Management Accounting Research, 1990, 1, 181- 197

The impact of environmental uncertainty,

managerial autonomy and size on budget
characteristics
Mahmoud Ezzamel*

This study provides the results of an empirical investigation into the impact of various
contextual variables on the design of the corporate budgeting system. The results
support earlier findings linking contextual variables to budgetary characteristics.
Further, the results indicate that compared with corporate context variables as
measured by managerial autonomy and size, the external variable represented by
perceived environmental uncertainty (PEU) had a greater impact on the design of
budget characteristics. The results also suggest that the impact of PEU was much
stronger for larger size companies compared with smaller size companies.
Key words: environmental uncertainty; managerial autonomy; size; budget
characteristics.

Budgets are typically viewed as formal authmizations to the budget-holders to expend

given amounts of resources on some specified activities. A traditional rationalistic view
of budgets, or what Argyris [l] calls the theories of control that practitioners espouse,
would ascribe the following roles to budgets: planning, reduction of uncertainty,
communication of information, motivation, and performance evaluation [2]. In this
conventional sense, budgets could be viewed as 'implicit contracts' in an agency
framework with the focus being on drawing optimal contracts which maximize the
organization's objective function.
The design of the corporate budgeting system has been investigated by many
researchers. Much of the research conducted to date focusses on lower levels of analysis
within the organization, usually individuals, and has several strands of emphasis [3].
One strand deals with the impact of budgetary characteristics on managerial behaviour
and attitudes [4-61. Another strand uses managerial-related variables, such as
leadership style and personality, to explain variations in budgetary characteristics e.g.
17-91. Despite the useful insights provided by that research, the results have been
difficult to integrate, and have also been conflicting, possibly because the impact of
higher-level variables, such as environmental uncertainty and corporate characteristics,
has been ignored [lo].
This limitation has led to the emergence of another strand of research in which the
*The Manchester School of Management, UMIST, P.O. Box 88, Manchester M60 IQD, U.K.
Received 15 May 1990; accepted 9 September 1990.

10455-5005/90/010181+ 17 $03.00/0 0 1990 Academic Press Limited

182 M. Ezzamel
impact of higher-level variables on budgetary characteristics has been explicitly
examined [lo-151. The results of these, and other similar, studies have illustrated the
usefulness of contingency theory in establishing various linkages between budget
system design and these contextual variables. But these results do not always converge,
thereby making it necessary to engage in further studies seeking to replicate and extend
earlier research. This provides one of the motivations for the present paper. Further,
most of these studies focussed on higher-level variables related to the corporate context
such as size, diversity, degree of decentralization, and technology thereby excluding
the potentially important role of external variables, such as environmental uncertainty,
in influencing the design of the corporate budgeting system. A second motivation for
this study is to examine the effects of environmental uncertainty vis-a-vis those of the
corporate context on budgetary characteristics.
Two sets of explanatory variables have been selected for empirical investigation:
(i) external variable: perceived environmental uncertainty (PEU); (ii) corporate con-
text variables: managerial autonomy, and organization size. The budgetary characteris-
tics selected are (1) participation in setting budgets, (2) budget goal difficulty,
(3) budget evaluation, (4) required explanation of budget variances and (5) inter-
actions between subunit managers and their superiors. In this paper, the unit of
analysis is the organization as a whole rather than its constituent subunits. The
framework for the study is presented in Figure 1.
The following section of the paper examines the experimental variables used in the
study, and develops the research hypotheses. This is followed by discussions of the
research method, the results, implications of the results and conclusions.

Experimental variables and hypotheses

This section contains a discussion of the independent contextual variables and the
dependent variables. A number of hypotheses for empirical investigation are derived.
The hypotheses indicate the potential effects of the contextual variables on budget
characteristics.

The contextual variables

This study identified perceived environmental uncertainty (PEU), organization size,
and managerial autonomy (organization structure), as the contextual variables in the
study of budget characteristics.
The results of previous research indicate that the design and use of budgets and
other accounting mechanisms are strongly related to the extent of PEU [16-211.

Contextual variables Budget characteristics

External variable:
(PEW
Corporate context variables:
(organization size;
managerial autonomy)
- 1. Participation
2. Goal difficulty
3. Evaluation
4. Required explanation
5. Interactions with superiors

Figure 1. A contingency model of the effects of corporate context on budget characteristics.

 Budget Characteristics 183

Researchers have also noted that the characteristics of information sought by

decision-makers are contingent upon the level of PEU [22, 231. Greater PEU implies
greater difficulty in predicting future events. This renders planning and control
activities more problematic [24, 251. For example, greater environmental instability
might require frequent revisions in budgets and standards.
Organization size has been identified by various researchers as a key contextual
variable with a strong effect on organization structure [26-291. Larger organizations
tend to be more specialized, more standardized and more formalized compared with
smaller organizations. Size causes structuring through its effect on intervening variables
such as the frequency of decisions and social control. As the scale of operations
increases so the frequency of recurrent activities and repetitive decisions increases, and
hence, they are standarized and formalized. The effects of organization size on the
design of accounting systems and on budget characteristics have been examined by
Khandwalla [18, 291, Bruns and Waterhouse [12] and Merchant [lo].
Managerial autonomy refers to the extent of decision making power delegated to the
managers of major subunits in the organization (i.e. divisions or departments). Local
managers in decentralized organizations have greater responsibility and power over
planning and control activities and greater access to information not available to
headquarters [30]. The importance of managerial autonomy as a contextual variable in
the study of accounting variables has been stressed by many researchers. Lorsch [31]
and Watson [32] suggest that accounting systems (e.g. budgets) should be consistent
with structural arrangements. More recently, accounting researchers examined the
effects of managerial autonomy on various accounting systems and information
characteristics [lo, 23, 331.

