Copyright
Peter Brownell.

Leadership behavior, budgetary participation and performance online

. (page 1 of 2)
Online LibraryPeter BrownellLeadership behavior, budgetary participation and performance → online text (page 1 of 2)
Font size
QR-code for this ebook


WORKING PAPER
ALFRED P. SLOAN SCHOOL OF MANAGEMENT



LEADERSHIP BEHAVIOR, BUDGETARY PARTICIPATION
AND PERFORMANCE

by

Peter Browne 11

and

Kenneth A. Merchant



WP 1174-80



November 1980



MASSACHUSETTS

INSTITUTE OF TECHNOLOGY

50 MEMORIAL DRIVE

CAMBRIDGE, MASSACHUSETTS 02139



LEADERSHIP BEHAVIOR, BUDGETARY PARTICIPATION
AND PERFORMANCE

by

Peter Brownell
and
Kenneth A. Merchant

WP 1174-80 November 1980



-1-

Leadership Behavior, Budgetary Participation
and Performance

INTRODUCTION

During the last two decades, a great deal of behavioral research in
managerial accounting has focused on the question of what constitutes
desirable design features of budgeting systems. The types of criterion
variables employed in studying this basic question have been as broad as
the array of design features studied. The effects of budgetary
participation on performance (Milani, 1975), goal difficulty on goal
commitment (Stedry and Kay, 1966), feedback on attitudes (Cook, 1967) and
acceptability of standards on felt cognitive dissonance (Foran and
DeCoster, 1974) represent just a few configurations of studied variables.

One particular area which has been studied in some depth is on the
effects of varying leadership style in budgetary settings. This area of
research developed out of one of the major conclusions of Argyris' (1952)
seminal work, namely that supervisors tend to employ budgets as means of
expressing their leadership style. Empirical advances along this line of
inquiry have been made possible by the development in the organizational
behavior field of measurement instrumentation for leadership style, most
notably the Leadership Behavior Descriptive Questionnaire (LBDQ) (Stogdill
and Coons, 1957).

While much has been learned from organizational behavior researchers
about the effects of varying leadership styles, the evidence from
accounting research on the question of the impact of leadership style in
budgetary settings has been mixed. This study is aimed, in part, at
clarifying some of the relationships which have been investigated with
limited success in the past.



nno^iq



-2-

PRIOR LITERATURE

An important, early budget-related leadership study was conducted by
Fertakis (1967) and reported on by DeCoster and Fertakis (1968). This
study aimed at investigating the impact on budget-related pressure of two
central dimensions from the LBDQ, consideration and structure initiation.
The results were surprising (at least to the authors) in that they revealed
that both dimensions were significantly and positively related to budget
pressure. It had been hypothesized that only structure initiation would be
so related.

A more recent study by Hopwood (1972, 1974) provided some insight into
DeCoster and Fertakis' result. Hopwood identified different combinations
of the two leadership style dimensions, consideration and structure
initiation, as representing what he called "profit conscious" and "budget
constrained", styles, as he studied the effect of leadership tendencies on
job-related tension. His most interesting result was that unbalanced
styles, that is high on one leadership dimension and low on the other, were
significantly related to the level of job tension. In particular, he
hypothesized that leadership style high on structure and low on
consideration would have a positive effect on job tension.

Although he did not test the question, Hopwood suggested that
leadership balance, via job tension, would probably have adverse effects on
performance. Otley, (1978) extended Hopwood 's work, and provided a test of
this specific question. However his results were contradictory to
Hopwood 's hypothesis, in that higher performance (in terms of budget
achievement) was associated with the tension producing, unbalanced
leadership style.



-3-

The results of both Hopwood's and Otley's studies are important
because they indicate that the most significant effects of a particular
leadership style dimension are most clearly uncovered when this dimension
interacts with, or is moderated by, a second dimension. Specifically,
Hopwood has suggested that unbalanced styles of leadership should be
avoided, although Otley's failure to confirm this result has left open the
question of the direction of the interactive effects. One purpose of this
paper is to provide further evidence on the question.

