Together with those of others (e.g.
Hwang, 1989; Chow et al., 1991, 1994;
Harrison, 1992, 1993; Otiey et al, 1994) it
also constitutes an attempt to develop a rela-
tively new area of research. This line of
research has provided some guidelines for sys
tematic evaluations of the effects of national
culture, not only in management accounting,
but also in other areas of accounting research.
BIBLIOGRAPHY
Alder, N. J., Umlcrstanding the Ways of UndcrstaadinR: Cnxs-culhml Manqtcment McthodoloRy
Rcvkwcd, in Fmncr, R. N. (cd.), Advances In Zntemationd Compa?raHue Management, pp. 31-67
(Greenwich CT: JAI Press, 1984).
RELIANCE ON ACCOUNTING PERFORMANCE MEASURES 379
AI&on, P. D., Testing for Interaction in Multiple Regression, Amerfcun J ournal of Sociology (1977)
pp. 144-153.
Bagoazl, R. P., Youjae, Y. % Phillips, L. W., Assessing Construct Validity in Orgamrational Research,
AdmWstrative Sckttce Quarterly (1991) pp. 421-458.
Brlers, M. L. t Hhst, M. K., The Role of Budgetary Information in Performance Evaluation, Accounting,
Orgunizutions and Society (1990) pp. 373-398.
Brownell, P., The Role of Accounting Data in Performance Evaluation, Budgetary Participation, and
Organlsational EITectiveness, J ournal ofAccountfng Research (Spring 1982a) pp. 12-27.
Brownell, P., Participation in the Budgeting Process: When it Works and it Doesn’t, J ournal of Account-
fng Literature (1982b) pp. 124-153.
Brownell, P., Budgetary Systems and the Control of Functionally Differentiated Organnational Activities,
J ournal of Accounting Research (Autumn 1985) pp. 502-512.
BrownelI, P. & Dunk, A. S., Task Uncertainty and its Interaction with Budgetary Participation and Budget
Emphasis: Some Methodological Issues and Empirical Investigation, Accounting, Organfzutfons and
Sockty (1991) pp. 693-703.
Brownell, P. % Hlrst, M K., Reliance on Accounting Information, Budgetary Participation, and Task
Uncertainty: Tests of a Three-way Interaction, J ournal of Accounting Reseurcb (Autumn 1986)
pp. 241-249.
ChenhaIl, R. H. % BrownelI, P., The Fffect of Participative Budgeting on Job Satisfaction and Perfor-
mance: Role Ambiguity as an Intervening Variable, Accounting, Organizations and Socfev (1988)
pp. 225-233.
Cherrington, D. J. & Cherrington, J. O., Appropriate Reinforcement Countingencies in the Budgeting
Process, J ournal of Accountfng Reseanb (1973) pp. 225-253 (SuppI.).
Churchill, G. A. Jr, A Paradlgm for Developing Better Measures of Marketing Constructs, J ournul of
Murketfng Reseurcb (February 1979) pp. 64-73.
Chow, C. W., Kato, Y. & Shield, M. D., National Culture and the Preference for Management Controls: an
Exploratory Study of the Finn-Iabour Market Interface, Accountfng, Orgumhztfons and Society
(1994) pp. 381-400.
Chow, C. W., Shields, M. D. & Chan, Y. K., The Effects of Management Controls and National Culture on
Manufacturing Performance: an Experimental Investigation, Accounting, Organizations and Society
(1991) pp. 209-226.
Cohen, J. & Cohen, P., Applied Multipk Regresston/Cowelation Analysis for tbe Behavioral Sciences,
2nd Edn (London: Lawrence Erlbaum Associates, 1983).
Cronbach, L. J., Coefficient AIpha and the Internal Structure of Tests, PsycbomeMka (September 1951)
pp. 297-334.
Cronbach, L. J., Statistical Tests for Moderator Variables: Flaws in Analysis Recently Proposed, Pry&c+
logical Bulierin (1987) pp. 414-417.
Daft, R. L. 8 Macintosh, N. B., A Tentative Exploration Into Amount and Equlvocallty of Information
Processing in Otganiaational Work Units, AdmjnnLsharfve Sclmce Quarterly (1981) pp. 207-224.
Dunk, A. S., Budget Emphasis, Budgetaty Partlclpatlon and ManagerlaI Performance: a Note, Accounting,
Organizations and Society (1989) pp. 321-324.
Early, P. C., Wojnarosli, P. % priest, W., Task PMnnIng and Energy Expended: Exploration of How Goals
IntIuence Performance, J ournal of Applied PsycboZogy (1987) pp. 107-114.
FNCOt , V. % Shearon, W. T., Budgetary Partlcipatlon, Locus of Control, and Mexican Managerial Perfor-
mance and Job Satisfaction, IIre Accounttng Revkw (J anuaq 1991) pp. 80-99.
GEsson, C. A., Dependence of TechnologicaI Routlnbation on Structure Vatiables in Human Service
Organizations, Administrative Science Quurterfy (1978) pp. 383-395.
Govindarajan, V., Approprlateness of Accounting Data in Performance Evaluation: an Empirical ExamIna-
uon of FnvhonmentaI UncettaInty as an Intervening Varlable, Accountfng, Organfzuttons and Society
(1984) pp. 125-135.
Hage, J. T. % Alken, M., Routine Technology, Social Structure and Organiaational Goals, Adminfstratiue
scknce Quurte@y (1969) pp. 366-377.
Harrison, G. L., The Cross-cultural GenemIlzxblIlty of the Rehtlon Between Participation, Budget Empha-
sis and Job-related Attitudes, Accounting, OrganLzathns and Society (1992) pp. 1-15.
Harrison, G. L., Reliance on Accounting Performance Measure s In Superior Evaluative Style - the
Influence of NatlonaI Culture and Personality, Accounttng, Orgunizutfons and Sodety (1993)
pp. 319-339.
380 CHONG M. LAU et al.
Heneman, H. G., Comparisons of Self and Superior Ratings of Managerial Performance, J ournal of
Applied pSrcbologv (October 1974) pp. 638-642.
Hirst, M. K., Some Further Evidence on the Effects of Budget Use and Budget Participation on Managerial
Performance, Australian J ournal of Management (1987) pp. 49-56.
Hit, M. K. & Lowy, S. M., The Linear Additive and Interactive Effects of Budgetary Goals Di&uky and
Feedback on Performance, Accounting, Organizations and Society (1990) pp. 425-436.
Hofstede, G. H., Culture’s Consequences: I nkrnational Differences in Work-related Values (Eleverly
Hills: Sage, 1980).
Hofstede, G. H., The Cultural Relativity of Organizational Practices and Theories, J ouwual of I nter-
national Business Studies (Fall 1983) pp. 75-89.
Hofstede, G. H., The Cultural Relativity of the Quality of Life Concepts, Academy ofManagement Review
(1984) pp. 389-398.
Hopwood, A. G., An Empirical Study of the Role of Accounting Data in Performance Evaluation, J ournal
ofAccounting Research (1972) pp. 156-182 (Suppl.).
Hwang, S. K., The Effect of Culture on the Relationship Between Participation and Motivation, Ph.D.
thesis, Macquarie University (1989).
Jaccard, J., Turrisi, R. & Wan, C. K., I nkruction Effects in Multiple Regression (Sage University Press,
1990).
Kahn, R. L., Wolfe, D. M., Quinn, R. P., Snoek, J. D. & Rosenthal, R., Cbgunizutional Stress: Studies in
Role Conflict and Ambiguity (New York: John Wiley, 1%4).
Kerlinger, F. N. & Pedhazur, E. J., Multiple Regression in Behavioral Research (Iiolt, Rinehart and
Winston, 1973).
Kompass: Register of I ndustry and Commerce of Singapore 1992 (Singapore: Kompass South East Asia,
1992).
Kren, L. t Liao, W. M., The Role of Accounting Information in the Control of Organizations: a Review of
the Evidence, J ournal of Accounting Literature (1988) pp. 280-309.
Lindsay, R. M. & Ehrenberg, S. C., The Design of Replicated Studies, 7&e Americun Stutisticiun (August
1993) pp. 217-228.
Lynch, B. P., An Empirical Assessment of Perrow’s Technology Construct, Administrative Science
Quarterly (1974) pp. 338-356.
Mahoney, T. A., Jerdee, T. H. & Carroll, S. J., Development of Managerial Performance: a Research
Approach (Cincionati: South Western Publishing, 1963).
Mahoney, T. A., Jetdee, T. H. & Carroll, S. J., ‘Ihe Jobs of Management, I ndustrial Relarions (February
1965) pp. 97-l 10.
Manzoni, J. F., Use of Quantitative Feedbqck by Superiors of Manufacturing Cost Centre Managers: a
Methodological and Empirical Update, Second European Management Control Symposium auly 1992).
Mia, L., Participation in Budgetary Decision Making, Task DifXculty, Locus of Control and Employee
Behaviour: an Empirical Study, Decision Science (Fall 1987) pp. 547-561.
Mia, L., The Impact of Participation in Budgeting and Job Difficulty on Managerial Performance and Work
Motivation: a Research Note, Accounting, Organizations and Society (1989) pp. 347-357.
Milani, K. W., The Relationship of Participation in Budget-setting to Industial Supervisor Performance
and Attitudes: a Field Study, Z&e Accounting Review (April 1975) pp. 274-284.
Oppenheim, A. N., Questionnaim Design and Attitude Measurement mew York: Basic Books, 1966).
Otley, D. T., Budget Use and Managerial Performance, J ournal of Accounting Research (Spring 1978)
pp. 122-149.
Otley, D. T., Hannakis, L. & Lindsay, R. M., Intluence in Budgeting, Locus of Control and organizational
Effectiveness: Cultural Differences, Accountfng and Business Review Uanuary 1994) pp. 29-42.
Perrow, C., A Framework for the Comparative Analysis of Organisations, Ameriuan Sociologicul Review
(1%7) pp. 194-208.
Perrow, C., Organizational Analysis: a Sociologicul View (Belmont, CA: Wadsworth, 1970).
Peter, P. J., Reliability: a Review of Psychometric Basics at&Recent Marketing Practices, J ournal of
Marketing Research (Febnuuy 1979) pp. 6-17.