Budget characteristics
Five characteristics of the corporate budgeting system have been selected as the
dependent variables in this study (see Figure 1). These characteristics were found to be
important by previous researchers [5, 6, 10, 12, 341. Other potentially important
budget characteristics such as budget incentive plans are outside the scope of this
paper. A summary of the proposed effects of contextual variables on each budget
characteristic is provided in Figure 2. Each of these characteristics is examined below.

Budget HI Hz H3
characteristics Environmental Organization Managerial
uncertainty size autonomy

Participation + + +
Goal difficulty + NR -
Evaluation + + NR
Required + + NR
explanation
interactions + - -
with superiors
NR = not relevant to that characteristic.
Figure 2. Proposed direct effects of contextual variables on budgetary characteristics.
Budget participation. Budget participation refers to the extent to which managers
engage in setting the budgets of their subunits. Bruns and Waterhouse [12] found that
budgetary participation was related to participation in other decision-making areas,
and that large organizations, presumably with more complex technologies, tend to
allow greater participation in budgets. They have also implied that managers in
decentralized organizations and in organizations facing highly uncertain contingencies
tend to perceive themselves as having high levels of budget participation, and as
appearing more satisfied with budget-related activities. The reverse was observed in
centralized organizations and in organizations operating in the face of lower levels of
uncertainty where the relative lack of emphasis on participation was perceived as being
more appropriate.
Similarly, Merchant [lo] noted that managers in the larger, more diverse,
decentralized companies participated more highly in budget setting. Participation by
middle and lower managers in budget-related activities is consistent with the spirit of
decentralization. In organizations characterized by large size, decentralized structure,
and greater PEU, middle and lower-level managers are likely to be better informed
about their local, specialized conditions. Hence, their involvement in budget prepara-
tion is likely to yield more realistic budgets and provide positive motivational
inducements [lo].

Goal difficulty.' Budget goals can vary in difficulty ranging from loose and easily
attainable to very tight and unattainable. According to Locke et al. [35], p. 127, 'goal
difficulty specifies a certain level of task proficiency measured against a standard'.
Compared with easy goals, the attainment of difficult goals requires the expending of
greater effort, and possibly the possession of more knowledge and skills. Several
researchers have investigated the effects of budget goal difficulty on employee
motivation e.g. Kenis [6], and on task performance e.g. Hofstede [16], Lock et al. [35],
Stedry [36] but few have considered the effects of contextual variables on goal
difficulty.
PEU makes managerial planning and control problematic [24, 25, 391. Under high
uncertainty future events become highly unpredictable and this could render budget
goals inappropriate. Hirst [37] has argued that task uncertainty is likely to be negatively
correlated with completeness of task knowledge and this could make budget goals
difficult to attain. Similarly, PEU is likely to be negatively correlated with completeness
of task knowledge and this could lead to difficult goals. It is also possible that in
situations of high PEU budget goals are deliberately made difficult in order to improve
performance. Various studies have shown that perceived budget goal difficulty and
performance are strongly related; the main ambiguity being whether or not perfor-
mance deteriorates after a certain degree of goal difficulty. Several studies [16, 381 have
suggested a bell shaped relationship between goal difficulty and performance, where up
to a certain degree goal difficulty leads to higher performance, but thereafter to lower
performance, see also Hopwood [9]. However, in a review of over 100 studies in the
areas of cognitive psychology and organization theory Locke et al. [35] found that the

'Budget goal difficulty emerged as one of the factors summarizing budget related behaviour in several studies
e.g. [6, 121, but this was not the case in Merchant's [lo] study.
 Budget Characteristics 185
evidence overwhelmingly indicated that difficult goals produced better performance
than medium or easy goals. Further, the impact of goal difficulty on performance is
likely to be moderated by other variables such as personality factors and aspiration
levels. In a laboratory study, Stedry [36] found that performance was positively
associated with budget difficulty for subjects who set their aspiration levels after
receiving the budget, thereby suggesting that in this case aspiration levels reflect the
performance objectives contained in the budget.
Previous evidence on the relationship between managerial autonomy and budget goal
difficulty is not straightforward. Bruns and Waterhouse [12] found that difficulty in
meeting the budget was positively, but not significantly, correlated with structuring of
activities, centralization of authority, and complexity of the control system. Onsi [ l l ]
and Cammann [40] reported that where managers participated actively in setting the
budgets, they had less motive to build slack presumably because they succeeded in
making their budget goals attainable. In contrast, they reported a positive relationship
between an authoritarian control system, where managerial autonomy is limited, and
budgetary slack. In the latter case, budget goals will be difficult to attain if subunit
managers are unable to build sufficient slack into the budget. It is thus plausible to
suggest that budget goal difficulty is lower in organizations with high managerial
autonomy compared with those with low autonomy. Greater managerial autonomy
could imply greater participation in budget setting and this enables managers to make
their budget goals attainable.

