HD28
.M414
i
ALFRED
P.
WORKING PAPER
SLOAN SCHOOL OF MANAGEMENT
THE MOTIVATIONAL IMPACT OF MA.NA&EfIENT-BY-EXCEPTION
IN A BUDGETMY CONTEXT
by
Peter Browne II
Dewey
THE MOTIVATIONAL IMPACT OF MANAGBIEKT-BY-EXCEPTION
IN A BUDGET/^ Y CONTEXT
by
Peter Browne 11
WP 1311-82
May 1982
ReviEed October 1982
'^AY 3 1 1983
REcavEo
1,0
Introduction
It
in
suggested
often
is
managerial
the
accounting
literature
that
control system effectiveness depends not only on what information the system
supplies to managers
responsible for exercising
control,
but
In the context of budgetary controls,
use the information.
managers
how
Argyris
(1952)
may have been the first to note that control system failure was more often
affected
style
the
by
information
of
characteristics of the system.
use
rather
than
the
by
technical
Subsequently, several empirical studies have
focused attention on this issue (e.g., DeCoster and Fertakis [1968]; Hopwood
Otley
[1972];
At
[1978]).
practical
the
level,
one
coimnon
style
of
information use is "management-by-exception" (MBE).
Exception-reporting is probably based on the notion that only variances,
unfavorable
both
particularly
if
and
they
should
favorable,
were
assessed
managerial
attract
as
being
significant.
attention,
MBE
Hence,
provides a basis for allocating, or rationing, managerial effort.
In
practice,
however,
such
systems
reduce
typically
preoccupation with unfavorable variances.
to
For example, Bittel
a
biased
(1964,
p.
5)
defined MBE as system of control which enables the manager to be "spared the
task of reviewing performance where things are going well (so as to) devote
his
attention
only
to
those
areas
which
his
require
really
managerial
Ronen and Livingstone (1975, p. 680) suggested that, under MBE,
attention."
"the response to favorable deviations not requiring corrective actions often
seems to be weaker than that to unfavorable deviations."
(1967,
emphasis
described
478)
p.
in
punisliment
the
and
characterized
feedback
reward,"
the
exception
is
on
while
phenomenon
reporting
Porter
one
a
rather
punishment
as
as
and
by
Birnberg and Nath
system
than
Zannotos
which
07453^1
in
some
(1978,
managers
"the
wliich
mixture
pp.
are
of
12-13)
given
successes."
for
than full credit
"full blame for failures and less
Itami
(1975) provided an analytical demonstration of how management, in the design
of
might
structures,
reward
subordinate
purposefully
asymmetrical
employ
rates of penalty and reward (for unfavorable and favorable budget variances,
respectively) so as to "modify" the subordinates' attitudes towards risk and
many budgetary control systems. "(p.
similar
like
and,
results
Baiman and Demski
85)
"Certainly
that
concluded
Itami,
obtained
(1980)
lovz-tail
the
conditional variance investigation system is often observed in practice
is
of
that "this is perhaps common practice in
He goes on to suggest
the latter.
part
behavior on the
likelihood of goal-congruent
the
thereby increase
probably the system which is most often used). "(p.
(It
parentheses in
194,
original)
Except
studies
for
these
that
is
this
last
results,
variant
of
the
assumption
general
MBE has adverse
other
the
in
behavioral consequences.
However, none offers any systematic empirical evidence of such effects.
The
purpose of this paper is to report the results of an empirical study of the
effects of MBE in a budgetary context on individual motivation, where MBE is
assumed to refer to
a
preoccupation with unfavorable variances.
I
(the
hypothesize that MBE will have adverse effects on motivation
hypothesis),
but
these
that
effects
will
mitigated
be
initially
level
the
by
general
of
participation in budget setting (the contingency hypothesis).
theoretical
As suggested by the variety of literature discussed above, a
framework
based on
economics
of
the
agency,
behavioral sciences,
were
possible
behavioral framework was chosen, but
the framework
as
candidates
well
for
as
the
later in the paper
one
based
present
I
study.
attempt
(and the niore important results of the study)
to
on
to
the
A
link
some of the
propositions which have emerged from the economic perspective.
In the next section of the paper,
2
I
develop the theoretical framework to
support
role
of
contingency
conclusions
then
preoccupied
follow
with
support
for
One
budget
unfavorable
contingency
the
variances.
MBE
results
The
but
hypothesis,
being
as
also
provide
the
general
for
not
and
strong
the
is
the
(the
results,
finding
of
on
MBE
of
method,
the
interesting
characterization
above
the
effects
the
dealing with
turn.
in
of
support
elaborates
following
moderating
in
Sections
hypothesis).
in
evidence
some
participation
budgetary
section
The
general hypothesis.
the
hypothesis.
A Theoretical Framev.'ork - The Expectancy Model of Human Motivation
2.0
and modification of behavior falls into two broad classes.
behaviorist
the
(e.g.,
maintenance
literature in psychology dealing with the acquisition,
Tlie
Hull
perspective,
[1929];
conceived
solely
theories
of
in
Spence
terms
classical
conceptualizations
characterstic
of
[1942]),
of
linking
the
and
in which
operant
learning
and
of
conditioning
motivation
v/hich
theorists
and motivation are
learning
stimuli
there is
First,
learning
early
of
and
responses.