Prien, E. P. & Liske, R. E., Assessment of Higher-level Personnel: III: a Comparative Analysis of Supervisor
ratings and Incumbent Sewratings of Job Performance, Personnel Psychology
Based on Harrison’s (Accountfng, Otganfzations and Sod&y, pp. 1-15, 1992; Accounting, Organ&&
tiom and Socfety, pp. 319-339, 1993) 6ndings on the general&ability of research results related to
budgetaty participation and budget emphasis between cultures, two tests on the three-way interaction
between budget emphasis, budgecuy partidpation and task characteristics (task uncertainty and task
diBiculty) atTect.injgo lx&ted tension and managerialp erformance were conducted in Singapore
Acown~ng,Orgnn~aHonrnndSOEIPtY,VoL. 2O,No.5,~p.359-381, 1995
Copyright 0 1995 PJsevier Science Ixd
PrlntcdinGnxt Britain.AUr&htsrcserved
0361~3682/95 $9.50+0.00
‘IWE IMPACT OF RELIANCE ON ACCOUNTING PERFORMANCE MEASURES ON
JOBRELATED TENSION AND MANAGERIAL PERFORMANCE: ADDITIONAL
EVIDENCE*
CHONG M. I&J
Edi th Cowan Uni versi ty
LIANG C. LOW
Si ngapore Pol ytechni c
and
IAN R. C. EGGLETON
The University of Western Austral i a
Based on Harrison’s (Accountfng, Otganfzations and Sod&y, pp. 1-15, 1992; Accounti ng, Organ&&
ti om and Socfety, pp. 319-339, 1993) 6ndings on the general&ability of research results related to
budgetaty participation and budget emphasis between cultures, two tests on the three-way interaction
between budget emphasis, budgecuy partidpation and task characteristics (task uncertainty and task
diBiculty) atTect.ing jolx&ted tension and managerial performance were conducted in Singapore. The
results support a high budget emphasis and high budgctaty participation combination in low task
difticulty skuations; and a high budgetary participation style, regardless of budget emphasis, in high
task dif6cuk.y situations.
The use of accounting data and the manner in
which those data are used by upper level man-
agement (superiors) to evaluate lower level
management (subordinates), and its effects on
subordinates’ job-related attitudes, budget-
related attitudes and performance have been
examined systematically by accounting
researchers commencing w&h the empiricaI
study by Hopwood (1972). while the research
findings of the earlier work of Hopwood (1972)
and OtIey (1978) were contIicting, more recent
work has apparent@ been able to resolve these
conflicting results (Brownell & Hirst, 1986;
BrowneII & Dunk, 1991). There is also evi-
dence to suggest the possibility of generalizing
the results of such research work done in the
AngIoAmerican block to the Asian block
CHwang, 1989; Harrison, 1992, 1993).
The focus of this paper is on the generaliz-
ability of the results of research on budgetary
participation and superiors’ evaluative styles
(specifically the Brownell & Him, 1986, and
Brownell & Dunk, 1991, studies) across
national cultures. We consider Harrison’s
’ The authors gratefully acknowledge the helpful comments of the editor and two anonymous mviewcrs. Special that&s
ucaacn~toAlvlDunLwho?asfsudintheevty~ofthisrrsearch.Anevllerverrsionofthlsplper~prrrnted
at the Account.ing % Finance Wo&shop, The University of Western AustfaLq the Accounting Research Group Seminar
Series, Edith Cowan University; and the 1994 University of New South Wales Management Accounting Research
Codel I Ce.
360 CHONG M. LAU et al.
(1992,1993) suggestion that research results on
budgetary participation are transferable across
the Anglo-American block and the Asian block,
whilst results on reliance on accounting perfor-
mance measures are not transferable. More
importantly, we provide evidence to support
Harrison’s research framework for examining
the Impact of national culture, not only on bud-
getary participation and reliance on accounting
performance measures, but also on other
aspects of the management accounting system.
With many nations in the Asian block experien-
cing fast economic growth, the adoption of
more sophisticated management accounting
systems and practices on a wider basis is
likely. Since a wealth of knowledge has been
accumulated through decades of research with
respect to the Anglo-American block, a systema-
tic approach to segregate those results that
could be generalized across cultures would facil-
itate the rapid adoption of improved manage-
ment accounting systems by Asian businesses.
Additionally, this study provides further
extensions to the research work in the area of
supervisory evaluative styles. Brownell & Dunk
(199 1, p. 703) suggested that the research work
in the area of supervisory evaluative style consti-
tutes an important “. . . organized critical mass
of empirical work in management accounting”.
Lindsay & Ehrenberg (1993, p. 223) considered
this area “one of the relatively few areas in man-
agement accounting research where there has
been any sequence of repeated studies”. From
the work of Hopwood (1972) to Harrison
(1993) management accounting researchers
have sought to study the relations between
superiors’ evaluative styles and subordinates’
attitudes and performance. Comprehensive
reviews of research work in this area have
been undertaken by Kren & Liao (1988) and
Briers & Hlrst (1990). Further research should
contribute to a better comprehension of these
relationships. Such understanding has impor-
tant implications for the design and implementa-
tion of management accounting control systems
as it may lead to changes in superiors’ budgetary
and evaluative styles to bring about desired
subordinates’ attitudes and job performance.
HYPOTHESIS DEVELOPMENT
Brownell & Hirst (1986) attempted to resolve
the conflict between Hopwood (1972) and
Otley (1978) by examining the three-way inter-
action between budget emphasis, budgetary
participation and task uncertainty. They sug-
gested that Brownell’s (1982a) finding that
compatible combinations of high (low) budget
emphasis and high (low) budgetary participa-
tion are associated with lower job-related ten-
sion and improved managerial performance
would hold in low task uncertainty situations,
but not in high task uncertainty situations. A
reward structure based primarily on budget
achievement is likely to motivate subordinates
only if they are heavily involved in the budget
setting process (Cherrington & Cherrlngton,
1973; Brownell, 1982a). If budget emphasis is
high, realistic and attainable budgetary goals
are crucial to subordinates because of the link
between budgetary goal achievement and
rewards. High budgetary participation is thus
necessary to ensure that the budgetary goals
set are attainable. In low task uncertainty envi-
ronments, complete task cause-effect knowl-
edge also facilitates the development of
unambiguous budgetary goals. Prior research
has found goal clarity to be an Important attri-
bute that impacts on performance (Early et al.,
1987). Consequently, high participation in a
high budget emphasis and low task uncertainty
situation is unlikely to be stressful to subordin-
ates as they are likely to agree with the choice
of evaluation criteria. Participation may
improve performance in a number of ways.
Kren & Llao (1988) suggested that participa-
tion may affect performance by acting
“through expectancies thus improving motiva-
tion; . . . (by providing) intrinsic valences for
performance, especially if it fosters group cohe-
siveness; . . . through the setting of higher bud-
get goals; by increasing commitment to budget
goals; and through the cognitive benefits
derived from sharing information during
participation” (p. 289).
In contrast, in low budget emphasis and low
task uncertainty situations, subordinates are
RELIANCE ON ACCOUNTING PERFORMANCE MEASURES 361
unlikely to be concerned with the accuracy of
budgetary goals because of the absence of a
link between budgetary goal achievement and
the reward system. In such situations, imposed
budgets may be more appropriate than partici-
pative budgets (Cherrington & Cherrington,
1973; Brownell, 1982a) because, with com-
plete task knowledge, there is no necessity
for subordinates to seek additional task knowl-
edge through participation. Superiors are also
unlikely to benefit from subordinates’ participa-
tion as information asymmetry is low. Participa-
tion therefore serves no useful purpose and
may be counterproductive, as subordinates
may be confused as to why they must waste
time participating in activities with seemingly
obvious solutions (Kren & Liao, 1988). Low
task uncertainty situations are likely to occur
in industries utilizing routine technologies for
the production of mature products, particularly
in developing Asian nations where labour
intensive industries are still fairly common.
In high task uncertainty environments,
budgetary participation will be beneficial
regardless of budget emphasis because “parti-
cipation may provide the opportunity for man-
agers to gain access to resources which can be
used to buffer task performance from the unan-
ticipated effects of others, and to introduce
new and better means for addressing tasks
which, if highly uncertain, will have character-
istics that change over time” (Brownell & Hirst,
1986, p. 242). It is also plausible that participa-
tion may enable superiors to gain input from
subordinates to improve operating techniques
(Kren & Liao, 1988). High budgetary participa-
tion therefore becomes the primary considera-
tion in high task uncertainty environments.
Brownell & Hirst’s (1986) results supported a
significant three-way interaction between bud-
get emphasis, budgetary participation and task
uncertainty affecting job-related tension, but
not managerial performance. Brownell &
Dunk (1991) attempted to resolve the failure
of Brownell & Hirst (1986) to support their
hypotheses with respect to managerial perfor-
mance. Task uncertainty comprises the subdi-
mensions of task difficulty (analysability) and
task variability (number of exceptions) (Van
de Ven & Delbecq, 1974). Of these two subdi-
mensions, task difhculty is the more appropri-
ate subdimension to measure to replicate the
findings of Brownell (1982a). When task dif&
culty is high, i.e. when the input and output
relations of the tasks are difficult to specify or
unclear, there is a greater need for budgetary
participation to counter the incomplete task
cause-effect knowledge. Task variability, on
the other hand, refers to the variability of sub
tasks. High task variability may arise because of
many different subtasks, but if each one is
easily analysable in terms of input and output
relations, the task cause-effect knowledge may
be complete (Perrow, 1967, 1970; BrownelI &
Dunk, 1991). With prescribed rules, policies
and procedures for performing the job
(Tushman 8~ Nadler, 1978) there would be
less need for participation. Brownell’s (1982a)
results are therefore expected to apply only to
situations that differ with respect to task diffi-
culty, not task variability. The results of
Brownell & Dunk (1991) supported a three-
way interactive and positive effect between
budget emphasis, participation and task diffi-
culty on performance but, as expected, only
in low task difficulty situations. There were
no significant interactive effects on perfor-
mance when tasks differed along the variability
continuum.
Hwang (1989) and Harrison (1992) extended
this body of research by studying the cross-
cultural generalizability of research results relat-
ing to budgetary participation. Hwang (1989)
examined the moderating effect of power dis-
tance on the relationship between budgetary
participation and motivation. Harrison (1992,
1993) examined the moderating effect of
national culture on the relationship between
budget emphasis and both job-related tension
and job satisfaction.
Harrison’s (1992) research framework was
based on two of Hofstede’s (1980) dimensions
of culture - power distance and individualism
-
which were hypothesized to be related to
budgetary participation. Hofstede (1980) sug-
gested that participation of the lower-placed
362 CHONG M. LAU et al.
groups in important affairs is not encouraged in
high power distance societies because such
societies elaborate, institutionalize and forma-
lize dominance behaviour. Budgetary participa-
tion is therefore likely to be “pseudo” and
unlikely to be beneficial. Low power distance
societies regard humans as equal and attempt
to minimize hierarchies. In such societies, the
participation of all individuals in affairs that
affect everybody is favoured. Budgetary partici-
pation is likely to be genuine and beneficial.