Budgetary evaluation. Budgetary evaluation refers to the extent to which subunit

managers are made responsible for budget variances in the evaluation of their
performance by top management. Budgetary evaluation can be general in the sense that
performance variances of subordinates are brought to their attention during perfor-
mance evaluation interviews and pay review with senior management. It can also be
punitive in the sense of causing manifest expressions of dissatisfaction from senior
management, possibly leading to lower motivation and unfavourable attitudes or to
favourable attitudes and behaviour [6].
Bruns and Waterhouse [12] found that budgetary evaluation was positively correlated
with structuring of activities and control system complexity, but negatively correlated
with concentration of authority (centralization and lack of autonomy'). Otley [21]
observed that in 'tough' operating environments, where making a profit is difficult,
upper management used budgets to pressure lower-level managers. In 'liberal'
operating environments upper management used budgetary information more for
problem-solving purposes. Further, Khandwalla [17] found that the intensity and type
of competition increase the need to monitor the results of organizational subunits, and
lead to greater use of sophisticated controls to achieve integration across subunits and
provide organization-wide norms of performance. Khandwalla [ 18, 291 also found that
the use of well developed budgets and other sophisticated controls was positively
correlated with environmental complexity and heterogeneity, size, and devolution of
decision-making .
These findings can be rationalized as follows. Small, highly centralized organizations

'In the Bruns and Waterhouse ([12], p. 195) study, lack of autonomy was defined as 'the extent to which
decisions are made outside the organization under study'.
can be controlled by informal, interpersonal mechanisms. In contrast, control in large,
decentralized organizations operating in highly uncertain environments is problematic
because they experience more complex communication and coordination problems.
Such organizations tend to use administrative control mechanisms which lead to
increased structuring of activities, greater formalization of communication channels,
and more reliance on standardized evaluation schemes such as budgets [12, 411.
Budgets are useful in communicating the intentions of various subunits to each other.
If budget targets are treated by subunits as organizational commitments, deviations
from budgets will be minimized and the amount of communication required to
maintain coordination will be reduced [lo, 421.

Required explanation of variances. This characteristic refers to the extent to which

subunit managers are required to explain variances from budgets and to report actions
taken to correct the causes of variances. Judicious performance monitoring necessitates
the separation of controllable and non-controllable variances. In situations of high PEU
actual results seldom conform to budget targets not only because of the level of
managerial competence but also as a result of the unpredictability of environmental
changes. In these situations, elaborate and sophisticated accounting and reporting
systems are employed [42] and detailed written explanations of causes of variances
together with progress reports of corrective actions taken are typically required. Bruns
and Waterhouse [12] found no significant relationship between required explanation
and concentration of authority (centralization and lack of autonomy). Merchant [lo]
reported significantly positive correlations between required explanation of variances
and organization size and diversity.

Interactions with superiors. This characteristic refers to the extent of interpersonal

interactions with superiors in the context of budget preparation and achievement. In
general, as PEU increases, particularly because of greater market diversification and
complexity, the knowledge of top management becomes less appropriate and
insufficient for centralized decision-making and direct, informal interpersonal com-
munications with lower-level managers. There will therefore be a tendency to rely
more on formal communication channels such as budgets and other formal reports.
However, situations of greater environmental dynamism, unpredictability, and hostility
call for action which is both timely and consistent with overall organizational interests.
In these situations, both lateral and hierarchical communication channels will be used.
Although many of these channels will be elaborate, regular, and highly formalized,
there will be a greater need for informal meetings, free consultations, and spontaneous
cooperation and discussions in order to deal effectively with fluid situations [41, 431.
Bruns and Waterhouse [12] found that interpersonal interactions were extensively
used in smaller or centralized organizations. Budget-related matters, including specific
means of budget attainment figured prominently in superior-subordinate interactions
(see also Macintosh [44]). They also found that lack of autonomy was negatively
correlated with superior-subordinate interactions during periods of budget preparation
and during discussions of the ways in which budget targets can be attained. In contrast
they found that large, decentralized organizations rely on formalized and standard
operating procedures and rules to organize work relationships. Similarly, Merchant
[lo] found that infrequent interactions with superiors was positively correlated with
 Budget Characteristics 187
size. Smaller, centralized organizations tend to be more homogeneous and communica-
tions tend to be based on informal interpersonal interactions. He observed that as
organizations grow larger and become more decentralized, communication is achieved
through the use of more formalized, indirect channels, structuring of activities,
standardized information and well-developed budgeting systems. Hence, these findings
suggest that formality of communication is likely to increase with size and extent of
decentralization.
The above discussion leads to the following three hypotheses
H, PEU will be positively associated with budget participation, goal difficulty,
budgetary evaluation, required explanation of variances, and interactions with
superiors.
H, Organization size will be positively associated with budget participation, budget
evaluation, and required explanation of variances, but negatively associated with
interactions with superiors.
H, Managerial autonomy will be positively associated with budget participation, but
negatively associated with goal difficulty and interactions with superiors.

Research method
Sample
Data was collected by administering a mailed questionnaire to the financial directors1 of
186 U.K. companies selected from the Times 1000. Of these 81 companies (44%) sent
complete responses. The unit of analysis was the organization as a whole. Respondents
covered a great variety of industrial classifications thus offering the potential for a wide
range for each of the experimental variables. The replies received were subjected to the
usual tests for randomness compared with the total sample and no discernible
differences were observed.

Tht measurements
In order to enhance the comparability and reliability of the findings of this study, the
questions contained in the questionnaire were developed from existing instruments.