It should be pointed out at this juncture that the notion that certain
leadership style dimensions interact with other dimensions in their effect
on various criterion variables is not new. Many studies in organizational
behavior have demonstrated this phenomenon (e.g. Fleishman and Ko, 1973;
Misumi and Toshiaki , 1965; Beer 1966; Skinner, 1969; Fleishman and Peters,
1962; Hunt and Hill. 1971; Dessler, 1972, 1973). However, in the context
of managerial accounting generally, and budget system design specifically,
our knowledge of the effects of these interactions is rudimentary at best.

A second purpose of this study is to attempt to unravel the effects on
performance of budgetary participation. Past literature is in considerable
conflict on the question of whether budgetary participation has desirable
consequences or not. Results in the literature range from strong positive
associations between participation and performance (e.g., Bass and Leavitt,
1963; Kenis, 1979), little observable association (French, Israel and As,
1960; Milani, 1975), and, indeed, results which indicate a negative
association (Stedry, 1960; Bryan and Locke, 1967).

One possible explanation for this mixture of results stems from the
failure of researchers to give proper consideration to the "conditional
factors" (Hopwood, 1976) which determine the effects of participation. In



-li-
the context of leadership style, Argyris (1952) made the specific
suggestion that an avowed attempt at high levels of budgetary participation
would, in the presence of inappropriate leadership style, be suspected by
subordinates as manipulative. Argyris coined the term "pseudo-
participation" to describe the resulting perception. Empirical support for
Argyris' contention was provided by French et al . (1966), who showed that
where budgetary participation was provided solely to permit superiors to
employ the resulting budget as a more powerful sanction device, performance
suffered. French, et al . employed two "threat" conditions to

operationalize the superior's style of budget use. The results showed that
under a "high" threat condition, performance was better with lower levels
of participation, while "low" threat levels combined better with higher
levels of participation, as far as performance is concerned. These results
suggested that participation may have very important, but hitherto
unstudied, interactive effects with leadership style characterized by the
LBDQ dimensions discussed above. Thus, a second purpose of this study is
to provide evidence on this question.
HYPOTHESIS

The two separate research questions which provide the focus for this
study can be formalized into three hypotheses. The first question,
concerning the impact of balanced versus imbalanced leadership styles is
left entirely in its null form due to the existence of contradictory
evidence in the literature to date:

H • Leadership balance, defined in terms of the degree of similarity

of levels of structuring and considerate behavior (LBDQ), has no

significant effect on performance.



-5-

The second question concerns the role of budgetary participation as a
moderating influence on the impact of leadership balance. Quite apart from
the results obtained from testing H., this question will focus on the role
of budgetary participation as an influence on the strength of the observed
relationships studied in H.. Two separate hypotheses are formulated for
this second research issue; one concerning the direct effects of
participation on performance, and one concerning the interaction between
leadership balance and participation. As far as the former is concerned,
the mixture of evidence in the literature again calls fo>- a hypothesis
statement in null form:

H^: Budgetary participation has no significant effects on
performance .
On the latter issue , the interaction between participation and leadership
style, seme positive expectations can be offered. Following Argyris (1952)
and French et al . (1966), it is expected that a significant interaction
will be observed between leadership balance and budgetary participation
affecting performance. Specifically, it is expected that in the presence
of unbalanced leadership styles (that is, styles which are reported high on
one leadership dimension and low on the other ) will seriously impair the
effectiveness of budgetary participation. In its null form, the hypothesis
can be stated as follows:

H_: There is no significant interaction between leadership style
balance and budgetary participation affecting performance.
METHOD

Generally, substantial strengthening of research method is afforded
through an approach which employs multiple measures on variables with a
single set of respondents. Campbell and Fiske (1959) refer to this



-6-
approach as a "multi-method/mono-trait" structure, permitting an assessment
of the reliability and validity of the research measures employed.
However, such an approach does not overcome the problem of generalizability
which might be confined to the population represented by the single sample
employed. Thus, for this study, two quite different research samples were
used to provide the data to test the hypotheses.
Samples