High individualism societies regard each indivi-
dual, and not the group, as the basic unit of
society. Since the focus is on the individual,
decision-making on an individual (superior)
basis is likely to be the norm. Budgetary parti-
cipation is unlikely to be favoured. In contrast,
low individualism societies regard the group to
which the individual belongs as the basic unit
of society. In such collective societies, partici-
pation of all forms, including budgetary partici-
pation, is favoured and likely to be beneficial. It
would seem that budgetary participation
should be encouraged in low power distance
and low individualism societies and discour-
aged in high power distance and high individu-
alism societies. However, Hofstede’s (1983)
study of 50 nations shows that a large majority
(47) have either a low power distance and high
individualism culture (which includes the
Anglo-American block) or a high power dis-
tance and low individualism culture (which
includes the Asian block). Within the Anglo-
American block, the low power distance
which favours budgetary participation is likely
to be counterbalanced by the high individual-
ism which does not favour participation. These
opposing factors could effectively neutralize
the cultural effects on subordinates’ reactions
to budgetary participation. Similarly, the
opposing effects on participation of the two
cultural dimensions within the Asian block cul-
ture could also neutralize their effects on sub
ordinates’ reactions to budgetary participation.
Thus, Harrison (1992) hypothesized that
research results on budgetary participation
obtained in the Anglo-American block could
be generalized to the Asian block. Harrison’s
(1992) results not only reaffirmed Hofstede’s
(1983) classification of Singapore as a nation
of high power distance and low individualism
and Australia as a low power distance and high
individualism nation, they also indicated no sig-
nificant two-way interaction between nation
and participation. This suggests that the effects
of participation on the relation between budget
emphasis and jobrelated tension on one hand,
and job satisfaction on the other, is not mod-
erated by differences in the Australian and
Singaporean cultures.
With regard to budget emphasis, Harrison
(1993) suggested that high power distance
societies are likely to prefer a high budget
emphasis evaluative style because a “prefer-
ence for a non-consultative, decisive leader-
ship style in high PD (power distance)
societies is also likely to produce a preference
for high RAPM (reliance on accounting perfor-
mance measures) in evaluative style. This style
accords with the decisive leadership character-
istic of coming to a decision and sticking to it”
(p. 322). Low individualism societies are also
likely to accept a high budget emphasis evalua-
tive style (which assumes that people can be
compared) because low individualism de-
emphasizes individual differences which
therefore facilitates comparability among indi-
viduals. Thus, high power distance and low
individualism Asian nations are likely to react
favourably to a high budget emphasis evaluative
style.
By contrast, subordinates in low power dis
tance societies will not react favourably to a
high budget emphasis evaluative style because
it “effectively blocks the subordinate from
representation in the evaluation process and
thereby frustrates the realization of equality in
the subordinate’s relations with his/her
superior” (Harrison, 1993, p. 322). Similarly,
subordinates in high individualism societies
are also likely to react unfavourably to a high
budget evaluative style because the assumption
of comparability among individuals with high
budget emphasis is inconsistent with the
emphasis on the uniqueness of individuals in
these societies. Consequently, low power
RELIANCE ON ACCOUNTING PERFORMANCE MEASURES 363
distance and high individualism societies such
as those in the Anglo-American nations are likely
to prefer a low budget emphasis evaluative
style.
Harrison (1992) therefore concluded that
“clear and different directional expectations
could be projected for the effect of culture
on the relation between budget emphasis and
the dependent variables in Singapore and
Australia. . . (but) no such directional expecta-
tion could be projected for the effect of culture
on the relation between participation and the
dependent variables” (p. 12).
Brownell & Hirst (1986) and BrownelI &
Dunk (1991) have provided plausible explana-
tions for the conflict between Hopwood (1972)
and Otley (1978) relating to job-related tension
and managerial performance, respectively.
Hwang (1989) and Harrison (1992, 1993)
have provided the theoretical justification to
extend these results for Anglo-American
nations to Asian nations. However, a number
of issues merit further investigation. First,
results of prior empirical studies on the posi-
tive effect of participation, particularly on man-
agerial performance, are Inconsistent and
therefore lack broad generalizability (Brownell,
1982b; Brownell & Hirst, 1986; Hirst, 1987;
Kren & Liao, 1988; Brlers & Hirst, 1990;
Lindsay & Ehrenberg, 1993). Kren & Liao
(1988) suggested that “existing evidence is still
limited and additional studies in different set-
tings and using different subject samples
would be very beneficial” (p. 298). Lindsay
& Ehrenberg (1993) concluded that “the
results add up to little by way of robust,
generalizable findings” @. 224). Otley et al.,
(1994) suggested that “replication is a much
neglected aspect of accounting research. The
value of repeating previous work is that it
allows a better assessment of the confidence
that should be placed in specific findings . . .”
@. 30). It would also reduce the chance of
accepting false hypotheses and results that are
“a total flash in the pan” (Lindsay & Ehrenberg,
1993, p. 220). Thus there is a need to cor-
roborate and solidify the results of earlier
studies and to establish the range of different
conditions (e.g. different organizations and dif-
ferent cultures) under which these findings will
hold. Second, Hwang’s (1989) study focused
primariIy on participation and motivation.
Harrison’s (1992, 1993) emphasis was on job
related tension and job satisfaction. Examining
managerial performance systematically may
support or reject his suggestion (Harrison,
1992) that the results of BrownelI (1982a)
could be generalized to high power distance
and low individualism nations. Furthermore, a
systematic examination of a three-way interac-
tion between budget emphasis, participation
and task uncertainty consistent with Brownell
& Hirst (1986) has not been undertaken in a
high power distance and low individualism
nation such as Singapore. Although there was
matching of task in Harrison’s (1992) study, the
effects of task uncertainty were not systematic-
aliy examined. Finally, the effects of budgetary
participation on the relation between budget
emphasis and managerial performance in high
task difficulty situations is also untested in high
power distance/low individualism nations.
Given these gaps in the current literature and
the significant results pertaining to job-related
tension and task uncertakry in BrownelI &
Hirst (1986); and managerial performance and
tusk di fJ i cul ty i n BrownelI & Dunk (1991), this
study examines the moderating effect of(i) task
uncertai nty and participation on the relation
between budget emphasis and j ob-rel ated ten-
si on and (ii) task d@i cuZty and participation on
the relation between budget emphasis and
manageri al performance using Singapore as
the test site.
The following two sets of null and related
alternative hypotheses are tested:
Set 1: Task uncertai nty and j ob-rel ated
tensi on
I & There i s no stati sti cal l y sigoificant three-way
interaction between budget emphasis, budget-
ary partidpation and task uncwkzinry affecting
j ob-Mated tensi on.
HA,, Compatible combinations of high sow) budget
emphasis and bigb (low) budgw participation
364 CHONG M. LAU et al.
are associated with lowerjob-rekated tension in
low task uncertainly situations.
Regardless of the types of supervisory evaluative
style, high budgetary participation is associated
with lower j ob-rel ated tensi on i n hi gh task
uncertainty situations.
Set 2: Task difficulty and managerial
performance
H
**a
HAab
There is no statistically sign&ant three-way
interaction between budget emphasis, budget-
ary participation and task dtQ@zr&y affecting
manageri al perfownance.
Compatible combinations of high (low) budget
emphasis and high (low) budgetary participation
are associated with improved rnunage&zZ per-
formance i n l ow task di ffi cul ty situations.
Regardless of the types of supervisory evaluative
style, high budgetary participation is associated
with improved manageri al performance i n hi gh
task d@i cul ty situations.
As the cultural differences between Singapore
and Australia are expected to affect only sub
ordinates’ reactions to budget emphasis, but
not subordinates’ reactions to participation
(Harrison, 1992, 1993) the extent to which
the results of Brownell & Hirst (1986) and
Brownell & Dunk (1991) will be supported in
Singapore will depend mainly on the influence
of culture on Singaporean managers’ reactions
to budget emphasis. With a high power
distance and low individualism culture in
Singapore (Hofstede, 1980, 1983; Harrison,
1992, 1993), Singaporean managers are
expected to react favourably to a high budget
emphasis evaluative style. As these reactions
are opposite to those of Australian managers,
our Singapore results are therefore expected to
be different from those of Brownell & Hirst
(1986) and Brownell & Dunk (1991). Although
the extent of these differences are indetermin-
able, directional predictions are possible.
With regard to low task difficulty (uncer-
tainty) situations, both Brownell & Hirst
(1986) and Brownell & Dunk (1991) have
found the high budget emphasis and high parti-
cipation combination to be superior (in terms
of lower job-related tension and improved man-
agerial performance, respectively). This combi-
nation is expected to be a superior combination
in Singapore as well since the high budget
emphasis component is compatible with the
Singaporean culture while the high participa-
tion component is unaffected by the culture.
Thus a high budget emphasis and high partici-
pation combination in low task difficulty
(uncertainty) situations is expected to be asso-
ciated with improved managerial performance
(lower jobrelated tension) in Singapore.
Apart from the high budget emphasis and
high participation combination, both Brownell
& Hirst (1986) and Brownell 81 Dunk (1991)
have also found the low budget emphasis and
low participation combination to be superior in
low task difficulty (uncertainty) situations.
Since a low budget emphasis strategy is incom-
patible with the Singaporean culture, this com-
bination may not be a desirable combination in
Singapore. Hence, unlike in Australia, a low
budget emphasis and low participation combi-
nation in low task difficulty (uncertainty) situa-
tions is not expected to be associated with
improved managerial performance (lower job
related tension) in Singapore.
With regard to high task difficulty (uncer-
tainty) situations, both Brownell & Hirst
(1986) and Brownell & Dunk (1991) have
found a high budgetary participation strategy,
regardless of budget emphasis, to be superior.
Since culture has no effect on participation, a
high budgetary participation strategy, regard-
less of budget emphasis, is also expected to
be a superior strategy in Singapore. In addi-
tion, since the Singaporean culture favours a
high budget emphasis evaluative style, a high
budgetary participation and high budget
emphasis combination is also likely to be a
desirable combination in Singapore.
METHOD
A questionnaire was administered to a sample
of 240 functional heads from 80 Singapore
manufacturing companies with more than 100
RELIANCE ON ACCOUNTING PERFORMANCE MEASURES 365
employees each.’ The companies were
sampled randomly from the list of manufactur-
ing firms published in Kompass: Register of
Industry and Commerce of Singapore
(1992). Singapore was chosen as a surrogate
for the Asian block of nations with high power
distance and low individualism cultural dimen-
sions (Hofstede, 1980, 1983, 1984; Harrison,
1992, 1993). As it is also English speaking, the
difficulties associated with the translation of
the questionnaire were avoided (Chow et al.,
1991; Harrison, 1992). Manufacturing compa-
nies were selected to determine if Harrison’s
(1992) results (on the cross-cultural generaliza-
tion of results on budgetary participation),
based only on departmental and retail stores,
could be extended to manufacturing compa-
nies where the use of budgets is common.