The
well-known
are
two
have
derived
from
a
behaviorist tradition.
A second, more recent perspective can be referred to as the cognitivist
view.
Here, the intervening cognitions
the focus of attention.
(between stimulus and response) are
While this perspective can be traced back at least
as far as Tolraan [1932],
the more recent work of Rotter
Bandura
[1972]
[1962]
expectations,
[1965]
has
integrated
of
and
beliefs,
traditional
Rotter [1966] suggests that the occurrence of a particular
behavior is dependent not only on an individual's
(the
roles
and other cognitive processes into extensions of
learning theory.
history
the
[1966]
[1954],
behaviorist view),
but
also
on
the
objective
reinforcement
individual's
expectation
that the behavior will result in a particular consequence and the subjective
assessment of the costs and benefits associated v/ith the
view).
cognitivist
constructs
These
readily
are
consequence
recognizable
(the
the
in
which provides the theoretical
expectancy theory conception of motivation,
linkage between MBE and motivation underlying the
choice
of
criterion
the
variable used in this study.
with
MBE,
its
reinforcement
objective
on
stress
variances,
can
individual,
and
unfavorable
history
of
the
expectancies associated with any new behavior.
will suffer.
both
the
subjective
the
motivation
either case,
In
particular formulation of
seen by examining the
This can be
shape
the expectancy value model of motivation chosen for this study.
The formulation, based on theoretical work of Georgopoulos
and subjected to empirical test by Galbraith and
Cummings
,
et al
(1957)
can
(1967),
be
expressed as follows:
M =
IV,
b
+
P,
(IV
1
a
+ ZP„. EV.)
2i
(1)
1
where
M
= motivation to work.
IV
= intrinsic valence associated with successful performance of
the task
(goal-accomplishment).
IV,
= intrinsic valence associated
EV.
= extrinsic
with goal-directed behavior.
valence associated with
extrinsic
i
contingent
re\/ard
1
on work-goal accomplishment.
P^
= the
P„.
= the
expectancy
that
goal-directed
behavior
accomplish
will
the
work-goal (a given level of specified performance).
expectancy that work-goal accomplishment will
lead
to
the
i
extrinsic reward.
MBE
is
hypothesized
formulation,
since
to
those
adversely
P-'s
affect
the
F^i
values
in
associated with positively valent
the
above
outcomes.
Lower under MBE
or reinforcers, are expected to be assessed systematically
goal-accomplishment and
due to an absence of historical association between
the receipt of positive reinforcers.
although
might also suffer a deterioration under MBE,
IV
For similar reasons,
that
2uch as
dissonance-resolving process,
a
suggested
by
Calder
That is, individuals who cannot
and Staw [1975], may mitigate this effect.
rewards
attribute their goal-directed effort to the expectation of extrinsic
rationalize their
because few such rewards are forthcoming under MBE, might
IV^.
or
goal-accomplishment,
although
Therefore,
value
of
motivation
may
internalized
the
of
terms
in
effort
of
expenditure
overall
the role of intrinsic versus extrinsic rewards in contributing
not suffer,
The use of the expectancy theory model of
to overall motivation will shift.
motivation chosen for this study permits an examination of the association
betv.'een
MBE and specific elements of the model.
Finally,
perceive
the
reduced assessments
non-receipt
of
P^
extrinsic
of
may
occur
rewards
as
because
individuals
resulting
also
failure of the reinforcing agent to acknowledge goal-accomplishment
may
from
a
itself.
Such a phenomenon translates into reduced assessments of P^.
here again,
But,
assessments
viewed
as
being
as
.
a
a
function of
negative
IV
IV
of
a
simple
reduced assessments of
The
function
positive
Indeed,
P^.
negative
value
intrinsic
function
of
early
of
of
goal
by
increased
can
goal-accomplishment
difficulty,
conceptions
the
offset
P^ may be
of
probability
and,
hence,
be
a
motivation
viewed
success
(e.g.,
of
Atkinson [1958]).
To summarize,
assessments
of
MBE is hypothesized to impact motivation through reduced
the
P2.
and
compensating adjustments to IV
P^,
.
with
the
possibility,
in
both
cases,
of
Budi^etary Participation - The Contingency Hypothesis
3.0
The
effects
years
the
of
researchers
by
been widely
budgetary participation have
managerial
in
although
accounting,
from
range
examined over
not
with
positive
effects
of
participation on performance and job satisfaction (e.g., Kenis,
[1979]),
to
consistent
weak
positive
(e.g.,
The
results.
Bryan
Milani
(e.g.,
and
findings
Locke
and
[1975]),
[1967]).
strong
even
the results
Indeed,
[1982]).
effects
of
(Brovmell
studies provide a basis
some
of
effects
the
that
is
participation are contingent on a variety of moderating
[1981],
associations
negative
to
conclusion
The
for
predicting interactive effects of participation and MBE.
Cherrington and Cherrington [1973], in a laboratory experiment, examined
the interaction between budgetary
participation and reward structure.
The
latter was reflected in the degree of reliance placed on budgeted results in
performance
was
evidence
Their
evaluation.
mildly
support
in
of
the
hypothesis that both performance and job satisfaction would be higher when a
high reliance on budget information
v;as
accompanied by high participation.