The first sample consisted of forty-eight middle-level, cost-center
managers, involved in production and distribution functions, from a single
manufacturing company. Sample 2 consisted of 201 middle-level manufactur-
ing managers in 19 firms in the electronics industry. The number of
managers questioned in each firm of sample 2 varied from three to 21
depending on the organization's size and willingness to have their managers
spend time on the study.
Measures

The variables requiring measurement in this study are leadership
style, budgetary participation and performance. Leadership style was
measured by using the Ohio State Leadership scale revised LBDQ (Leadership
Behavior Description Questionnaire), Form XII (Stogdill, 1963). This
instrument uses 20 items to measure the subordinates' perceptions of
his/her superior's behavior along two dimensions:

(1) Consideration — of subordinates' ideas and feelings, and

(2) Initiating Structure — the extent to which the superior def fines

the subordinates' roles.

A sample item f-om the Consideration scale is: "He does little things to
make it pleasant tc be a member of his group," and for the Initiating
Structure scale is: "He lets subordinates know what is expected of them."
Each item is scaled from 1 (Always) to 5 (Never). The measures for each



-7-
dimension of leadership style are computed by summing the scores for the
appropriate 10 questions, and thus the theoretical range for each dimension
is 10-50.

A leadership balance variable was constructed from these scores.
Leadership style was considered balanced and coded +1 if the measures of a
particular superior's style were above the mean on both leadership
dimensions _or below the mean on both dimensions. A style was considered
unbalance, and coded -1, if one dimension scored above the mean, the other
below the mean .

Budgetary participation was measured by using different, but of course
related, instruments. In the first phase, a direct attempt at measurement
of participation was made through the use of Milani's (1975) 6-item, Likert
scale. A sample item is: "How often does your superior seek your
requests, opinions and/or suggestions when the budget is being set?" This
was scored on a scale of 1 (Never) to 7 (Very frequently).

In the second phase, participation was measured as part of a slightly
modified version of the budget-related behavior instrument developed by
Swieringa and Moncur (1975). This instrument provides a list of 44 related
activities and asks managers to indicate how often each takes place using a
scale from 1 (never) to 5 (always). These data were factor-analyzed, using
the SPSS (Stastical Package for the Social Sciences; Nie et . al, 1975)
inferred factor approach with oblique direct oblimin rotation (with 6=0).
Two budget participation factors emerged with eigenvalues greater than 1.0,
as reported in an earlier paper (Merchant, 1980), and these were labeled:
(1) influence on budget plans, and (2) personal involvement in budgeting.
The items loading en these two factors are presented in Table 1. The
factor scores for these two factors were used as the budgetary
participation variables.



-8-
Performance

The performance measure employed in Phase I was a self-rating of
overall performance. This measure was also used in phase II, but it was
supplemented by asking for ratings from the managers' immediate superiors.
The 201 managers in the Phase II sample reported to 131 different
superiors. Each of these superiors was asked to provide a rating of
overall performance and ratings in each of 15 areas of performance which
were shown by Mahcney (1967) and Mahoney and Frost (197*0 to provide useful
discriminations among organizational performances. The ratings in the 15
performance areas were then factor analyzed to reduce the dimensionality.
Two factors with eigenvalues greater than 1.0 were extracted, accounting
for 60.6? of the total variance. The factor loadings after oblique direct
oblimin rotation (with 6=0) and the percentage of variance explained by
each of the factors are shown in Table 2. The factor scores for these two
factors were used as measures in phase II, and they may be described as
follows :

1. Integration/Results : Extent to which the organization achieves high
performance and meets its commitments.

2. Adaptation : Extent to which the organization suggests and is able to
adapt to new ideas and directions.

Thus, phase II used four performance measures: (1) a self-rating of
overall performance, (2) an overall performance rating from the immediate
superior, (3) a superior-rating of integration/ results , and (4) a superior-
rating of adaptation.
Results
Hypothesis 1

Questionnaires were collected from all forty-eight managers in the
phase I study, a satisfying 100% response rate. However, two sets of data



-9-
were omitted due to improper completion of the performance measure, leaving
the sample size for testing of hypotheses at N=46. In phase II, responses
were received from 170 of the 201 manufacturing managers (85*), and 85 of
the 99 superiors (86?). This yielded a complete set of data, including
participation, leadership style, and self- and superior ratings of
performance for 131 managers out of the original sample of 201 (65%).