The contextual variables

PEU was measured using a 23-item question developed by Miles and Snow [45], based
on the work of Dill [46]. The items cover suppliers actions ( a = 0.77), competitors
actions ( a = 0.78), customers demands ( a = 0.63), availability of credit and interest
rate changes in financial markets ( a = 0.75), changes in laws or policies by government
regulatory bodies ( a = 0.91), and actions of labour unions ( a = 0.91). All values of the
Cronbach alpha statistic [47] were safely above the lower limits of acceptability which
are usually considered to be 0.50 or 0.60 [48]. Using a scale ranging from 1 to 7 each
respondent was asked to indicate the perceived level of uncertainty for each item for the
company as a whole, with increased score representing increased PEU. The 23 items
were factor analysed, and six factors (which explained 74% of the total variability in the
data belonging to PEU) were extracted and item loadings corresponded to the a prion'
classification used.
'Although the questionnaire was sent to the finance directors, they were asked to solicit answers on each
question from the relevant corporate manager. Later telephone communications betweeen the author and the
respondents confirmed this process.
188 M. Ezzamel
Managerial autonomy was measured by a 17-item question developed from earlier
research by Baumes [49], Tomkins [SO] and Ezzamel and Hilton [51, 521. The items
cover five decision-making areas: operating policy ( a = 0.82), financing policy
( a = 0.63), investment policy ( a = 0.53), internal audit policy ( a = 0.51), with the
alpha coefficient for each of these last two variables just above the lower limits of
acceptability, and finally corporate policy, e.g. influence over the selection and ranking
of corporate plans, (a= 0.91). For each item respondents had to indicate, on a scale
ranging from 1 to 5, the extent of autonomy granted by the centre to subunit managers,
with 1 representing 'no' autonomy and 5 representing 'full' autonomy. The 17 items
were factor analysed and five factors were extracted (which explained 72% of the total
variability of the data) and item loadings corresponded to the a primi classification
used.
Three measures of size were considered: turnover, capital employed, and number of
employees. The three measures were highly correlated with P < 0.001. The analysis
was performed using each of these variables as a proxy for size one at a time but the
results were virtually identical. In consistency with previous research (e.g. Merchant
[lo]), only those results obtained when the number of employees was used as a proxy
for size are reported.
Because of the multidimensional nature of PEU and managerial autonomy, it was
decided to compose a single index for each of these two variables using factor scores,
the latter being derived from a linear combination of the subject's response and factor
loadings from the questions loading on each factor. In this sense, factor scores are
treated as if they were raw scores representing those responses on the questions which
exhibited high loadings on that factor, and should thus be considered as surrogates for
the intensity of the contextual variables in the statistical tests (see Haka [53]). All
relevant factor loadings were positive, indicating positive correlations with the
underlying original responses. Thus, the factor score for a given respondent with
respect to PEU would represent the degree to which that respondent perceived hislher
firm's PEU to be unpredictable, whereas the factor score for managerial autonomy
would reflect perceptions of the extent to which local managers have discretion in
decision-making.

Characteristic Items

Participation Control over budget setting

(a= 0.66) Changes included in budget
Influence on budget plans
Influence on finalized budget
Goal difficulty Difficulty in attaining budget
(a= 0.66) Effort required in meeting budget
Evaluation Importance of variances in evaluation
(a= 0.74) Importance of variances in pay raises
Superior dissatisfaction for exceeding budget
Adverse comments caused by variances
Required explanation Required explanation of variances
(a= 0.62) Required reporting of corrective actions
Interactions with Involvement with superiors in budget
superiors preparation
(a= 0.56) Involvement with superiors in budget problems

Figure 3. Budgetary characteristics (with Cronbach alpha's) and items.

 Budget Characteristics 189
Table 1
Descriptive statistics for the contextual and dependent variables

Independent/ Mean Standard Theoretical Actual

dependent variables deviation range range
(n = 81)

Contextual
variables:
Environmental
uncertainty
Size
Managerial
autonomy
Dependent
characteristics:
Participation
Goal difficulty
Evaluation
Required explanation
Interactions with
superiors

The dependent variables

The five budget characteristics were measured by a 14-item question. The items
represent a subset of the instrument developed by Swieringa and Moncur [7]. The
selection of these items was guided by the results of previous research. Each of the
items was found to be a significant budgetary variable by one or more researchers e.g.
[6, 10, 121. The score for each item ranged from 1 to 5. Figure 3 contains a list of the
specific items which correspond to each budgetary characteristic, together with the
Cronbach alpha statistic for internal reliability. The 14 items were factor analysed. Five

Table 2
K&l's tau intercorrelationsfor independent and dependent variables
- -

Independent/ Intercorrelations
dependent variables
(n = 81) 1 2 3 4 5

Independent variables:
1. Environmental
uncertainty
2. Size
3. Managerial autonomy
Dependent variables:
1. Participation
2: Goal difficulty
3. Evaluation
4. Explanation
5. Interactions
factors were extracted, which explained 66.4% of the total variability in the data and
the factors corresponded to the a prim' classification. Again, factor scores were derived
for each of the five budget characteristics.
Table 1 contains some descriptive statistics for each of the budget characteristics and
contextual variables.
Kendall's tau intercorrelations of the dependent and independent variables are
reported in Table 2. The correlations between the independent variables are
insignificant with the exception of the correlation between managerial autonomy and
size. In contrast, six out of the ten intercorrelations for the dependent variables were
found to be significant, which is consistent with the results of Merchant [lo] but not
with those of Kenis [6].