One reason for this is that in providing the individual v/ith the opportunity
to influence budgeted
results,
management achieves more
legitimate grounds
for strong sanctions in the event of sub-standard budget-related performance.
In a similar type of investigation, Brownell [1982] examined the role of
budget
participation as
evaluative
styles.
Two
a
moderator
prior
of
effects
the
investigations
style had produced contradictory results
of
(Hopwood
of
the
different
impact
[1972]
superior
evaluative
of
and Otley
[1978]).
In his study, Brownell found results which supported the hypothesis that a
heavy
reliance
accounting
on
information,
and
on
budgets
in
particular,
would not adversely affect performance as long as this evaluative style was
accompanied
by
a
high
level
of
participation
in
budget
setting.
The
in
thread
common
both
studies
above
the
of
is
notion
the
that
participation in budget setting may alleviate otherwise adverse consequences
stemming from a style of use of accounting information which places a high
emphasis
budget
on
A
administration.
achievement
similar
effect
participation
of
and
evaluation
performance
in
is
reward
hypothesized
in
connection with the use of MBE.
Hypotheses and Method of Study
4.0
Hypotheses
4.1
The above can be summarized in the two hypotheses to be tested in this
In null form we have:
study.
There will be no relationship between MBE and motivation.
H
There will be no interaction between MBE and budgetary
H
2
participation affecting motivation.
Method
4.2
Data for the study were collected through a survey questionnaire to 224
- two in the
middle-level managers drawn from three separate corporations
electronics industry and one in the steel industry.
from
variety
a
functional
administration.
and
research,
of
fields
including
one
selection
The
The managers were drawn
production,
marketing,
criterion
budgeting.
activities of the managers should be controlled via the use of
However,
final
sample
selection
was
up
left
to
top
the
that
was
management
in
each
corporation and so the sample was not strictly random.
Of
the
224
questionnaires
which 122 were usable.
distributed,
The average age
140
were
returned
of respondents was
(62.5%),
37.0
of
years and
years.
their average tenure with their respective companies was 0.1
Measures
of
three
variables
were
motivation, MBE, and budget participation.
obtained
in
the
questionnaire:
Motivation
4.2.1.
The objective of the questionnaire items was to elicit measures on each
construct of the model presented in Equation
of
constructs
the
Equation
For expositional purposes,
1.
classified
are
1
into
basic
tv70
groups
-
valences, and instrumentalities/expectancies.
Valences:
valences
The
dV,
IV
,
approach
and
,
by Lawler and Suttle
to
(LS)
adapted
was
EV.)
measurement
the
(1973)
of
from
three
the
classes
procedure
the
of
developed
in such a way to distinguish between
the
Seventeen outcomes from the LS set were used in
three classes of valence.
the prese^nt study and these were a priori classified as either intrinsic (8)
or
extrinsic
outcome,
The
(9).
outcomes
listed
are
the
in
(extremely desirable to extremely undesirable/
nine
appendix.
respondents were twice asked to indicate on a scale
preference for that outcome.
First,
For
from
each
one
to
strength of their
the
respondents were asked to value each
outcome as it might result from "working hard" (goal-directed behavior), and
second, to value the outcomes as they might result from "meeting or beating
budgeted goals"
eight
IV
responses
(goal-accomplishment).
to
the
intrinsic
was measured by averaging
IV,
items
from
measured by averaging
was
the
first
set
of
the
responses.
t
from the second set of responses.
Two
points
should
be
noted
here.
First,
and
IV,
b
IV
are
a
incorporated into the expectancy model by way of a single average score, as
implied
by
the
model
presented
in
Equation
(1).
difficult to ascertain why theoretical discussions of
have
overlooked
the
issue
of
multiple
intrinsic
Nonetheless,
the
it
is
expectancy model
valences,
and
hence
possibility of their entering the formulation by way of a summation.
the
llie
two options (mean versus summation) are clearly different in that they place
different weights on the intrinsic versus extrinsic elements of the model.
Second, the use of the same eight intrinsic outcomes in the measurement
of
IV
either
and
presupposes
IV
goal-directed
that
behavior
^
given
any
outcome
could
and
goal-accomplishment
from
result
that
value
the
placed on the outcome (valence) might not be independent of its source.
Table
(after
1
presents the means and standard deviations of the eight outcomes
scale
adjustment)
independence of the
among
the
two
sets
tv/o
for
each
of
IV
The
.
degree
of
sets is reflected in the coefficients of correlation
(last
difference of mean tests.
column)
and
for
t-statistics
the
eight
paired
The results reveal that while a significant degree
INSERT TABLE
of association
and
IV,
exists between the
1
HERE
two classes
intrinsic valence,
of
their
respective mean scores were usually quite different.
Turning
to
a
consideration
of
the
EV
.
,
the
the nine extrinsic outcomes were assessed from the
(outcomes resulting from goal-accomplishment).
valences
s econd
Table
and standard deviations of the nine extrinsic valences.