Eighty companies, representing about 15% of
all manufacturing companies in Singapore
with more than 100 employees each, were con-
sidered adequate to ensure external validity.
Only companies with more than 100 employ-
ees each were chosen because smaller compa-
nies may not be organized on a work-unit
(departmental) basis (Brownell & Dunk,
1991). The requirement for companies to be
organized on a work-unit basis is important
because the measurement instruments for
task uncertainty (Withey et al., 1983) and task
difficulty (Van de Ven & Delbecq, 1974) used in
this study are designed to be used on a work-
1 Copies of the questionnaire are available from the authors
unit basis (Van de Ven & Delbecq, 1974;
Withey et al, 1983; Brownell & Dunk, 1991).
Three respondents from three different func-
tional areas including the heads of produc-
tion, sales or marketing and a staff functional
area (e.g. personnel, administrative or account-
ing) were selected from each company allow-
ing the research results to be generalized
across functional areas.* The names of the func-
tional heads were obtained by telephone calls
to the companies to ensure that the question-
naire would reach, and be answered, only by
the selected respondents. Only functional
heads were selected as they were regarded as
surrogates for work-units. Moreover, one res-
pondent from each functional area was selec-
ted to ensure independence of observations.
A questionnaire with a covering letter
explaining the objectives of the study and
assuring anonymity of response was mailed to
each respondent. A self-paid addressed envel-
ope was also enclosed for the completed ques
tionnaire to be returned directly to one of the
researchers in Singapore. Follow-up of the
questionnaire was by telephone calls to those
who had not responded after 3 weeks.
A total of 129 responses were returned yield-
ing a response rate of 53.8%. Of the 129
responses, 3 were excluded from the study
for incomplete responses and 14 for unreliable
responses.3 This leaves the study with 112
usable responses.4 Of these, 49 were from the
* The staff functional heads (non-production and non-marketing or sales) were not selected randomly as most companies
contacted were reluctant to supply the researchers with a complete list of their staIf functional heads. Consequently, the
questionnaire was mailed to the staff functional heads suggested by the companies and these were genetally from
administration, personnel or accounting.
3 The questionnaire includes a seven-item measure for task dilIiculty. Of the seven items, two items are polar reverse. The
mean score of the two polar reverse items was compared with the mean score of the other five items for each respondent
to isolate any oudiers. Fourteen responses have a difference of 4 or more out of a maximum of 6. The coefficient of
correlation (r) of the mean score of the two poIar reverse questions and the mean score of the other five items of these 14
responses was a very high negative 0.83. This justihes the exclusion of these 14 outIiers from the study. There are no
signiticant differences between the regression results based on 112 responses (after the exclusion of the 14 responses) and
those based on 126 responses (before the exclusion of the 14 responses).
* The presence of non-response bias was investigated by the test suggested by Oppenheim (1!966, p. 34). No statistIcaIIy
significant differences in the mean scores between the earIy and Iate responses were found.
366 CHONG M. LAU et al.
production function, 19 from the sales or mar- styles separately and found significant differ-
keting function and 44 from staff functions. ences in job-related tension and negative atti-
The mean age of the respondents was 38 years tudes between the extent (Budgetconstrained
and 90% of them had tertiary qualifications.
and Budget-profit styles) and the manner
They had held their current positions for an
(Frofit-conscious style) of the reliance on
average of 4.5 years and had been employed
accounting measures in performance evalm-
by their respective companies for 7.7 years.
tion, there is also a theoretical justification to
The average length of experience in the areas
evaluate budget emphasis on the scores of the
they managed was 9 years and the average
“Meeting the budget” item alone instead of
number of employees in their areas of
summing the two accounting items. We tested
responsibility was 109.
our hypotheses using both approaches and
found no significant differences in the results.
Measurement of variables
This could be due to the relatively high correla-
Budget emphasis. Consistent with Brownell
tion between the two accounting items in our
& Hirst (1986), budget emphasis was measured
study.5 Given the absence of significant differ-
with a ten-item instrument which is a modified
ences in our results and the high correlation
version of that developed by Hopwood (1972).
between the two accounting items, the results
Since the ten-item instrument also incorporates
based on the combined scores were used in
all the eight items included in the instrument
this study so that any differences in the results
used in Brownell & Dunk (1991), it could also
between this study and those of Brownell &
be regarded as being consistent with Brownell
Hirst (1986) and Brownell & Dunk (1991)
& Dunk (1991). The rating form, rather than
could be attributed to cultural differences
the ranking form, was used as the former per-
and not to a change in the measurement
mits the evaluative style to be measured as a
instrument.
continuum (Otley, 1978; Govindarajan, 1984;
Budgetaly participation. The Milani (1975)
Brownell & Dunk, 1991; Harrison, 1992) and
six-item, seven-point Like&scaled instrument
avoids the problem (reported by BrowneIl,
was used to measure budgetary participation.
1985) encountered by respondents with the
This instrument was used by both Brownell &
ranking instrument. Consistent with Brownell
Hirst (1986) and Brownell & Dunk (1991). It
& Dunk (1991), the scores of the two account-
has also been tested and used extensively in
ing items (“Meeting the budget” and “My con-
other studies, and has provided high internal
cern with costs”) were combined to provide a
reliability (Brownell, 1982a, 1985; Chenhall h
score for budget emphasis. The objective of
Brownell, 1988; Dunk, 1989; Mia, 1989; Frucot
combining the two items is to operationalize
& Shearon, 1991; Harrison, 1992). The
budget emphasis “along a continuum from
Cronbach alpha (Cronbach, 1951) of 0.91
accounting to non-accounting style of evalua-
obtained for this study indicates high internal
tion” (Browneli & Dunk, 1991, p. 702). How-
reliability.
ever, since Hopwood (1972) had categorized Task uncertainty. The nine-item, seven-point
the Budget-constrained and Profit-conscious Likert-scaled instrument developed by Withey
’ The two accounting items have a coef5cient of correlation of 0.56 (p 4 0.01). As Eu as the “Meeting the budget” item is
concerned, this represents the highest correlation it has with any bf the other nine items. As far as the “My concern with
costs” item is concerned, this is also the highest correlation it has with any of the other items except for the item,
“Concern with quality” which also has a correlation coelkient of 0.56 with the “My concern with costs” item. Brownell
(1985) has argued that the legitimacy of w the two account@ items is dependent on a signi5cant correlation
between the responses to the two items. Hopwood’s (1972) Budget-profit styie was also made up of respondents who had
ranked both “Meeting the budgets” and “My concern with costs” in the top three at c gor i es and he found no s@Wicant
differences in his results between the Budget-pro5t style and the Budget~onstmined style.
et al. (1983) was employed to evaluate task despite the low Cronbach alpha scores. The
uncertainty. This instrument was employed by Cronbach alpha obtained in this study was
Brownell & Hirst (1984) and represents an addi- 0.41. Deleting the two polar reverse items
tive scale for the composite measure of task increases the Cronbach alpha to 0.52 which
difficulty and task variability. The scale was is close to that obtained by Brownell & Dunk
used by Brownell & Hirst (1986) because it (1991). However, as there were no significant
was consistent with Thompson’s (1967) defti- differences between the regression results
tion of task uncertainty and because the scale based on the scores of the seven-item instru-
had been tested and found to be reliable by ment and those based on the scores of the
Withey et al. (1983). The Cronbach alpha revised five-item instrument, the results based
(Cronbach, 1951) of 0.85 obtained in this on the score of the seven-item instrument were
study indicates satisfactory internal reliability reported to ensure consistency with Brownell
and is close to the 0.82 obtained by the & Dunk (1991). Given the low alpha in our
Brownell & I-I&t (1986). study, the reliability of task difficulty is ques-
Tusk dificulty. Task difficulty was evaluated
tionable. Hence, caution should be exercised
with the Van de Ven & Delbecq (1974) seven-
in interpreting the results of tests involving
item instrument. A seven-point rating scale con-
this variable. Remedial strategies are discussed
sistent with that used by Brownell & Dunk
in the final section of this report.
(1991) was used.6 This subscale was tested by J ob-related tension. Consistent with Brownell
Withey et al. (1983) along with five other simi- & Hirst (1986) job-related tension was mea-
lar scales developed by Hage & Aiken (1969) sured with a seven-point Likert-scaled, fifteen-
Lynch (1974) Glisson (1978) Van de Ven & item, job-related tension index (Kahn et al,
Ferry (1980) and Daft & Macintosh (1981). 1964) developed by the Institute for Social
The Van de Ven & Delbecq (1974) scale and Research at the University of Michigan. The
the Daft & Macintosh (1981) scale were scores on this index were found to be corre-
regarded by Withey et al. (1983) as the two lated positively with role conflict, role ambi-
most appropriate scales for research as they guity and anxiety (Kahn et aZ., 1964, pp. 424-
both have reasonable measurement properties 427). The Cronbach alpha obtained in this
in terms of the extent to which the two instru- study was 0.87.
ments measure the same construct as other Managerial perjomance. The ninedimen-
instruments (from separate studies of work- sional seven-point Likert-scaled Mahoney et al.
unit technology), and in terms of their relative (1963, 1%5) self-rating measure was employed
convergent and discrlminant validity compared to evaluate this dependent variable. The scale
with other similar instruments. In their study, comprises eight performance dimensions and
Van de Ven & Delbecq (1974) reported a reli- one overall effectiveness dimension. This self-
ability alpha of 0.86 for the task difficulty rating scale has been criticized for the ten-
instrument. However, the Cronbach alpha dency of respondents to be too lenient on
(Cronbach, 1951) obtained by Brownell & themselves, thereby resulting in a small range
Dunk (1991) with this scale was a low 0.57. in the scores being observed (Prien & Liske,
Another study by Mia (1987) with this scale 1%2; Thornton, 1968; Mia, 1989). However,
achieved a similarly low alpha value of 0.55. according to Brownell (1982a), it has the
Since task difficulty was an important variable advantages of overcoming the problem of
in Brownell & Dunk (I991), the same additive superiors’ tendency “to evaluate on only one
scale instrument was used in this study to facil- cognitive dimension” and to “clearly capture
itate comparability of studies between cultures, the multidimensional structure of performance
RELIANCE ON ACCOUNTING PERFORMANCE MEASURES 367
o In their e work, Van de Ven % Delbecq (1974) used a ten-item interval scale.
368 CHONG M.
without introducing the problem of excessive
dimensional&y” (pp. 17-18). Heneman (1974)
also found a very high correlation between
superior-rating and self-rating scales. Consis-
tent with both Brownell & Hirst (1986) and
Brownell & Dunk (1991), only the score of
the overall effectiveness dimension was used
for managerial performance in this study.