Tests and results

~ests '
The effects of the contextual variables on budget characteristics were tested by means of
calculating Kendall's tau correlation coefficients, as it has been suggested that
non-parametric tests are more suited to questionnaire data (e.g. Scapens and Sale [33]).
The results are reported in Table 3.

Results
The results presented in Table 3 strongly support H,, offer only weak support for H,,
and no support for H,.
As stated in H,, the results in Table 3 indicate that PEU is positively associated with
budget participation, budget evaluation, required explanation of variances, and
interactions between subordinates and superiors concerning budget preparation and
problems. The only discrepancy between the results and H, relates to budget goal

Table 3
Kendd's tau correlations between the dependent and independent variables (n = 81)
Dependent variables Independent variables
(budgetary characteristics)
Uncertainty Size Autonomy
HI H2 H3

Participation
Goal difficulty
Evaluation
Require explanation
Interactions with
superiors

'The underlying distributions of all the variables used in this study were tested for normality. The
histograms and various descriptive statistics (e.g. skewness; kurtosis) were examined. The X2 and the
Kolmogorov-Smirnov statistics were also computed. The hypothesis that the underlying distributions were
normal was only rejected in the case of managerial autonomy and PEU.Logarithmic transformations of both
these variables (and of size) did not result in any significant improvements in the results. As suggested in the
text, Kendall's tau was used in preference to parametric statistics and this represents yet another means of
dealing with non-normally distributed variables.
 Budget Characteristics 191
difficulty which did not exhibit a significant relationship with PEU as hypothesized.
Either the hypothesized relationship has been mis-specified, or the measurement
instrument was not sufficiently sensitive to proxy budget goal difficulty. Further
empirical work is needed to test this relationship further.
The results show only very weak support for H,, as interactions with superiors was
the only budget characteristic which exhibited a significant relationship with man-
agerial autonomy. As hypothesized, this variable was not relevant to budget evaluation
nor to required explanation of variances. But contrary to expectations, it was not
significantly associated with budget participation nor with budget goal difficulty.
Detailed correlations between managerial autonomy and the individual items of each of
these two characteristics showed only one significant relationship in the hypothesized
direction (budget achievement requires effort, P < 0.10). The results therefore suggest
that managerial autonomy, as defined here, has a limited impact on the design of the
corporate budgeting system.
The results relating to H, are disappointing, given the importance of organization
size in explaining variations in the characteristics of the corporate budgeting system
reported by previous studies e.g. [lo, 12, 18, 29, 411.
Taken together, the results support the notion that the organization's budgeting
system is related to the contextual variables facing the organization. This is consistent
with the results of previous research [lo, 12, 341. The results also indicate that PEU, as
a proxy for variables external to the organization, has a much stronger impact on the
design of the corporate budgeting system compared with the variables relating to the
corporate context, as proxied by managerial autonomy and size. But the analysis
conducted thus far fails to address several important questions. First, would the various
dimensions of both PEU and managerial autonomy have differential effects on budget
characteristics compared with those of the single scalar used for each of the two
contextual variables? Second why has the impact of organization size on budget
characteristics been so negligible compared with the results of previous research (e.g.
Merchant [lo])? Third, would the relative impact of PEU and managerial
autonomy on budget characteristics differ for populations of organizations of different
sizes? Fourth, would the presence of a significantly high correlation between
organization size and managerial autonomy bias the correlation coefficients between the
dependent and independent variables? These questions are addressed below.
First, Kendall's tau correlation coefficients were calculated between each budgetary
characteristic and each of the six dimensions of PEU and the five dimensions of
managerial autonomy using factor scores as described previously. In the case of PEU,
uncertainty with respect to suppliers, competitors and financial markets had by far the
strongest relationship with budget characteristics, with all significant correlations
exhibiting the hypothesized signs. In contrast, the impact of PEU in relation to
government regulatory bodies and actions of labour unions was muted, being
significant only in the case of interactions between superiors and subordinates. In the
case of managerial autonomy, the detailed correlations added little to the results
obtained for the aggregate scalar, thereby supporting the conjecture that managerial
autonomy had a marginal impact on the design of the corporate budgeting system.
Second, in order to test for the possibility that sample heterogeneity may have
dampened the effect of size on the dependent variables, correlation coefficients were
calculated for three broad industrial classifications: consumer g o o d d u r a b l e (n = 30),
consumer goods-nondurable (n = 36) and miscellaneous (n = 9). The results were
similar to those obtained earlier in the sense that size remained an unimportant
explanatory variable of budget characteristics, except in the case of budget participation
for the consumer goods-durable group and interactions with superiors for the
miscellaneous group. It should be noted, however, that Merchant's sample was drawn
from one industrial classification (electrical industry) and is therefore much more
homogeneous than each of the three subsamples used here.
Third, the sample was partitioned into two size groups: smaller (number of
employees less than 5000; n = 45), and larger (number of employees equal to or greater
than 5000; n = 36). This analysis yielded two interesting insights: (a) PEU and
managerial autonomy were significantly correlated with more budget characteristics in
the case of the larger size sample compared with the smaller size sample, and (b) the
correlation coefficient between managerial autonomy and goal difficulty for the larger
size sample was negative as hypothesized but was positive for the smaller size sample.
A plausible explanation of the first finding is that large firms tend to be diversified
(e.g. Merchant [lo]) and diversified firms tend to operate with higher PEU. Further,
organizations operating in the face of greater environmental instability, complexity and
diversity tend to allow subunit managers greater measures of autonomy [24, 54, 551.
Some support for this argument can be found in the data, where the mean score and
standard deviation for PEU for the larger group were ( x = 79.14; S.D. = 14.25)
compared with ( x = 75.69; S.D. = 14.06) for the smaller group (even though the t-test
for the difference in means is not statistically significant the difference has the right
sign). Similarly, the mean score and standard deviation for managerial autonomy for
the larger group were ( x =43.92; S.D. = 10.24) compared with (X = 37.53; S.D. =
10.45) for the smaller group with the t-test significant at the 5% level.
The positive correlation coefficient between goal difficulty and managerial autonomy
is more perplexing. This may be indicative either of pseudo autonomy or, alternatively,
of the inability of subunit managers to build sufficient slack into their budget targets.
Clearly, more research is needed to corroborate this result.
Fourth, partial correlation coefficients were calculated in order to assess the impact
on budget characteristics of controlling for the effects of managerial autonomy and size
one at a time. When managerial autonomy was controlled for, the only difference in the
results was that size was significantly positively correlated when required explanation of
budget variances (P < 0.10). No significant differences were observed, however, when
organization size was controlled for. It appears, therefore, that the impact of
multi-collinearity between managerial autonomy and size on budget characteristics was
insignificant.'