INSERT TABLE
2
HERE
2
associated
set of
presents
with
responses
the
means
Instrumentalities/Expectancies:
of
nine
specific
outcome
extrinsic
instruraentalities
Equation
LS,
these
assessed by asking respondents to indicate on
assessment
the
goal-accomplishment
associating
Again following
(P„).
for
calls
(1)
instrumentalities
scale from one
a
seven (always), how often "meeting or beating the budget" would
with
were
(never)
to
result
in
Three additional items incorporated in this set of questions
each outcome.
elicited measures of
P,
,
the
probability
expectancy
or
that
For ease
behavior would result in goal-accomplishraent.
goal-directed
interpretation,
of
the twelve responses were converted to probabilities in the range of zero to
one and
in
Table
the
means and standard deviations
The
3.
three
assessments
of
of
were
P,
nine
the
P«
'
s
are
significantly
(p
presented
<
0.01)
).:
therefore derived by averaging the three responses.
and
The three correlations,
the
Table 3.
INSERT TABLE 3 HERE
The
final
individual
measure
constructs
of
motivation
following
was
Equation
obtained
1.
statistics on the final measure.
INSERT TABLE
10
4
HERE
Table
by
4
aggregating
reports
the
descriptive
Management-By-Exception (MBE)
4.2.2.
A review of previous literature uncovered no suitable empirical measure
study involved nine questionnaire
The measure developed for this
for MBE.
One item simply asked respondents to rate on a seven-point scale the
items.
frequency (one equals never, seven always) with which unfavorable variances
consisted
pairs
four
of
investigation
variances
of
similar
of
were
from
taken
items
Four
items.
the
relating
1975;
Sv/ieringa and Moncur,
superiors
discuss
budget
1975;
items
1968; Bruns
and Merchant,
with
the
to
Budget-Related
so-called
Behavior questionnaire (see, for example, DeCoster and Fertakis,
and Waterhouse,
items
The remaining eight
receive more attention than favorable variances.
An
1981).
when
me
variances
example
is
occur."
Each of four such items were included twice in the questionnaire,
"My
once for favorable variances, and once for unfavorable variances.
Since
which
the
are
definition of MBE used
pre-occupied
with
here
unfavorable
involves
variances,
managerial assessments
the
to
response
the
"favorable variance" version of each item was subtracted from the response
to
the
"unfavorable
difference scores.
variance"
version
the
means
for
the
same
item,
to
obtain
four
All scores were converted to probabilities, and Table
reports first the mean scores for
by
of
each
of
the
the single,
four
pairs
overall assessment,
of
specific
5
followed
items,
their
differences, and the significance of those differences (paired test).
INSFJIT TABLE 5 L-ERE
As noted in the introduction,
11
the
results
presented in
Table
5
wholly
support
empirical observation that MBE,
Che
defined
as
in
study,
this
is
All four of the pairwise comparisons among the
quite prevalent in practice.
eight specific questionnaire items
reveal a
significantly higher frequency
of the "budget-related behavior" as it relates to unfavorable variances than
as
the
behavior relates to favorable
variances.
interpreted,
Roughly
the
result for the single item addressing MBE, indicates that, on average, more
attention is given to unfavorable variances than to favorable variances 81%
Given the consistent pattern of results from the entire nine
of the time.
item instrument,
I
response to all nine.
To do so,
I
results
are
2
suggestion
(1980) who provided a convincing rationale
defined
here,
is
analytical result which suggests
top
(agents),
investigation strategy
they offer (p.
of
that,
v.'hich
of
Baiman
3
Demski
and
as
derived
an
they
mentioned,
for monitoring risk-averse managers
(principals)
management
would
do
well
employ
to
of
an
Beyond this,
focuses on unfavorable variances.
194) an appealing argument suggesting the greater likelihood
encountering risk-averse individuals among middle-level management
objects
the
for the prediction that MBE,
previously
As
prevalent.
MBE using
of
and then divided by five.
with a
consistent
measure
summed the differences between the four
paired items plus the single item score,
These
average
single
constructed a
monitoring
-
and
the
subjects
of
this
research)
than
(the
among
top-management (the monitors themselves).
4.2.3.
The
Budgetary Participation
participation
measures
developed
by
Milani
(1975)
and
Hofstede
(1967) were both employed in this study, as a crude means of validating one
another.
Previous use of both measures in a single study
(Brovmell,
1982)
revealed that the two measures correlated significantly (0.74).
The Milani measure is a six item Likert type
12
scale,
each
item
calling
for a response from one
construction
the
of
scale
The
seven.
to
score,
overall
and
designed for an additive
is
factor
performed
previously
a
analysis of the scale (Brownell, 1982) provides adequate confirmation of the
sin^ile factor nature of the measure.
The
measure
llofstede
calling for one response.
But
since
hypothesis
the
Table
6
only
at
r
=
of
0.59
were
participation
budgetary
concerning
tests
both measures.
correlated
measures
tv;o
the validation
because of
scale
single
In my previous use of both measures
tests
Milani's for hypothesis
anchored
fully
eight -point,
an
is
I
,
structure.
its
in
relied on
study,
this
performed
using
reports descriptive statistics for each.
INSERT TABLE
6
HERE
Results
5.0
H
The
:
first
between MBE and
motivation.
yielded a coefficient
not
proposed
hypothesis
r
The
= -0.09 which,
significant.
statistically
simple
an
negative
overall
between
correlation
relationship
the
measures
while in the predicted direction, is
Hence,
null
the
hypothesis
of
no
relationship cannot be rejected.