Regressing the score of the overall effective-
ness item on the scores of the other eight
items provided a coefficient of determination
of 0.61 (p < 0.001) which is higher than the
0.55 (desirable lower limit) suggested by
Mahoney et al. (1963). The remaining 39% of
unexplained variance is probably job specific
or related to individual differences and sug-
gests that factors other than those covered by
the questionnaire also influence overall man-
agerial performance. Given the cross-sectional
nature of our sample and the fact that we have
used the instrument in a different culture from
that within which it was developed, this level
of R2 is not surprising.
Descriptive statistics for the variables meas-
ured in our study are shown in Table 1 and the
intercorrelations among the variables are
LAU et al.
shown in Table 2. The Pearson pairwise corre-
lations among the independent variables are
not high (less than 0.5). In addition, because
interaction regression models were used in
this study, the approach of centring the inde-
pendent variables (by using deviation scores) is
adopted to reduce the correlations between
the product terms and the component parts
of the independent variables (Southwood,
1978; Cohen & Cohen, 1983; Cronbach, 1987;
Jaccard et cd., 1990). Tolerance of greater than
0.1 was achieved for all independent variables
in the regression models used. Multicollinearity
was therefore not considered to be a problem.
RESULTS AND DISCUSSION
Hierarchical regression models were used to
analyse our results. This approach was selected
to facilitate the partitioning of the proportion
of the total variance accounted for by the addi-
tive model, the two-way interactive model and
three-way interactive model (Cohen & Cohen,
1983). Since the main effects of the three-way
interaction model are arbitrary and have no
TABLE 1. Descriptive statistics for the variables in the study (n = 112)
Theoretical range Actual range
Variable Mean S.D. Min. MaX. Min. MaX.
Budgetary participation 30.6518 7.6320 6 42 6 42
Budget emphasis 11.66% 2.1619 2 14 2 14
Task uncertlinty 37.2320 9.2440 9 63 18 57
Task difficulty 24.4911 4.6557 7 49 14 39
Job-related tension 47.6964 12.0720 15 105 15 80
Managerial performance 5.79% 0.8398 1 7 4 7
TABLE 2. Correhtion matrix among dependent and independent variables
Jobrelated
Variable Performance tension Participation Budget emphasis
Job-related tension -0.31”
Participation -0.39” 0.15
Budget emphasis -0.42” 0.13 0.45”
Task uocertainty 0.09 0.03 0.09 0.01
Task difficuky 0.21. -0.22. -0.01 -0.07
* p c 0.01.
l p s 0.05.
RELIANCE ON ACCOUNTING PERFORMANCE MEASURES 369
theoretical meaning as we have used non-ratio
scales (Allison, 1977; Southwood, 1978;
Schoonhoven, 1981) the hierarchical
approach also permits the main effects and
the interactive effects of the independent vari-
ables to be analysed separately. The following
models were used:
Yi = bo +blPi +bzB* +b$i +et
(1)
Yr = bO +b,P, +b2B, +b3T, +b4pI Bi +
bP& +b6 BA +ei (2)
Yi = bO +b,P, +b2B, +b& +bflJ .3, +
bP& +b&J , +b7P&Wi +ei, (3)
where Yi is job-related tension (HOI) or manage-
rial performance @I,&; Pi is budgetary partici-
pation; Bi is budget emphasis; T, is task
uncertainty ml> or task difficulty @I&; and
ef is the error term.
Equations (1) (2) and (3) were used to ana-
lyse the main effects, the two-way interactive
effects and the three-way interactive effect,
respectively. Deviation scores based on the
overall mean score minus the raw scores
were used for all independent variables in the
regression models. Hence, low raw scores have
positive deviation scores and high raw scores
have negative deviation scores. This enables
the coefficient of the two-way interaction
between budget emphasis and participation in
low task uncertainty situations to be predicted
(we elborate upon this later). Tests on the ade-
quacy of the multiple linear regression models
indicate that the inherent assumptions of the
models were satisfied by the data7
Hypothesfs HOI
Hypothesis HoI states that there is no statis-
tically significant three-way interaction between
budget emphasis, budgetary participation and
task uncertafnty affecting job-related tensfon.
To test this hypothesis, job-related tension was
regressed on budget emphasis, participation
and task uncertainty. The results in Table 3
(equation (3)) indicate that the three-way inter-
action between budget emphasis, participation
and task uncertainty affecting job-related
tension is not significant as reflected by the
insignificant b, coefficient (p d 0.791). Accord-
TABLE 3. Results of regression of jobrelated tension on budget participation, budget emphasis and task uncertainty
Equation (I) Equation (2)
Bquation (3)
(additive)
(tJvov=y) (three-way)
Variable Coeff. Est.
P
Est.
P
Est.
P
constant
bo
47.6964
Budget participation (P) bl 0.1889
Budget emphasis (B)
bz 0.4390
Task uncen.dnty (2’)
bs
0.0269
PXB
b4
PXT
b5
BXT
b6
PXBXT
b7
R2 0.030
F value 1.110
PG
0.348
J? explained by twoway interaction terms = 4.5%
If explained by three-way interaction term = 0%
0.000 48.1%2
0.267 0.2405
0.463 0.6208
0.829 0.0733
-0.0409
-0.0308
0.0309
0.075 0.075
1.414 1.211
0.216 0.303
0.000 48.21% 0.000
0.161 0.2305 0.192
0.329 0.6303 0.325
0.566 0.0593 0.669
0.429 -0.0500 0.422
0.052 -0.0279 0.152
0.619 0.0206 0.778
0.0024 0.791
‘The tests to check for the three important assumptions of the selected regression models (normality of residuals,
homogeneity of variance of residuals and the appropriateness of the linear models) indicate that the assumptions inherent
in the regression analysis were satisfied in all equations.
370 CHONG M IAU et al .
ingly, hypothesis HoI cannot be rejected. This
result is inconsistent with Brownell & Hirst’s
(1986) finding. No significant results were
found in the two-way interaction model nor
in the additive model.
our failure to control for the effect of uncer-
tainty avoidance on job-related tension.
Hypothesis Ho2
To determine if Brownell & Dunk’s (1991)
observation that the subdimension of task diff?-
culty, rather than task uncertainty, may be a
more relevant variable to measure task chamc-
teristics, job-related tension was regressed on
budget emphasis, participation and task dfffi-
c&y. This three-way interaction was also
insignificant (p s 0.49).
A possible explanation for our failure to
replicate the results of Brownell & Hirst
(1986) could be due to differences in the
sampling methods between the two studies.
Brownell & Hi&s (1986) sample was selected
from a single manufacturing company, whereas
our sample was selected from 80 different man-
ufacturing companies. Alternatively, cultural
differences could have infhtenced the results.
Our study has considered the influence of the
two underlying cultural dimensions of power
distance and individualism on employees’ reac-
tions to budgetary participation; but has not
considered the influence of the cultural dimen-
sion of uncertainty avoidance. Uncertainty
avoidance refers to the ways societies adapt
to uncertainty in life. Low uncertainty avoid-
ance countries have lower anxiety levels in
the population and lower job stress (Hofstede,
1980, p. 176). High uncertainty avoidance
nations have higher anxiety in the population
and higher job stress. Since the dependent vari-
able in HoI is job-related tension, it is feasible
that cultural differences in respect to uncer-
tainty avoidance could influence it. The range
of uncertainty avoidance index for the 40
nations studied by Hofstede (1980) is 8-112.
Significantly, Singapore has a score of 8 while
Australia has a score of 51. Given this substan-
tial difference in the score between the two
nations, it is feasible that our failure to repli-
cate Brownell & Hirst (1986) could be due to
Q, states that there is no statistically sign&
cant three-way interaction between budget
emphasis, budgetary participation and task dff-
@ulty affecting managerial momrance. To
test this hypothesis, managerial performance
was regressed on budget emphasis, participa-
tion and the subdlmension of task difficulty.
The results in Table 4 (equation (3)) show a
significant three-way interaction between bud-
get emphasis, participation and task difficulty
affecting managerial performance (p 6 0.032).
Hypothesis H,,, can therefore be rejected. This
implies that Brownell’s (1982a) results cannot
be generalized across all levels of task difficulty.
To test Brownell & Dunk’s (1991) assertion that
task difficulty and not task uncertainty is the
task characteristic associated with perfor-
mance, the three-way interaction between bud-
get emphasis, budgetary participation and task
uncertainty affecting managerial pe$or-
mance was also tested. As expected, no signif-
icant three-way interaction was found.
Table 4 indicates that the two-way interac-
tions (equation (2)) and the three-way interac-
tion (equation (3)) only enhance the predictive
ability of the model by 4.6% and an additional
3%, respectively. This suggests that the two-
way interactions and the three-way interaction
(although statistically significant) may have less
practical significance, relative to the main
effects (26.8%). However, we are unable to
compare the main effects of our study with
those of Brownell & Dunk (1991) as they
have not disclosed the main effect results in
their article.
To determine if compatible combinations of
high (low) budget emphasis and high (low)
participation is associated with improved per-
formance only in low task difficulty situations,
task difficulty was dichotomized at its mean.’
’ There arc no si gni f i c ant di f f er enc es i n t he r egr essi on r esul t s based on dichotomizing task diiliculty at the mean and those
obtained when task difiiculty is dichotomized at the median.
RELIANCE ON ACCOUNTING PERFORMANCE MEASURES 371
TABLE 4. Results of regression of e performance on budget participation, budget emphasis and task diGculty
Variable Coeff.
Equation (1)
(additive)
Est.
P
Equation co Equation (3)
(*~~y~ (thrrcway)
Est.
P
Est.
P
constant
Budget participation (9
Budget emphasis (s>
Task difficulty (T,
PXB
PXT
BXT
PXBXT
b0
5.7946 0.000 5.7699 0.000 5.7593 0.000
bl -0.0284 0.006 -0.0321 0.002 -0.0365 0.001
bz
-0.1138 0.002 -0.1234 0.002 -0.1078 0.006
b,
0.0350 0.021 0.0307 0.039 0.0162 0.311
b4
0.0031 0.306 0.0044 0.155
bs
0.0054 0.032 0.0055 0.026
b6
-0.0042 0.630 -0.0113 0.223
b7
0.0019 0.032
R2 0.268 0.314 0.344
F value 13.152 8.010 7.781
P S 0.001 0.001 0.001
J? explained by two-way interaction terms = 4.6%
I? explained by three-way interaction term = 3.0%
The following regression models were used to
test the main effects and the twoway interac-
tion effect between budget emphasis and parti-
cipation for each of the low and high task
difGculty situations:
Yi = co + clPi + czBr + el
(4)
Yi = co + clPi +c& +c$&l, +ei,
(5)
where Y, is managerial performance; Pi is bud-
getary participation; & is budget emphasis; and
ef is the error term.