Discussion
This paper has sought to investigate the relationships between a number of budget
characteristics and some contextual variables. The framework for the study is presented
in Figure 1.
'To test further for the possible effects of interactions between managerial autonomy and sue, a new
explanatory variable, AUTOSIZE (autonomy x size) was constructed and correl~tionswith the dependent
variables were calculated, but none was significant. It should be noted that such composite variables are
difficult to interpret. Further, combining collinear variables together can violate the true specifications
linking the independent and dependent variables [68].
 Budget Characteristics 193
The results suggest that:
(i) PEU is positively correlated with budget participation, use of budgets for
performance evaluation, required explanation of variances and interactions with
superiors, but shows no significant relationship with budget goal difficulty.
(ii) Managerial autonomy is negatively correlated with interactions between superiors
and subordinates.
(iii) Organization size is not significantly correlated with any of the budget characteris-
tics studied.
Overall, this evidence is both plausible and consistent with the results of previous
research. First, the plausibility of the evidence is reflected in the notion that external
variables and corporate context variables impact on the design of the corporate
budgeting system. In designing their accounting systems organizations should therefore
take cognizance of both the internal and external contexts in which they operate.
Second, the results lend support to the findings of previous research e.g. [lo, 12, 341
which indicate that the characteristics of the corporate budgeting system are related to
the contextual variables of the organization. Managerial autonomy exhibited weaker
relationships with the dependent variables than hypothesized. The relationships
between size and the dependent variables were insignificant, as this study failed to
replicate the results of previous research relating to size. Of the three independent
variables, PEU exhibited the strongest relationships with budget characteristics. This
appeared to be particularly true for larger size firms compared with smaller size firms.
Further, the results lend some support to those who contend that environmental
variables have a most pronounced impact on organizational structure. In particular, the
results can be interpreted as supportive of the population ecology model according to
which the environment is viewed as a selective mechanism and organizations that
survive are perceived to be only those which are suited to environmental selection
criteria [56, 571. However, without a more exhaustive examination of the impact of
other contextual variables (e.g. technology; strategy) on the corporate budgeting system
the above conjecture has to remain tentative.
The designers of accounting and budgeting systems need to consider carefully the
implications of various higher-level variables, relating not only to the corporate context
but also to forces external to the organization, for the accounting systems they design.
However, these findings should be evaluated in the light of the limitations of the
study. First, the model used in this study (see Figure 1) is not complete. It could be
expanded to include other corporate contextual variables such as technology and
interdependence [14, 33, 541, diversity [lo], strategy [58, 591 and the controller's
personality [60]. Second, in common with questionnaire-based studies, this study was
limited to organizations which were willing to participate. Despite the good response
rate obtained, the results may have been affected by self-selection bias. Third, while
established measurement instruments were used in this study, they may still be
somewhat crude and this may have biased the results.

Conclusion
This paper examined the relationships between a number of budget characteristics and
various contextual variables. Despite the limitations of the model and measurement
instruments used, the results are useful. First, at a broad level the results support the
194 M. Ezzamel
findings of previous research which indicate that budget characteristics vary in relation
to the contextual variables with which the organization operates. Second, compared
with corporate contextual variables, PEU appears to have the strongest impact on the
design of the corporate budgeting system. Within the corporate context, managerial
autonomy appears to be more powerful in explaining variations in budget characteris-
tics compared with organization size, particularly in the case of larger size firms.
One limitation of the analysis undertaken here is that it emphasizes traditional
budget roles while ignoring other important roles. For example, the analysis did not
deal explicitly with the political-negotiation mode through which budgets are
'negotiated' between budgeters and budgeteers, given the asymmetric distribution of
power between these groups [61-641. Argyris [ l , 651 has called for broadening our
understanding of control by examining the theories espoused and the theories used by
practitioners; this he calls 'double-loop learning'.
It should also be noted that, in common with questionnaire-based research, the
present study has some potential shortcomings. These include the possibility of model
mis-specification, the problems associated with measurement of contextual variables,
and the exclusion of other, potentially important, variables.
To reduce these limitations, additional studies using more complete frameworks and
more refined measurement instruments are needed in order to improve the validity of
the results reported here. Future research could proceed at different levels of analysis
such as major divisions or strategic business units. Moreover, in order to control for
some of the potential biases and measurement errors of questionnaire-based studies
[66], 'softer methodologies' employing case study research could contribute sig-
nificantly to our understanding of these complex phenomena [67]. Not only would such
research help in trangulating the results reported here, but it would also add a new
dimension to our understanding of corporate budgeting practices by capturing the
dynamics of organizational processes [63, 641.
Acknowledgements: The author acknowledges the useful advice provided by the late Ken Hilton
and by Norman Macintosh at the design stage of this study, and by Chris Burke, Richard
Macve, Bob Scapens, Steve Ward and an anonymous reviewer on earlier drafts. The paper has
also benefitted from various comments made by the participants of the EAA Conference,
Stuttgart, March 1989, the International Conference on Research in Management Control
Systems, London, July 1989, and the Staff Seminar at the University of Warwick, December
1989. The author also acknowledges the financial support provided by the School of Business,
Queen's University, Canada.