In order to further explore this
with each of
P_'s,
the
the
nine
eight
EVs
and
intrinsic
the
three
results are presented in Table 7.
13
result, MBE was correlated separately
valences
(both
components
used
IV^
to
,
the
nine
construct
P^^.
The
and
IV^^)
INSERT TABLE
Taken
individually,
in
coefficients
the
HERE
7
Table
fail
7
uncover
to
compensating effects of HBE on the separate components of the model.
jointly,
hov;ever,
do
results
the
propositions offered earlier.
MBE and each of the nine
expected
relationship
on
MBE
page 4).
(see
occurrence
this
less
is
existed
signed
IV,
v/as
eight are
,
A binomial test
than
and
of
of
significantly),
(one
a
or
EV.
no
particular
(both
contingent
relationship
positive
correlations
significant.
as
probability of
the
while
that
IV
eight
the
highly
is
also
theoretical
the
nine correlations betv;een
indicated that
either
evidence
Six
one
and
for
support
of
negative
Note
0.02.
some
found.
0,11)
<
(p
s
'
between MBE
goal-accomplishment),
and
P„
some
offer
In particular,
any
Taken
This
are
betv/een
positively
result
provides
tentative support for a slight variation of the dissonance-resolving process
discussed earlier.
There I suggested that MBE could operate to shift valued
outcomes from those contingent on external agents
associated
with
goal-accomplishment
Finally,
all
implying
that
three
the
goal-directed
(IV
and
behavior
EV.),
assessments of
use
of
MBE
assessments of goal-difficulty.
is
are
P.
are
more
(EV.
(^^k^
highly
)
to
those
^^
internal
opposed
to
under
MBE.
valued
negatively associated with MBE,
associated
with
increased
subjective
This possibility was also alluded to in the
earlier discussion.
H
:
The
second
hypothesis
was
that
increased
budgetary
participation
would favorably influence the relationship between MBE and motivation.
14
As a
this
first test of
participation scores
of
sets
two
the
hypothesis,
vjere
measures
dichotomized into their upper and lower halfs, and the correlation
between
computed
(high
condition
correlations
MBE
and
Tabic
low).
versus
8
participation are
under high
reports
consistent
are
results
The
test).
for
approximately
zero,
<
(p
measures
of
4
while
0.05,
The
under
one-tail
participation.
for example, Winkler and Hays
Using the Fisher transformation (see,
p.
both
results.
the
low participation they are both significantly negative
participation
each
under
observed
motivation
653), the two correlations under each participation measure
[1975]
differ at p
< 0.05, providing support for the rejection of H^.
INSERT TABLE 8 HERE
A
motivation
means
efficient
more
on
the
MliE,
dichotomized
multiplicative interaction term.
then
be
ur,ed
to
estimate
of
regressions
simple
testing
of
the
H^
involves
participation
a
scores
The coefficients from this
intercept
motivation
on
slope,
and
MBE
-
one
regression
(+1/-1),
a
regression can
respectively,
for
of
and
each
of
of
two
the
two
participation groups.
The regression equation was:
where
Y
= a + 3,X^
Y
= Motivation
X.
= MBE, rescaled, mean = zero.
-f
B^X^ + 33X^X2,
(2)
X„ = Budgetary Participation (+1 = high, -1 = low)
In
this model,
high values of MBE (positive) will combine with +1
15
for
high participation
produce a positive interaction
to
terra.
values
Lov?
of
MBE (negative) will combine with -1 for Low participation, also to produce a
positive interaction
Motivation should generally be higher in these
terra.
conditions, so a positive interaction coefficient, 3^,
v/as
The results of the regression are presented in Tables
expected.
9
(for the Milani
c
participation measure) and
(for
10
the
Hofstede
reveal positive interaction coefficients,
the
0.10
level.
These
results
each is
but
consistent
are
Both analyses
measure).
significant
with the
above
at
only
correlation
tests.
INSERT TABLES
As
a
Milani
final
measure
relationship
analysis,
(Table
between
the
9)
9
AND 10 HERE
from
coefficients
utilized
v;ere
motivation
and
MBE
to
the
reconstruct
separately
using
regression
a
functional
high
under
the
and
low
participation:lligh
Low
These
participation:
Y = 10.46 + 0.35X^
participation:
Y =
two functions intersect at
in Equation
it
is
2,
somewhat
(3)
9.46 - 6.11X
(4)
X,=-0.15.
Since
this point of intersection is
unlikely
(at
least
16
on
the
basis
X^ has
X - 1.240x,
of
the
a
mean
of
zero
indicating that
present
sample)
to
higher
on MBE which
score
encounter a
under
coefficient
motivation
for
that,
higher
general,
in
be
to
positive
significant,
the
indicates
participation)
(on
3„
Indeed,
participation.
low
low
sufficiently
is
participation is associated with higher motivation.
Conclusions
6.0
of
results
The
were
study
this
disappointing.
generally
overall pattern tends to be consistent with the
the
particulcirly
contingency
significance and explanatory power of
limitations of the study
feel,
I
,
theoretical propositions,
levels
the
hypothesis,
statistical
of
major
The
tests were marginal.
the
the
VThile
revolve around the conceptualization of
Three particular problems v/ith the model used in this
the expectancy model.
study are as follov/s.
it
First,
IV,).