Equations (4) and (5) were used to analyse
the main effects and the two-way interaction
effect, respectively. Since deviation scores
based on the overall mean less raw scores
were used for all independent variables, high
participation and high budget emphasis have
negative deviation scores and low participa-
tion and low budget emphasis have positive
scores. The product of the deviation scores of
low budget emphasis and low participation
would be the same as the product of the devia-
tion scores of high budget emphasis and high
participation. This enables the direction of c3,
the coefficient for the twc~~3y interaction
between budget emphasis and participation
to be predicted. Since compatible combina-
tions of high (‘low) budget emphasis and high
(low) participation are expected to be asso-
ciated with improved performance only in
low task difficulty situations, c3 was expected
to be positive in low task difficulty situations.
The results of the regression are shown in
Table 5 for the low task difhculty condition
and in Table 6 for the high task ticulty
condition.
As expected, the interaction between parti-
cipation and budget emphasis on managerial
performance was found to be sign&ant and
the coefficient positive for the low task diffr-
culty subsample (p G 0.016 in Table 5, equa-
tion (5)), but not for the high task diBiculty
subsample (p G 0.944 in Table 6, equation
(5)). Note that with the low task difficulty sub-
sample (Table 5), the J? of the regression
model with the interactive term (0.271 in equa-
tion (5)) is considerably higher than the Ra of
the additive model (0.191 in equation (4)).
However, in the case of the high task difficulty
subsample (Table 6), the value of I? of the
model witb the interactive term (0.295 in equa-
tion (5)) is the same as that of the additive
model (0.295 in equation (4)). These results
support hypothesis HAzp that budget emphasis
interacts with participation onty in low task
difficulty situations. Thus the null hypothesis
372 CHONG M. LAU et al.
TABLE 5. Results of regression of managerial performance on budget participation and budget emphasis - low task
difficulty subsample
Equation (4) Equation (5)
(additive)
@weway)
Variable Coeff. Est.
P
Est.
P
constant CO 3.%15 0.000 5.8070 0.000
Budget participation (P) cl 0.0089 0.468 -0.0172 0.164
Budget emphasis (B) cz 0.1408 0.003 -0.1450 0.002
PXE
c3
0.0117 0.016
ti 0.191 0.271
F value 6.729 6.929
PS 0.002 0.001
ti explained by twc+way interaction terms = 8%
TABLE 6. Results of regression of managerial performance on budget participation and budget emphasis - high task
difliculty subsample
Equation (4) Equation (T)
(additive)
@weway)
Variable Caeff. Est.
P
Est.
P
constant CO 3.2606 0.000 5.7024 0.000
Budget participation (P) Cl 0.0528 0.004 -0.0527 0.004
Budget emphasis (E) c2 0.0704 0.231 -0.0677 0.344
PXB
c3
-0.0003 0.944
Rz 0.295 0.295
F value 10.24 6.692
PC 0.001 0.001
J? explained by nvoway interaction terms = 0%
Hc,a may be rejected in favour of the two alter-
native hypotheses HMl and Hub. _
The results in Table 6 also indicate that in
high task difficulty situations, only the coeffi-
cient for participation was positive and signifi-
cant (P
=Z 0.004, equation (4)). Budget
emphasis, however, had no significant main
effect (p d 0.231 in equation (4)) nor interac-
tive effect (p S 0.944 in equation (5)). These
results support hypothesis HAa,-, that participa-
tion is important in high task difficulty
situations regardless of budget emphasis.
There are, however, important differences
between our results and those of Brownell &
Dunk (1991), particularly in low task difficulty
situations. First, the interaction between parti-
cipation and budget emphasis resulted in a 20%
increase (0.276-0.073) in Ra in the Brownell &
Dunk study. The corresponding increase in R*
in our study was only 8% (0.271-0.191) (Table
5). Second, Brownell & Dunk (p. 701) found no
sign&ant main effect (p s 0.406) with budget
emphasis in the low task difficulty situation. By
contrast, our results indicate a significant main
effect for budget emphasis (p G 0.003 in equa-
tion (4), Table 5). The Ra of the additive model
(equation (4)) was a substantial 19% (more
than twice that of the interaction effect), most
of which was accounted for by budget empha-
sis since participation was not significant (p S
RELIANCE ON ACCOUNTING PERFORMANCE MEASURES
373
TABLE 7. CeII frequencies, mean and standard deviation for
managerial performance across high and low budget
participation and budget emphasis - low task difficulty
subsample
Budget emphasis
Participation
Low
High
LOW
=eh
n = 12 n = 10
7 = 5.58 7 = 6.00
o,, = 0.67 a, = 0.67
n=9 n = 29
p = 5.44 9 = 6.17
0, = 0.73 0, = 0.66
TABLE 8. CeII frequencies, mean and standard deviation for
managerhI performance across high and low budget
participation and budget emphasis - high task diflicuky
subsample
Budget emphasis
Participation
Low
High
LOW
ma
n = 13 n=9
9 = 4.85 7 = 5.56
a, = 0.90 0, = 0.53
n=9 12 = 21
ji = 5.78 7 = 6.14
o, = 0.83 o,, = 0.85
0.468). The combination of (i) a signiticant
interaction effect between budget emphasis
and participation on performance in low task
difficulty situations (Brownell & Dunk, 1991);
and (ii) a positive cultural effect on subordin-
ates’ reaction to budget emphasis in Singapore
(Harrison, 1992, 1993) may have augmented
the effect of budget emphasis on performance
in Singapore, resulting in a significant main
effect not found by Brownell 81 Dunk (1991)
in Australia. These results support Harrison’s
(1992, 1993) theory and findings that culture
has a significant interaction effect with budget
emphasis. Overall, our results in low task dif&
cuhy situations therefore suggest that manage-
rial performance in Singapore was driven by
both budget emphasis and the interaction of
budget emphasis and participation. By con-
trast, Brownell & Dunk (1991) found that
performance in Australia was driven mainly
by the interaction of budget emphasis and
participation.
To assist the interpretation of the results of
equation (5) in the low and high task difficulty
subsamples presented in Tables 5 and 6,
respectively, the levels of both participation
Graph (a) Low task difficulty Graph cb) High task difficulty
Per
6.0
5.5
5.0
f
5
high budget
6.17
5.58 -
low -
5.44
budget emphasis
I I
I I
P&
6.0
5.5
5.0
6.14
I I
LOW
=sh
Participation
LOW
High
Participation
Fig. 1. Three-way interaction graphs
374 CHONG M
and budget emphasis were dichotomized at the
mean. The results (based on the raw scores) are
reported in Tables 7 and 8, and illustrated
graphically in Fig. 1.
Note that for the low task difficulty sub-
sample in Table 7, the mean performance
score is higher in the low budget emphasis
and low participation cell (5.58) than in the
low budget emphasis and high participation
cell (5.44). Similarly, the mean performance
score is higher in the high budget emphasis
and high participation cell (6.17) than in the
high budget emphasis and low participation
cell (6.00). These results are consistent with
hypothesis HAza that, in low task difliculty situa-
tions, compatible combinations of high (low)
budget emphasis and high (low) participation
are associated with improved managerial
performance.
However, the results in Table 7 and Fig. l(a)
also indicate that the mean performance of a
high budget emphasis and low participation
combination (6.00) is higher than that of a
low budget emphasis and low participation
combination (5.58). These results are different
from those of Brownell & Dunk (1991) as they
had observed reverse results. Taken together
with the result that high budget emphasis and
high participation (6.17) is superior to low bud-
get emphasis and high participation (5.44), the
results suggest that, regardless of participation,
a high budget emphasis is always superior to a
low budget emphasis in terms of its impact on
managerial performance in low task difficulty
environments. To test for this possibility, the
point of intersection of the regression lines
for (i) participation and performance for high
budget emphasis and (ii) participation and per-
formance for the low budget emphasis sub-
sample was computed (Kerlinger & Pedhazur,
1973, p. 247). This point represents the level of
participation at which the level of performance
for high budget emphasis is equal to that of low
budget emphasis. The point of intersection for
IAU et al.
the participation score is 19 which is within
the actual observed range of 6 and 42, the
research range of interest. This interaction is
regarded as disordina19 indicating that high
budget emphasis is superior to low budget
emphasis (in terms of performance) for partici-
pation scores of more than 19, and low budget
emphasis is superior to high budget emphasis
(in terms of performance) for participation
scores of less than 19. Only 4 respondents
(7%) in the low task difficulty subsample had
participation scores of 19 or less. Conse-
quently, even though we cannot conclude
that a high budget emphasis is always superior
to low budget emphasis for all levels of partici-
pation, our results indicate that a high budget
emphasis is superior to a low budget emphasis
for a large majority of the Singaporean
managers.
To explore further the levels of participation
at which performance can be significantly
attributed to the level of budget emphasis,
the range of participation within which no sig-
nificant differences in the level of performance
could be attributed to the level of budget
emphasis was computed using the Johnson-
Neyman technique (Kerlinger & Pedhazur,
1973, p. 256). The range of non-significance
(alpha = 0.05) was between 16.6 and 32.6.
This means that for participation scores in
excess of 32.6, high budget emphasis is statis-
tically superior to low budget emphasis in
terms of performance. For participation scores
of less than 16.6, low budget emphasis is statis-
tically superior to high budget emphasis.
Within the range between 16.6 and 32.6,
neither level of budget emphasis is superior.
Since the mean level of participation for the
low task difficulty subsample (31) is within
this range, we can conclude that a high budget
emphasis is not always superior to low budget
emphasis at all level of participation.
However, as the mean of 31 is near the
upper @vel of the non-significance zone, our
’ The results of the test suggested by Schoonhoveu (1981> also indicate tbat budget emphash has a nonmonotonic effect
on managerial pexfomance as the point of infIexion of 19 for tbe raw pattic@ation score lies within the observed range of
participation for the low task dit%culty subsample.
RELIANCE ON ACCOUNTING PERPORMANCE MEASURES 375
Fig. 2.
6.5
6.0 Y6.77994.CXM2K
Y=4.8629+0.0379K
6
19 31 42
mean
Kuticipaticm
Results of Johnson-Neyman test - low task di!iiculty subsamples. The shaded area represents the non-signikancc
7.one, Pi i s the raw score.
result also suggests that for many respondents,
performance is influenced mainly by high bud-
get emphasis, rather than by the intemction of
high budget emphasis and participation. A
further analysis of our low task difficulty sub
sample indicates that there were 4 respondents
with participation scores of 16 or less; 24 with
participation scores of between 17 and 32; and
32 with parddpation score in excess of 33.
This also suggests that for a large majority of
our respondents ln our low task difficulty sub
sample, a high budget emphasis is either (a)
associated with improved performance (53%);
or (b) not detrimental to performance (40%).