References
1. Argyris, C. Organizational learning and management information systems, Accounting,
Organizations and Society, 2(2), 113- 123, 1977.
2. Bourn, M. and Ezzamel, M. Budgetary devolution in the National Health Service and
universities in the United Kingdom Financial Accountability &' Management, 3(1), 29-45,
1987.
3. Ezzamel, M. and Hart, H. Advanced Management Accounting: An Organizational Emphasis,
London, Cassell, 1987.
4. DeCoster, D. T. and Fertakis, J. P. Budget-induced pressure and its relationship to
supervisory behaviour. Journal of Accounting Research, Autumn, 237-246, 1968.
5. Milani, K. The relationship of participation in budget-setting to industrial supervisors
performance and attitudes: a field study, The Accounting Review, April, 274-284, 1975.
6. Kenis, I. Effects of budgetary goal characteristics on managerial attitudes and performance,
The Accounting Review, October, 707-721, 1979.
 Budget Characteristics 195
7. Swieringa, R. J. and Moncur, R. H. Some Effects of Participative Budgeting on Managerial
Behaviour, National Association of Accountants, 1975.
8. Hopwood, A. Leadership climate and the use of accounting data in performance evaluation,
The Accounting Review, July, 485-495, 1974.
9. Hopwood, A. Accounting and Human Behaviour, Englewood Cliffs, NJ: Prentice-Hall, 1974.
10. Merchant, K. A. The design of the corporate budgeting system: influence on managerial
behaviour and performance, The Accounting Review, October, 813-829, 1981.
11. Osni, M. Factor analysis of behavioural variables affecting budgetary slack, The Accounting
Review, July, 535-548, 1973.
12. Bruns, W. J. Jr. and Waterhouse, J. H. Budgetary control and organizational structure.
Journal of Accounting Research, Autumn, 177-203, 1975.
13. Dermer, J. Management Phnning and Control Systems, Homewood, ILL, Irwin, 1977.
14. Daft, R. L. and Macintosh, N. B. A tentative exploration into the amount and equivocality
of information processing in organizational work units. Administrative Science Quarterly 26,
207-244, 1981.
15. Macintosh, N. B. and Daft, R. L. Management control systems and departmental
interdependencies: an empirical study, Accounting, Organizations and Society, 12 (I),
213-227, 1987.
16. Hofstede, G. H. The Game of Budget Control, Assen, Van Gorcum, 1967.
17. Khandwalla, P. N. The effect of different types of competition on the use of management
controls, Journal of Accounting Research, Autumn, 275-285, 1972.
18. Khandwall, P. N. The Design of Organizations, San Diego, Harcourt, Brace, Jovanowich,
1977.
19. Hayes, D. C. The contingency theory of managerial accounting, The Accounting Review,
January, 22-39, 1977.
20. Arnigoni, E. Planning management control systems, Journal of Business Finance and
Accounting, Autumn, 279-291, 1978.
21. Otley, D. T. Budget use and managerial performance, J o u m l of Accounting Research,
Spring, 122-149, 1978.
22. Gordon, L. A. and Narayanan, V. K. Management accounting systems, perceived
environmental uncertainty and organization structure: an empirical investigation,
Accounting, Organizations and Society, No. 1, 33-47, 1984.
23. Chenhall, R. H. and Morris, D. The impact of structure, environment and interdependence
on the perceived usefulness of management accounting systems. The Accounting Review,
January, 16-35, 1986.
24. Lawrence, P. R. and Lorsch, J. W. Organization and Environment, Harvard University
Press, 1967.
25. Duncan, R. B. Characteristics of organizational environments and perceived environmental
uncertainty, Administrative Science Quarterly, September, 313-327, 1972.
26. Pugh, D. S., Hickson, D. J. and Hinnings, C. R. An empirical taxonomy of structures of
work organizations, Administrative Science Quarterly, March, 115- 126, 1969.
27. Blau, P. M. Decentralization in bureaucracies. In Zald, M. N. (ed) Power in Organizations,
pp. 150-174. Nashville, TN, Vanderbilt University Press, 1970.
28. Hall, R. H. Organizations:Structure and Process, Englewood Cliffs, NJ, Prentice-Hall, 1977.
29. Khandwalla, P. N. Mass output orientations of operation technology and organizational
structure. Administrative Scieme Quarterly, March, 74-97, 1974.
30. Waterhouse, J. H. and Tiessen, P. A contingency framework for management accounting
system research, Accounting, Organizations and Society, No. 1, 65-76, 1978.
31. Lorsch, J. W. Introduction to the structural design of organizations, In Dalton, G. W.
Lawrence, P. R. and Lorsch, J. W. (eds), Organizational Structure and Design, pp. 1-16,
Homewood, ILL, Irwin, 1970.
32. Watson, D. J. H. Contingency formulations of organizational structure: implications for
managerial accounting. In Livingstone, J. L. (ed), Managerial Accounting-The Behavioural