Second,
issue
The
of
classes
it
is
(as
why
unclear
only
is
valence
and IV,) are
plausible
is
for
but
outcomes,
(IV
provides
for
outcomes
(for
"weights"
v/hat
,
IV,
,
unclear v/hether the
apply
should
and
of
classes
tvjo
to
each
each
incorporated
EV)
of
intrinsic
in
this
study)
that
a
particular
three
the
the
model?
outcomes
It
and
IV^
of
in
intended to be viewed as mutually exclusive.
assumed
extrinsic
multiple
model
intrinsic
the
single
(IV
seems more
intrinsic
outcome
could arise from both goal-directed behavior and goal accomplishment, and be
valued
differently
according
to
each
source.
Third,
and
perhaps
most
important, the model fails to allow for the possibility of extrinsic rev/ards
associated with goal-directed behavior.
likely to react to job-effort
Both
superiors
and
peers
(goal-directed behavior) as they are
are
to
as
job
performance (goal-accomplishment) on the part of an individual.
The importance of this last issue is in connection with a reconciliation
17
of the behavioral and agency theoretical perspectives on MBE.
MBE involves
model,
variance
a
In the agency
investigation strategy which will yield
performance,
(agent's)
further signal regarding the subordinate's
a
a
signal
which might indicate, for example, that an unfavorable variance was due to a
lack of subordinate effort as
subordinate's
opposed to factors beyond the
r
control.
So long as the subordinate perceives that the signal generated by
is
investigation
the
promise
this
is
of
not
an investigation is
true
or
v;hether
appropriate
Demski,
risk-averse
for
gives
investigation
it
strategies
subordinates.
the
that
showed
(1979)
receive
MBE)
(i.e.
attention
more
of
a
of
the
(Baiman
and
outcome
"lottery"
particularly
find
subordinates
risk-averse
particularly
are
uncertain
The
characteristics
the
which
192-193)
pp.
variances
then
exerts,
he
Holmstrom
Baiman and Demski (1980) extended this result and
investigation
low-tail
effort
the
motivating.
unfavorable
not
than favorable variances.
showed that
of
independent
Hence, it follov/s that the use of MBE would be associated with
distasteful.
greater effort on the part of such subordinates to perform at a level which
will avert the unfavorable variance which triggers the investigation.
the
In
P„EV,
context
the
capture
might
,
of
the
with
this
associated
with
goal-directed
such
a
will
reward
probability
goal-directed
administering
that
be
the
effort
EV^^.
superior's
on
P3
forthcoming,
the
would
part
and
or,
P^
in
investigation
of
therefore
the
is
an
the
will
capture
the
uncertainty
reward
probability
context,
reveal
subordinate,
term,
a
extrinsic
agency
the
of
the
of
represents
EV
behavior,
inclusion
effects
motivational
investigation.
associated
the
model,
expectancy
the
that
the
substantial
basis
probabilistic
for
outcome
of the variance investigation process and we might predict that it would be
positively associated with MBE.
18
Interestingly,
which
specifically
from my
study
do
Staw (1977) advocates a version of
incorporates
not
permit
a
^^^^h
measurement
^^^^'
of
it.
the
expectancy model
Unfortunately,
It'
seems
that
the
data
such
an
expanded model might provide a fruitful basis for some future integration of
the behavioral and econorao
c
perspectives on MBE, which, on the basis of the
results presented in Table 5, appears to be in widespread use in budgetary
control systems.
19
TABLE
I
Means and Standard Deviations of Intrinsic Valences, and
Assessment of Differences among the Valences (Two-tailed, paired sample t-test)
Outcome'*'
TABLE
2
Means and Standard Deviations of Extrinsic Valences
Mean
Outcome*
Std. Dev.
1.
3.09
0.84
2.
0.70
0.88
3.
3.10
0.87
4.
2.25
1.10
5.
3.16
0.81
6.
2.30
0.83
7.
1.41
1.07
8.
-1.09
1.30
9.
1.79
1.27
* See appendix for list of outcomes numbered in table.
21
TABLE 3
Means and Standard Deviations of Probabilities/Instrumentalities
and Correlations among ?i items
TABLE 4
Mean, Standard Deviation, and Range of Final Motivation Measure
Mean
TABLE
5
Descriptive Statistics for MBE
Mean
Item
Std. Dev.
0.14
0.81
Unfavorable variances receive
more managerial attention than
favorable variances
Unfavorable
variance;
Favorable
variances
Difference
0.58
0.34
0.24
12.22
My explanation of *
budget variances is
included in performance reports.
0.63
0.53
0.10
6.16
My superiors discuss
budget items with me
when * variances occur.
0.77
0.49
0.28
13.92
I am required to submit
an explanation in writing
about causes of large _2
budget variances.
0.56
0.50
0.06
2.77
I am required to trace
the cause of £ variances to groups or
individuals within my
department.
Descriptive Statistics for Final Measure of MBE
Mean
0.299
+
Std. Dev,
0.125
A paired sample t-test was used.
24
Maxiraura
0.67
TABLE
6
Descriptive Statistics for MiLani and llofstede Measures
Mean
Milani
Hofstede
Std. Dev.
Min.