Only for a small minority (7%) is a low budget
emphasis associated with improved perfor-
mance. This may explain our results in Table
7 which indicate that a high budget emphasis
and a low participation combination (6.0) is
better than a low budget emphasis and a low
participation combination (5.58). In this
regard, our results are di&rent from those of
Brownell & Dunk (1991, p. 700) who found
that a low budget emphasis and low partidpa-
tion combination (5.53) is superior to a
high budget emphasis and low participation
combination (5.0).
Table 7 also indicates that the number of
respondents in the high budget emphasis cate-
gories (39) is also substantially more than those
in the low budget emphasis categories (21).
Together with our finding that participation
does have a moderating effect on the relation-
ship between budget emphasis and perfor-
mance, our results suggest that in high power
distance and low individualism societies, the
best strategy in low task d@iculty envhm-
men& is a bigb budget emphasis and bigb
particiftfon combfnation. This is substan-
tiated by the large numbef of respondents in
this category (29) which is nearly three times
that of each of the other three categories.
These results are consistent with Browneli h
Dunk’s (1991) finding that a high budget
emphasis and a high participation is a pre-
ferred combination. Although Brownell &
Dunk also found a low budget emphasis and
low participation combination to be desirable,
376 CHONG M
this combination is not popular in Singapore
because the Singaporean culture favours high
budget emphasis (Harrison, 1992, 1993). The
incompatibility of low budget emphasis with
the Singaporean culture is probably the reason
for the low number of respondents in the low
budget emphasis and low participation cate-
gory (12), even though this combination may
in some cases be associated with improved
performance.
For high task difficulty, an inspection of the
cells in Table 8 and Fig. l(b) reveals that parti-
cipation is important regardless of the level of
budget emphasis. With low budget emphasis,
the high participation and low budget empha-
sis cell mean (5.78) is higher than that of the
low participation and low budget emphasis celI
(4.85). With high budget emphasis, the high
participation and high budget emphasis cell
mean (6.14) is higher than that of the low par-
ticipation and high budget emphasis cell
(5.56). The argument for hypothesis HAZb is
that when task difficulty is high, participation
is associated with improved performance
regardless of whether it is matched with bud-
get emphasis or not. These results are consis-
tent with the theory that participation exerts
most of the effect on performance in high
task difficulty (Brownell & Dunk, 1991) and
culture does not inlluence subordinates’ reac-
tions to participation (Harrison, 1992, 1993).”
Our results in Table 8 and Fig. l(b) show that
for both low and high levels of participation,
high budget emphasis is superior to low budget
emphasis (5.56 vs 4.85, and 6.14 vs 5.78).
These results are consistent with Harrison’s
(1992, 1993) theory that Singaporean man-
agers react more favourably to a high budget
emphasis than a low budget emphasis. How-
ever, since Table 6 shows that budget empha-
sis has no significant main effect (p G 0.231;
equation (4)), nor significant interactive effect
(p d 0.944; equation (5)), the positive effect of
budget emphasis on performance in a high task
LAU et al.
difficulty situation is therefore not statistically
significant. Moreover, from the results in Table
8 and Fig. l(b), it is not possible to conclude
that a high budget emphasis always results in
superior performance than does a low budget
emphasis, as a low budget and high participa-
tion combination (5.78) is still superior to a
high budget emphasis and low participation
combination (5.56). These results are consis-
tent with hypothesis HAZb that regardless of
the types of supervisory evaluative style, high
budgetary participation is associated with
improved managerial per$ormance in high
task difficulty situations. These results suggest
that the positive impact of culture on budget
emphasis in Singapore (Harrison, 1992, 1993)
is not sufficient to influence the results of
Brownell & Dunk (1991) in high task difficulty
situations.
CONCLUSIONS AND LIMITATIONS
The replication in another culture of
research performed in one culture is important
for an understanding of the impact of culture
(Alder, 1984; Frucot & Shearon, 1991; Harrison,
1992, 1993; Otley et al., 1994). With regard to
managerial performance, our results confirm
that (i) high budget emphasis and high budget-
ary participation in low task df@x.&y situa-
tions and (ii) high budgetary participation in
high task dificulty situations, regardless of
budget emphasis, are both associated with
improved managerial performance. These
results are consistent with, and hence rein-
force, those of Brownell & Dunk (1991). The
sign&ant results on managerial performance
associated with task drfficulty and not with
task uncertainty also provide evidence to
substantiate BrownelI & Dunk’s (1991) sugges
tion that task dfffi cul ty is the more appro-
priate choice of instrument to measure task
characteristics.
lo To determine if our results were tienced by the moderate level of task difiiculty in our sample, we also dichotomized
our sample based on the task dif6culty theoretical mean of 28. This resulted in 80 cases for low task difticulty and 32 cases
for high task difiiculty. There were no differences in the general patterns of our results.
RELIANCE ON ACCOUNTING PERFORMANCE MEASURES 377
However, our results also differ from those
of Brownell & Dunk (1991). First, our results
indicate that a low budget emphasis and low
participation combination is not superior to a
high budget emphasis and low participation
combination in low task difficulty situations.
Second, in a low task difficulty environment,
budget emphasis has a highly signiticant main
effect, which is substantially more than that of
the interaction between budget emphasis and
participation. Differences in our results and
those of Brownell & Dunk (1991) appear to
be caused not by subordinates’ reactions to
budgetary participation, but by subordinates’
positive reactions to budget emphasis. The
extent and the direction of the differences in
the results are consistent with Harrison’s
theory on the influence of national culture on
budget emphasis and participation. Hence our
results provide additional evidence to substanti-
ate Harrison’s (1992, 1993) findings on the
impact of national culture. Since our results
are based on a sample of manufacturing com-
panies, incorporating managers from various
functional areas, they also suggest the possi-
bility that Harrison’s (1992, 1993) conclusions
can be generalized to a broader range of indus-
tries and functional areas. More importantly,
these confirmations provide additional evi-
dence in support of Harrison’s research frame-
work for examining the impact of culture on
other aspects of management accounting sys-
tems. Further research work along this line,
not only in the budgetary area, but also on
other aspects of managerial accounting, would
enable Asian businesses to adopt Western man-
agerial accounting systems with greater confi-
dence and, when and if appropriate, in a more
discriminatory manner.
Further research using more systematic tests
of the impact of culture on the relation
between budgetary participation and manage-
rial performance would also be worthwhile.
For example, the use of a pooled sample
drawn from different cultural blocks, together
with tests on the effects of budget emphasis,
participation, task difficulty and culture affect-
ing managerial performance, may provide
further evidence to evaluate Harrison’s conclu-
sions. Other dependent variables, such as job
satisfaction and other job-related attitudes and
budget-related attitudes, could also be studied.
Moreover, both our and Harrison’s (1992,
1993) studies have been conducted in low
power distance and high individualism, and/
or high power distance and low individualism
blocks. Similar studies on nations in the other
two cultural blocks characterized by high
power distance and high individualism, and
low power distance and low individualism
would provide a more rigorous test of
Harrison’s (1992, 1993) framework. Finally,
our failure to replicate the results of Brownelf
& Hirst (1986) in Singapore suggests that the
relationship between budget emphasis,
national culture and job-related tension may
be a complex one. Further research which con-
siders the effect of the cuhural dimension of
uncertainty avoidance may be worthwhile.
Although the results of this study provide
some support for those of Brownell & Dunk
(1991) and Harrison (1992, 1993) generalizing
the effect of culture on the research work on
budgetary participation and budget emphasis
should be viewed with caution, First, as with
most studies on supervisory evaluative styles
based on questionnaire surveys, our study has
the usual limitations associated with that
methodology. Hidden biases and random
errors can also inflate associations (see Bagozzi
et d, 1991, p. 424). Second, comparisons of
research results based on research work done
in different time periods may be distorted by
inter-temporal changes. Third, the responses
we observed for the task difficulty measure
were limited to the moderate range, restricting
the generalizability of our results to organiza-
tions involving similar task difficulty situa-
tions. Fourth, as our sample was selected
from manufacturing companies, caution
should be exercised in generalizing our results
to other industries. Further research involving
the service industries and the public sector
would be worthwhile. Budget emphasis is
likely to remain high in the public sector
because of the appropriation process and the
378 CHONG M. JAU et al.
resultant accountability requirements imposed
on managers.
As the focus of our paper is to generalize the
results of BrowneII & Hirst (1986) and BrownelI
& Dunk (1991) to Singapore, the choice of
measurement instruments was problematic.
We have tried to use the same measurement
instruments as those in BrowneIl & Hirst
(1986) and BrowneIl and Dunk (1991) so that
the differences between our results and those
of the two prior studies could be attributed to
cuhurai difference. This means that some mea-
sures selected may have inherent Limitations.
With regard to budget emphasis, Kren & Liao
(1988) pointed out several methodological
problems including the iimitations associated
with the use of self-reported values; the use
of different instruments in different studies;
and the use of different operational defini-
tions. Some researchers (e.g. BrowneIl, 1985;
Briers h Hirst, 1990; Manzoni, 1992) have
also expressed a concern on the need to cap
ture not only the etient but also the munner in
which accounting data are used by superiors in
performance evaluation to overcome the pro-
blem of construct vaiidity. With regard to man-
agerial performance, it should be noted that
the use of overall job performance has been
regarded as problematic (Briers & Hirst, 1990;
Hirst & Lowy, 1990). Concerns have included
(i) whether budgetary performance, rather
than overall job performance, may be the
more appropriate dependent variable @Iirst &
Lowy, 1990); (ii) whether the relationship
between supervisory styles and job perfor-
mance may only be a spurious one (Otley,
1978); (iii) whether job performance should
be treated as an independent variable or a
dependent variable (BrownelI, 1982a); and
(iv) whether the use of the &f-rating scale is
appropriate (Govindarajan, 1984).
With regard to task diflicuhy, our study and
some of the other accounting studies (e.g. Mia,
1987; Browneli & Dunk, 1991) using the Van
de Ven & Delbecq (1974) instrument have
reported low alpha Cronbach scores. It is per-
haps worth noting that Van de Ven & Delbecq
(1974) in their original study on the develop
ment and testing of their task difIicuity instru-
ment report an alpha coefficient of 0.86. It
seems that the use of this scale in a different
situation has signi6cantIy lowered its reliability.
As the task difficulty instrument has a history of
low Cronbach alpha values in cross-sectional
studies, further use of this instrument in future
research should proceed with caution. Further
research with other existing task difiicuhy
instruments (e.g. Van de Ven 8r Ferry, 1980;
Daft & Macintosh, 1981) or the development
of a new instrument with improved construct
validity (Churchih, 1979; Peter, 1979) is
needed. The incorporation of alternative mea-
sures of task uncertainty and task difficulty
would also assist in determining the validity
of Browneil & Dunk’s (1991) explanation on
the failure of Brownell 8r Hirst (1986) to con-
firm their hypotheses relating to manageriai
performance.