Hypothesis 1 was tested via examination of the Pearson correlation
coefficient between the relevant variables, leadership balance and
performance. The results are presented in Table 3. The results are in
partial conflict between phases I and II. In phase I, a significant
negative association emerged, indicating that performance was better under
unbalanced leadership conditions. In phase II, no relationship between
leadership balance and performance was uncovered, except for a significant
positive association with adaptability. This last result suggests that
managers maintain higher levels of adaptability under balanced leadership
styles.

The conflict between the results of the two phases of this study is
similar to a previously noted conflict between Hopwood (1972) and Otley
(1978). Hopwood suggested (but did not show) that poorer performance would
result from a leadership style high on structure but low on consideration
than from more balanced styles. This suggestion is partially in line with
the results of phase II of this study. However, Otley provided conflicting
results which showed that this leadership style was in fact preferred over
others in terms of performance. Otley's result is partially in line with
the results of phase I of this study.

To explore this question further, the effects of each leadership
dimension were examined individually under balanced and under unbalanced



-10-
conditions. Each of the two dimensions was correlated with performance
under each of the two balance conditions and the results are presented in
Table 4. The results of this analysis are interesting. In both phases,
consideration is seen to have a strong favorable impact on performance in
unbalanced conditions while structure has seriously adverse consequences in
the same circumstances. Where leadership style is balanced, however,
structure has either no impact (phase I) or a tendency towards a positive
impact (phase II). These results suggest that structure is resented (with
adverse performance consequences) unless accompanied by a commensurate
level of consideration, in which case structure appears effective. In
other words, the evidence suggests that the effects of each leadership
dimension are moderated by the level of the other. This result is
consistent with previous literature which has shown the existence of
interactions between leadership dimensions. Returning to the main thrust
of hypothesis 1, however, a conservative assessment of the evidence from
both phases suggests that the null hypothesis of no relationship between
balance and performance cannot be rejected. In interpreting this result,
however, the vastly different effects of consideration and structure on
performance in balanced versus unbalanced conditions are worthy of note.
Hypothesis 2

The second hypothesis concerned the direct effects of budget
participation on performance. Table 5 sets out the relevant results. The
results of both phases are consistent. Participation is postively
associated with performance (particularly the self rating version in both
phases). This result is consistent with some previous literature (e.g.
Kenis, 1979) and permits the rejection of hypothesis 2.



-11-

Hypothesis 3

Hypothesis 3 suggested the possibility of an interaction between
leadership balance and participation affecting performance. The results
(presented in Table 6) provide some support for the rejection of hypothesis
3. In phase I, participation has significant positive association with
performance only in the balanced leadership condition. Its effects in
unbalanced conditions still appear positive but are not statistically
significant. A similar result emerges in phase II, particularly on the
involvement dimension affecting the two performance dimensions —
efficiency and adaptability. Strong positive associations in the balanced
condition disappear in the unbalanced condition with the effect of
involvement on adaptability actually tending toward negative.

It should be noted, however, that the self rating measure of
performance in phase II shows an opposite result. That is, participation
shows stronger positive associations in the unbalanced condition. In the
absence of this last result, a tentative rejection of hypothesis 3 would
have been justified, but this conflict illustrates the danger of
generalizing from a single sample. The most likely explanation for these
results is an interaction with one or more contextual influences not
measured in this study.