Foundations, pp. 65-80, Columbus, OH, Grid Inc., 1975.
33. Scapens, R. W. and Sale, J. T. An international study of accounting practices in
divisionalized companies and their associations with organizational variables. The Accounting
Review, April, 231-247, 1985.
1% M. Ezzamel
34. Collins, F. The interaction of budget characteristics and personality variables with budgetary
response attitudes. The Accounting Review, April, 324-355, 1978.
35. Locke, E. A., Shaw, K. N., Saari, L. M. and Latham, G. P. Goal setting and task
performance: 1969- 1980. Psychological Bulktin, 90(1), 125- 152, 1981.
36. Stedry, A. C. Budget Control and Cost Behaviour, Englewood Cliffs, NJ, Prentice-Hall,
1960.
37. Hirst, M. H. The effects of setting budget goals and task uncertainty on performance: a
theoretical analysis. The Accounting Revim, October, 774-784, 1987.
38. Stedry, A. C. and Kay, E. The effects of goal difficulty on performance: a field experiment.
Behavioural Science, 11(6), 459-470, 1966.
39. Burns, T. and Stalker, G. M. The Management of Innovation, London, Tavistock, 1968.
40. Cammann, C. Effects of the use of control systems. Accounting, Organizations and Society,
301-313, 1976.
41. Miller, D. and Friesen, P. H. Organizations: A Quantum Vim, Englewood Cliffs, NJ,
Prentice-Hall, 1984.
42. Galbraith, J. R. Organization Design, Reading, MA, Addison-Wesley, 1977.
43. Mintzberg, H. The Structuring of Organizations, Englewood Cliffs, NJ, Prentice-Hall, 1979.
44. Macintosh, N. B. The Social Sofmare of Accounting and Znfmtion System, New York,
John Wiley, 1985.
45. Miles, R. E. and Snow, C. C. Organizational Strategy, Structure and Process, New York,
McGraw-Hill, 1978.
46. Dill, W. R. Environment as an influence on managerial autonomy. Administrative Science
Quarterly, 2, March, 404-443, 1958.
47. Cronbach, L. J. Coefficient alpha and the internal structure of tests. Psychometrika,
September, 297-334, 1951.
48. Nunnally, J. C. Psychomenic Themy, New York, McGraw-Hill, 1967.
49. Baumes, C. G. Divisional Financial Executives, Studies in Business Policy No. 101 (National
Industrial Conference Board), 1961.
50. Tomkins, C. Financial Planning in Divisionalised Companies, London, Haymarket Publish-
ing, 1973.
51. Ezzamel, M. A. and Hilton, K. Divisionalization in British industry: a preliminary study.
Accounting and Business Research, Spring, 197-214, 1980.
52. Ezzamel, M. A. and Hilton, K. Can divisional discretion be measured? Journal of Business
Finance and Accounting, Summer, 31 1-329, 1980.
53. Haka, S. F. Capital budgeting techniques and firm specific contingencies: a correlational
analysis, Accounting, Organizations and Society, 12(1), 31-48, 1987.
54. Thompson, J. C. Organizations in Action, New York, McGraw-Hill, 1967.
55. Galbraith, J. R. Designing Complex Organizations, Reading, MA, Addison-Wesley, 1973.
56. Hannan, M. T. and Freeman, J. H. The population ecology of organizations, American
Journal of Sociology, March, 929-964, 1977.
57. Aldrich, H. Organizations and Environments, Englewood Cliffs, NJ, Prentice-Hall, 1979.
58. Simons, R. Accounting control systems and business strategy: an empirical analysis.
Accounting, Organizations and Society, 12, 357-374, 1987.
59. Ezzamel, M. The impact of organization strategy on budget characteristics and corporate
performance, Working Paper, The University of Manchester Institute of Science and Technology,
1990.
60. Brownell, P. Participation in budgeting, locus of control and organizatonal effectiveness. The
Accounting Rarim, October, 844-860, 1981.
61. Wildavsky, A. Budgeting: A Comparative Themy of Budgeting Process, Boston, MA, Little
Brown, 1975.
62. Bourn, M. and Ezzamel, M. Organizational cultures in hospitals in the national health
service. Financial Accountability and Management, 2(3), 203-225, 1986.
63. Ezzamel, M. and Bourn, M. The roles of accounting information systems in an organiza-
tion experiencing financial crisis. Accounting, Organizations and Society, 15(5), 399-424,
1990.
64. Ezzamel, M. and Bourn, M. Budget allocations in universities: incrementalism and
organizational change. Working Paper, The University of Manchester Institute of Science and
Technology, 1990.
 Budget Characteristics 197
65. Argyris, C. Single-loop and double-loop models in research on decision making,
Administrative Science Quarterly, 21, September, 363-375, 1976.
66. Kerlinger, F. N. Foundations of Behavioural Research, 2nd ed., Holt, Rinehart and Winston,
1973.
67. Kaplan, R. S. The evolution of management accounting. The Accounting Revim, July,
390-418, 1984.
68. Asher, H. B. Casual Mohling, 2nd ed., London, Sage, 1983.