Nevertheless, despite the limitations listed
above, this study has provided additional evi-
dence on the area of superior’s evaluative
styles. Together with those of others (e.g.
Hwang, 1989; Chow et al., 1991, 1994;
Harrison, 1992, 1993; Otiey et al, 1994) it
also constitutes an attempt to develop a rela-
tively new area of research. This line of
research has provided some guidelines for sys
tematic evaluations of the effects of national
culture, not only in management accounting,
but also in other areas of accounting research.
BIBLIOGRAPHY
Alder, N. J., Umlcrstanding the Ways of UndcrstaadinR: Cnxs-culhml Manqtcment McthodoloRy
Rcvkwcd, in Fmncr, R. N. (cd.), Advances In Zntemationd Compa?raHue Management, pp. 31-67
(Greenwich CT: JAI Press, 1984).
RELIANCE ON ACCOUNTING PERFORMANCE MEASURES 379
AI&on, P. D., Testing for Interaction in Multiple Regression, Amerfcun J ournal of Sociology (1977)
pp. 144-153.
Bagoazl, R. P., Youjae, Y. % Phillips, L. W., Assessing Construct Validity in Orgamrational Research,
AdmWstrative Sckttce Quarterly (1991) pp. 421-458.
Brlers, M. L. t Hhst, M. K., The Role of Budgetary Information in Performance Evaluation, Accounting,
Orgunizutions and Society (1990) pp. 373-398.
Brownell, P., The Role of Accounting Data in Performance Evaluation, Budgetary Participation, and
Organlsational EITectiveness, J ournal ofAccountfng Research (Spring 1982a) pp. 12-27.
Brownell, P., Participation in the Budgeting Process: When it Works and it Doesn’t, J ournal of Account-
fng Literature (1982b) pp. 124-153.
Brownell, P., Budgetary Systems and the Control of Functionally Differentiated Organnational Activities,
J ournal of Accounting Research (Autumn 1985) pp. 502-512.
BrownelI, P. & Dunk, A. S., Task Uncertainty and its Interaction with Budgetary Participation and Budget
Emphasis: Some Methodological Issues and Empirical Investigation, Accounting, Organfzutfons and
Sockty (1991) pp. 693-703.
Brownell, P. % Hlrst, M K., Reliance on Accounting Information, Budgetary Participation, and Task
Uncertainty: Tests of a Three-way Interaction, J ournal of Accounting Reseurcb (Autumn 1986)
pp. 241-249.
ChenhaIl, R. H. % BrownelI, P., The Fffect of Participative Budgeting on Job Satisfaction and Perfor-
mance: Role Ambiguity as an Intervening Variable, Accounting, Organizations and Socfev (1988)
pp. 225-233.
Cherrington, D. J. & Cherrington, J. O., Appropriate Reinforcement Countingencies in the Budgeting
Process, J ournal of Accountfng Reseanb (1973) pp. 225-253 (SuppI.).
Churchill, G. A. Jr, A Paradlgm for Developing Better Measures of Marketing Constructs, J ournul of
Murketfng Reseurcb (February 1979) pp. 64-73.
Chow, C. W., Kato, Y. & Shield, M. D., National Culture and the Preference for Management Controls: an
Exploratory Study of the Finn-Iabour Market Interface, Accountfng, Orgumhztfons and Society
(1994) pp. 381-400.
Chow, C. W., Shields, M. D. & Chan, Y. K., The Effects of Management Controls and National Culture on
Manufacturing Performance: an Experimental Investigation, Accounting, Organizations and Society
(1991) pp. 209-226.
Cohen, J. & Cohen, P., Applied Multipk Regresston/Cowelation Analysis for tbe Behavioral Sciences,
2nd Edn (London: Lawrence Erlbaum Associates, 1983).
Cronbach, L. J., Coefficient AIpha and the Internal Structure of Tests, PsycbomeMka (September 1951)
pp. 297-334.
Cronbach, L. J., Statistical Tests for Moderator Variables: Flaws in Analysis Recently Proposed, Pry&c+
logical Bulierin (1987) pp. 414-417.
Daft, R. L. 8 Macintosh, N. B., A Tentative Exploration Into Amount and Equlvocallty of Information
Processing in Otganiaational Work Units, AdmjnnLsharfve Sclmce Quarterly (1981) pp. 207-224.
Dunk, A. S., Budget Emphasis, Budgetaty Partlclpatlon and ManagerlaI Performance: a Note, Accounting,
Organizations and Society (1989) pp. 321-324.
Early, P. C., Wojnarosli, P. % priest, W., Task PMnnIng and Energy Expended: Exploration of How Goals
IntIuence Performance, J ournal of Applied PsycboZogy (1987) pp. 107-114.
FNCOt , V. % Shearon, W. T., Budgetary Partlcipatlon, Locus of Control, and Mexican Managerial Perfor-
mance and Job Satisfaction, IIre Accounttng Revkw (J anuaq 1991) pp. 80-99.
GEsson, C. A., Dependence of TechnologicaI Routlnbation on Structure Vatiables in Human Service
Organizations, Administrative Science Quurterfy (1978) pp. 383-395.
Govindarajan, V., Approprlateness of Accounting Data in Performance Evaluation: an Empirical ExamIna-
uon of FnvhonmentaI UncettaInty as an Intervening Varlable, Accountfng, Organfzuttons and Society
(1984) pp. 125-135.
Hage, J. T. % Alken, M., Routine Technology, Social Structure and Organiaational Goals, Adminfstratiue
scknce Quurte@y (1969) pp. 366-377.
Harrison, G. L., The Cross-cultural GenemIlzxblIlty of the Rehtlon Between Participation, Budget Empha-
sis and Job-related Attitudes, Accounting, OrganLzathns and Society (1992) pp. 1-15.
Harrison, G. L., Reliance on Accounting Performance Measure s In Superior Evaluative Style - the
Influence of NatlonaI Culture and Personality, Accounttng, Orgunizutfons and Sodety (1993)
pp. 319-339.
380 CHONG M. LAU et al.
Heneman, H. G., Comparisons of Self and Superior Ratings of Managerial Performance, J ournal of
Applied pSrcbologv (October 1974) pp. 638-642.
Hirst, M. K., Some Further Evidence on the Effects of Budget Use and Budget Participation on Managerial
Performance, Australian J ournal of Management (1987) pp. 49-56.
Hit, M. K. & Lowy, S. M., The Linear Additive and Interactive Effects of Budgetary Goals Di&uky and
Feedback on Performance, Accounting, Organizations and Society (1990) pp. 425-436.
Hofstede, G. H., Culture’s Consequences: I nkrnational Differences in Work-related Values (Eleverly
Hills: Sage, 1980).
Hofstede, G. H., The Cultural Relativity of Organizational Practices and Theories, J ouwual of I nter-
national Business Studies (Fall 1983) pp. 75-89.
Hofstede, G. H., The Cultural Relativity of the Quality of Life Concepts, Academy ofManagement Review
(1984) pp. 389-398.
Hopwood, A. G., An Empirical Study of the Role of Accounting Data in Performance Evaluation, J ournal
ofAccounting Research (1972) pp. 156-182 (Suppl.).
Hwang, S. K., The Effect of Culture on the Relationship Between Participation and Motivation, Ph.D.
thesis, Macquarie University (1989).
Jaccard, J., Turrisi, R. & Wan, C. K., I nkruction Effects in Multiple Regression (Sage University Press,
1990).
Kahn, R. L., Wolfe, D. M., Quinn, R. P., Snoek, J. D. & Rosenthal, R., Cbgunizutional Stress: Studies in
Role Conflict and Ambiguity (New York: John Wiley, 1%4).
Kerlinger, F. N. & Pedhazur, E. J., Multiple Regression in Behavioral Research (Iiolt, Rinehart and
Winston, 1973).
Kompass: Register of I ndustry and Commerce of Singapore 1992 (Singapore: Kompass South East Asia,
1992).
Kren, L. t Liao, W. M., The Role of Accounting Information in the Control of Organizations: a Review of
the Evidence, J ournal of Accounting Literature (1988) pp. 280-309.
Lindsay, R. M. & Ehrenberg, S. C., The Design of Replicated Studies, 7&e Americun Stutisticiun (August
1993) pp. 217-228.
Lynch, B. P., An Empirical Assessment of Perrow’s Technology Construct, Administrative Science
Quarterly (1974) pp. 338-356.
Mahoney, T. A., Jerdee, T. H. & Carroll, S. J., Development of Managerial Performance: a Research
Approach (Cincionati: South Western Publishing, 1963).
Mahoney, T. A., Jetdee, T. H. & Carroll, S. J., ‘Ihe Jobs of Management, I ndustrial Relarions (February
1965) pp. 97-l 10.
Manzoni, J. F., Use of Quantitative Feedbqck by Superiors of Manufacturing Cost Centre Managers: a
Methodological and Empirical Update, Second European Management Control Symposium auly 1992).
Mia, L., Participation in Budgetary Decision Making, Task DifXculty, Locus of Control and Employee
Behaviour: an Empirical Study, Decision Science (Fall 1987) pp. 547-561.
Mia, L., The Impact of Participation in Budgeting and Job Difficulty on Managerial Performance and Work
Motivation: a Research Note, Accounting, Organizations and Society (1989) pp. 347-357.
Milani, K. W., The Relationship of Participation in Budget-setting to Industial Supervisor Performance
and Attitudes: a Field Study, Z&e Accounting Review (April 1975) pp. 274-284.
Oppenheim, A. N., Questionnaim Design and Attitude Measurement mew York: Basic Books, 1966).
Otley, D. T., Budget Use and Managerial Performance, J ournal of Accounting Research (Spring 1978)
pp. 122-149.
Otley, D. T., Hannakis, L. & Lindsay, R. M., Intluence in Budgeting, Locus of Control and organizational
Effectiveness: Cultural Differences, Accountfng and Business Review Uanuary 1994) pp. 29-42.
Perrow, C., A Framework for the Comparative Analysis of Organisations, Ameriuan Sociologicul Review
(1%7) pp. 194-208.
Perrow, C., Organizational Analysis: a Sociologicul View (Belmont, CA: Wadsworth, 1970).
Peter, P. J., Reliability: a Review of Psychometric Basics at&Recent Marketing Practices, J ournal of
Marketing Research (Febnuuy 1979) pp. 6-17.
Prien, E. P. & Liske, R. E., Assessment of Higher-level Personnel: III: a Comparative Analysis of Supervisor
ratings and Incumbent Sewratings of Job Performance, Personnel Psychology