Ma>
32.99
5.40
14
42
5.46
1.38
1
8
25
TABLE
7
Correlations betv/een MBE and Expectancy-
Model Components
Outcome"^
N = 122
TABLE
8
Correlation between MBE and motivation
under high versus low participation
High Participation
Low Participation
Difference*
*
'
Milani (N=12l)
Hofstede (N=120)
0.02 (n=65)
0.04 (n=67)
-0.23 (n=56)
-0.24 (n=53)
t-1.83, p < 0.05
Tests performed under the null hypothesis that
in the high participation group.
27
t=2.01, p
p
<
0.05
= the sample correlation
Table
9
Results of Regression based oa Milan! measure
Coefficient
Table LO
Results of Regression based on Hofstede Measure
Coeffic
Appendix
List of Outcomes
Eight intrinsic and nine extrinsic outcomes were
used
in
this
study.
To
avoid cluttering the tables, the outcomes were numbered as follows:
Intrinsic
Extrinsic
1.
Pay Raise
1«
Personal growth and development
2.
High Pay
2.
Setting higher standards for yourself
3.
Respect from boss
3.
Giving help to others
4.
Respect from other employees
4.
Time at
5.
Receiving more compliments
5.
Feelings of security
6.
Greater chances for independent 6.
v.'ork
passing fast
Setting higher standards for others
thought and action
7.
Fevrer chances to make friends
7.
Feelings of accomplishment
8.
Special reward or recognition
8.
Being tired
9.
Promotion
30
Footnotes
1.
These raw scores were reversed and rescaled by subtracting five from all
"neutral" responses score zero,
As a result of this procedure,
scores.
while responses in the
"desirable" direction
positively
score
(one
to
four) and responses in the undesirable direction score negatively (minus
r
one to minus four).
2.
The
complement of the score on
the
item can be viev;ed
single
the
as
probability that both favorable and unfavorable variances receive equal
attention,
or
that
variances
favorable
of
indicate
0.5
Therefore,
it
the
might
be
MBE
of
presence
attention
more
receive
than
In this sense, only scores in excess
unfavorable (i.e., "reverse" MBE).
argued that
as
defined
should
0.5
be
this
in
study.
from all
subtracted
scores on the single measure before its addition to the sum of the four
difference
scores.
Such
(0.1) from the
constant
would
procedure
a
result
in
final scores for MBE and vrould,
subtracting
a
therefore, not
affect the results in any way.
3.
Of the ten intercorrelations among the five items used to construct the
MBE measure, five were significant at
p
0.01,
<
one
at
p
<
0.02,
and
the remaining four were either insignificant or signficant at no better
than
p
<
0.10.
These
last
four
intercorrelations
first of the four difference measures
this
result,
(omitting
the
an
alternative,
uncorrelated
(see
four-item
item).
Table
MBE
However,
5).
measure
none
of
all
On
was
the
involved
the
the
basis of
constructed
test
results
reported in the results section of this paper is affected by the choice
between the four- and five-item measures.
31
involving
and
Hilani measure
the
N
to
=
tests
for
120
involving
the
This was due, in each case, to improper completion of
llofstede measure.
the participation measure involved.
5,
None
of
participation
and
(X^)
between
correlation
model,
in
In the Hilani model,
MBE
and
interaction
the
Hofstede
variables
independent
the
intercorrelated.
term
MBE
these
was
(X,)
and
(X.
the
model
either
significantly
v/as
the biserial correlation between
-0.01.
X„),
three
correlations
the
risk-averse
product-moment
The
was
term
interaction
participation
between
and
0.09,
were,
In
0.08.
respectively,
the
-0.03,
0.00, and 0.10.
6.
In
other
words,
investigation
motivated to avoid.
subordinates,
"reward."
investigation,
represents
a
subordinate,
penalty
which
implies
subordinate
that
the
Bairaan and Demski also show that,
the investigation strategy (lottery)
This
risk-tolerant
for
strategy
is,
high-tail
a
perform
to
at
a
variance.
32
level
strategy
at
a
which
level
low-tail
is
for risk-tolerant
should be used as
a
motivate
a
which
will
a
subordinate
will
which
produce
induces
a
the
favorable
References
Argyris,
C,
The Impact of Budgets on People.
New York:
Controllership
Foundation, 1952.
Atkinson, J. W.
of
Motives,
,
"Towards Experimental Analysis of Human Motivation in Terms
Expectations and Incentives."
Motives in Fantasy Action and Society.
Bainan, S. and J. S. Deniski
,
in
J.
New York:
\J.
Atkinson,
Editor,
Van Nostrum, 1958.
"Economically Optimal Performance Evaluation
Studies on Economic Consequences of Financial and
and Control Systems."
Managerial Accounting:
on Corporate
Effects
Incentives
supplement to the Journal of Accounting Research (1980):
and
Decisions,
184-220.
Bandura, A., "Social Learning through Imitation," in M. R. Jones, Editor,
Nebraska Symposium on Motivation.
,
University of Nebraska Press, 1962.
"Influence of Models' Reinforcement Contingencies on the
Acquisition of Iraitiative Responses."
Psychology (June 1965):
,
Journal of Personality and Social
589-595.
Social Learning Theory.
New York:
General Learning Press,
1972.
Birnberg, J. G. and R. Nath, "Implications of Behavioral Science for
Managerial Accounting."