As a further test for interactions, the correlation coefficients
across the balanced and unbalanced group were tested for differences. The
coefficients were normalized by a procedure outlined by Hoel (197^, p. 167)
and tested for significant differences. Table 7 presents the results of
this analysis. The results support the discussion above.



-12-
SUMMARY AND CONCLUSION'S

The results of this two-phase study show some consistent and some
inconsistent findings. Leadership balance, defined as leadership behavior
being high on both the consideration and initiating structure dimensions,
or low on both, was negatively related to performance in phase I, but a
tendency towards a positive association was noted in phase II. Some
reconciliation of this conflict was achieved by examining the effects of
each leadership dimension on performance in each of the two groups
(balanced and unbalanced). The results showed that initiating structure
was negatively related to performance of individuals reporting unbalanced
leadership styles on the part of their superior. Apparently, structuring
styles require a commensurate degree of considerate behavior in order to
avoid adverse consequences.

The basic conflict between the results of Phases I and II remain as a
curiosity, however. It is possible that interactions with other task-
related variables not studied may hold the explanation. The managers
forming the phase I sample were drawn from a company which faced a
relatively stable production technology while those in phase II were drawn
from a dynamic, fast-growth industry. Higher-level variables, such as
technological or environmental stability, may therefore also interact with
leadership styles.

In both phases, budgetary participation was found to be strongly
associated with performance. While some previous evidence (e.g. Eryan and
Locke, 1967) suggests a reverse relationship, the result is consistant with
the bulk of the evidence from previous literature.

The partial conflict noted between the results of phases I and II on
hypothesis 3 is related to the conflicting results noted on hypothesis 1.



-13-

In both phases of the study, participation was found to exert a stronger
positive influence in the presence of leadership conditions which were
problematic. To elaborate, there was strong evidence in phase I of the
study that performance was better under unbalanced leadership conditions.
Participation, when introduced to the analysis, was found to have strongest
positive effects in the balanced leadership condition. In phase two, where
some evidence of better performance was found in the balanced leadership
condition, participation was seen to positively influence performance more
in the unbalanced condition. In both cases, therefore, participation
tended to ameliorate the performance consequences of less effective
(balanced in phase I, unbalanced in phase II) leadership styles.

The explanation as to why the two phases of this study are in conflict
with regard to the direct effect of leadership balance must await research
results incorporating the effects of other, potentially interactive,
variables, as discussed above. For this same reason, it is necessary to
introduce some important caveats which should be borne in mind when
considering the consistent results of the two phases of this study.
Strictly interpreted, the results cannot be generalized beyond
manufacturing-related functional areas in U.S. corporations. It is not
obvious, for example, that the results can be generalized to other
functions such as marketing for, as Hayes (1977) pointed out, budgeting is
most relevant and important for control of the manufacturing function.
Further research must also address the issue of whether the results hold at
other managerial levels, such as vice-president, and indeed whether the
results hold at all in corporations in other countries. On this last
point, Hofstede (1967) concluded an extensive survey of budgeting practice
with the cross-cultural caveat that "the game of budget control (as he
describe it) is a Western game" (p. 281, parentheses added).



-14-

On the positive side, the strength and value of the results on which
phases I and II are in agreement should not be overlooked. Given the
vastly different industrial settings studied, and methods used to gather
data in each phase of this study, consistent results emerging in each phase
should be viewed as particularly strong since mutual validation of the
instruments used in each phase effectively results. This point is
particularly relevant in the case of the vastly differing approaches
employed to measure budgetary participation.

Future research, aimed at clarifying some of the conflicting results
of this study, is needed in much the same way as this study was motivated
by conflict from past research. Similarly, we should not lose sight of the
need to give careful thought to research designs which build in a basis for
assessment of empirical measures used. Without the study of both higher-
order variable interactions and research instrument validity, it will be
impossible to attribute conflicting results to the two possible causes of
higher-order interaction and "instrumentation-interaction".



-15-
Table 1


1

Online LibraryPeter BrownellLeadership behavior, budgetary participation and performance → online text (page 1 of 2)