The Accounting Review (July 1967):
Bittel, L. R., Management by Exception:
the Managerial Job.
New York:
33
468-479.
A Systematizing and Simplifying of
McGraw-Hill, 1964.
BrownelL, P., "Participation in Budgeting, Locus of Control and
Effectiveness."
Organizational
The
Review
Accounting
(October
1981);
844-859.
,
Budgetary
"The Role of Accounting Data in Performance Evaluation,
Participation and
Organizational
Accounting Research (Spring 1982):
Effectiveness."
Journal of
12-27.
BrunSj U. J. Jr. and J. H. Waterhouse, "Budgetary Control and
Journal of Accounting Research (Autumn 1975):
Organization Structure."
177-204.
Bryan, J. F. and E. A. Locke, "Goal Setting as a Means of Increasing
Calder, B. J. and B»
Motivation;
274-277.
Journal of Applied Psychology (June 1967):
Motivation."
II.
Some
Staw, "Interaction of Intrinsic and Extrinsic
Methodological
Social Psychology (January 1975):
Notese"
Journal
of
Personality
and
76-80.
Cherrington, D. J. and J. 0. Cherrington, "Appropriate Reinforcement
in
Contingencies
Accounting:
the
Budgeting
Selected Studies,
Research (1973):
Process."
supplement
to
Empirical
the
Research
in
Journal of Accounting
225-253.
DeCoster, D. T. and J. P. Fertakis, "Budget-Induced Pressure and its
Relationship to Supervisory Behavior."
(Autumn 1968):
237-246.
34
Journal of
Accounting Research
Galbraith, J. lU, and L. L. Cummings, "An Empirical Investigation of the
Determinants
Motivational
Behavior and Human Terformance (August 1967):
Effects
Interactive
Performance:
Task
of
Between Instrumentality-Valence and Motivation-Ability,"
Organizational
237-257.
Georgopoulos, B. S., G. M. Hahoney, and N, W. Jones, "A Path Goal
Approach
.
1957):
to
Productivity,"
Journal
of
Applied
Psychology
(December
345-353.
Hofstede, G. H., The Game of Budget Control.
Hopv/ood, A. G..
"An Empirical Study of the Role of Accounting Data in
Performance, Evaluation."
Studies,
Van Gorcum, 1967.
Assen:
supplement
to
Empirical
the
Research
Journal
of
in
Accounting:
Accounting
Research
Selected
(197 2):
156-182.
Hull, C. L., "A Functional Interpretation of the Conditioned Reflex."
Psychological Reviev/ (January 1929):
498-511.
Itami, H., "Evaluation Measures and Goal Congruence Under Uncertainty,"
Journal of Accounting Research (Spring 1975):
73-96.
Kenis, I., "Effects of Budgetary Goal Characteristics on Managerial
Attitudes
•
and
Performance."
The
Accounting
Review
(October
197 9):
707-721.
Lawler, E. E. and J. L. Suttle, "Expectancy Theory and Job Behavior."
Organizational Behavior and Human Performance (June 1973):
35
482-503.
"Moral Hazard and Observability."
Holrastrom, B.,
Bell Journal of Economics
74-91
(Spring 1979):
Merchant, K. A., "The Design of the Corporate Budgeting System:
on Managerial Behavior and Performance."
The Accounting Review (October
813-829.
1981):
Milani, K.
Influences
\1 .
"The Relationship of Participation in Budget-Setting to
,
Industrial Supervisor Performance and Attitudes:
"Budget Use and Managerial Performance."
,
Research (Spring 1978);
Tlie
274-284.
Accounting Review (April 1975):
Otley, D. T.
A Field Study »"
Journal of Accounting
122-149.
Porter, M. and Z. Zannetos, "Administrative Regulation Versus Market
in
Regulation
the
Diversified
Company,"
Sloan
School
Paper
l.'orking
987-78, MIT (April 1978).
Ronen, J. and J. L. Livingstone, "An Expectancy Theory Approach to the
..ocivational
Impact
of
Budgets,"
Accounting
Review
(October
1975):
671-685.
Rotter, J. B., Social Learning and Clinical Psychology.
Englewood Cliffs:
Prentice-Hall, 1954.
,
Control
of
"Generalized Expectancies for Internal versus External
Reinforcement."
Psychological
Applied (1966, whole No. 609):
36
1-28.
Monographs:
General
and
Spence,
iC.
W.
,
"Theoretical Interpretations of Learning." in F. A. tloss,
Editor, Comparative Psychology.
Englewood Cliffs:
Staw, B. M., "Motivation in Organizations:
Prentice-Hall, 19A2.
To\;ard Synthesis and Redirection"
in B. M. Staw and G. R. Salancik (eds.) New Directions in Organizational
Behavior.
Chicago:
St. Clair Press,
1977.
Swieringa, R. J. and R. H. Honcur, Some Effects of Participative
Budgeting on Managerial Behavior.
National Association of Accountants,
1975.
Tolman, E.
C,
Purposive Behavior in Animals and Men.
New York:
Century, 1932.
Winlcler, R. L. and W. L. Hays, Statistics:
and Decision.
New York:
Probability Inference
Holt, Rinehart and V/inston,
37
1975.
Appleton
eKSt^^^l
Du,
745931
* BK S
M
S m.i«,
,
"oao DDe 3a6 sas