3‘ .
.91‘.‘

 

.. "tau. 1.
I: 9 .v
:35. 1.;

.q .x

 

 

IA ($1.51.. .

 

 

IHESIS
'L

MICHIGANS

' IIIIIIIIIIIIIII

 

 

/ Nivensnv LIBRARIES

III II III II I IIII III III

 

II

This is to certify that the

dissertation entitled

IMPLEMENTING FINER COST ALLOCATION METHODS: IMPACT OF
COMPLEXITY, COMPETITION, STRATEGY, MANAGERIAL SUPPORT,
AND IMPLEMENTATION OBSTACLES

presented by

Win Gilkey Jordan

has been accepted towards fulﬁllment
of the requirements for

Doctor of Philosophy degreein Business Administration

 

 

 

Mad ¢ [IQ/kw

M a ]01' professor

Date June 28, 1996

 

MS U is an Afﬁrmative Action/Equal Opportunity Institution 0-1277 1

_. ‘v -.4 o

_ .._ A_ﬁ-.._q_ ,

 

LIBRARY

Michigan State
University

 

 

 

PLACE IN RETURN BOX to romovo this checkout from your rooord.
. TO AVOID FINES rotum on or baton date duo

DATE DUE DATE DUE DATE DUE

      
  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

iii

MSU loAnNﬁrmotivo Action/Equal Opportunity Institwon
«cu-mm

-7 m—

IMPLEMENTING FINER COST ALLOCATION METHODS: IMPACT OF
COMPLEXITY, COMPETITION, STRATEGY, MANAGERIAL SUPPORT,
AND IMPLEMENTATION OBSTACLES
By

Win Gilkey Jordan

A DISSERTATION
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Accounting

1996

ABSTRACT
IMPLEMENTING FINER COST ALLOCATION METHODS: IMPACT OF
COMPLEXITY, COMPETITION, STRATEGY, MANAGERIAL SUPPORT,
AND IMPLEMENTATION OBSTACLES
By

Win Gilkey Jordan

This descriptive study examined whether certain ﬁrm and industry
characteristics increase the likelihood of a ﬁrm’s implementing ﬁner cost
allocation (FCA) methods. Characteristics examined included degree of
complexity in products and processes, level of competition experienced,
degree to which certain strategies are followed, strength of managerial
support for FCA methods, and strength of implementation obstacles facing
FCA methods.

Most data came from 124 survey responses from manufacturing plants
in various industries. Through manual correlation analysis, principal
components factor analysis, and item reliability analysis, 50 measures were
reduced to 15 which formed four independent constructs (no construct was
found for competition). Factor scores were generated for each response. The
factor scores and six selected measures were analyzed against FCA methods
implementation scores in a series of multivariate regression models.
Depending on the measures included, missing data caused the analyses to be

based on 111 or 112 observations.

Of ﬁve constructs hypothesized as inﬂuencing FCA method
implementation, the constructs for three exhibited signiﬁcance and correct
signs: complexity (99%), implementation obstacles (95%), and managerial
support (90%). The data therefore support the three associated hypotheses.
An additional complexity measure (an increased number of labor rates
reﬂecting complexity of human resource skills) was also found to be a
signiﬁcant variable of correct sign.

The lack of a cohesive competition construct precluded directly testing
the hypothesized competition relationship. However, three competition
measures (ﬁrm portion of HHI, perceived level of competition, and lower
proﬁt margins) used as variables in the regressions showed signiﬁcant
relationships but of mixed signs, thus indicating a lack of support for the
competition hypothesis.

A partial construct for strategy was signiﬁcant at the 90% conﬁdence
level and two strategy measures (process innovator and cost-based pricing)
used as variables in the regressions were signiﬁcant at the 95% conﬁdence
level. However, the signs for the three variables were opposite that
hypothesized, thus indicating a lack of support for the strategy hypothesis.

When only derived constructs were used, the regression model
explained only 7% of variation. When the additional measures for
complexity, competition, and strategy were included, the model explained

27% of the variation.

I dedicate this dissertation to Valerie Neville Jordan. Her continuing
support and encouragement throughout the years have given me the courage
to turn my dreams into reality. As wife, mother of our children, companion,
partner, advisor, and best friend, she has molded my life and made it worth
living. When I have become the best that I can be, perhaps I shall begin to

be worthy of her.

iv

ACKNOWLEDGMENTS

My greatest acknowledgment is to those who sacriﬁced so much that I
might successfully undertake my doctoral program: my family members. My
heartfelt thanks goes to Valerie, my wife, and our children (Kendra,
Kimberly, Nathan, Aaron, Jocelyn, Adam, Joshua, and Julia) for supporting
me throughout the entire program. Their being uprooted so that I could
enter a doctoral program, then experiencing the privations that accompany
being the family of a doctoral student, were sacriﬁces of such magnitude that
this culminating dissertation is as much theirs as mine.

I express my gratitude to my dissertation committee members —
Susan Haka, Fredric Jacobs, Ronald Marshall, and Stanley Fawcett — for
their insight, help, and patience throughout the dissertation process.

I appreciate the time and efforts of the survey participants, as well as
those who helped validate the survey. They provided information which
would not have been available in any other fashion.

I alsp acknowledge the ﬁnancial assistance of the Michigan State
University Department of Accounting in funding the publication and mailing

of the surveys, as well as numerous telephone calls.

My appreciation extends also to my friends and colleagues at the
University of Idaho for their encouragement and support during the analysis
and writing phases. Special mention goes to Christopher Williams and Kirk
Steinhorst of the University of Idaho Statistical Consulting Center for their
generous help during the analysis.

I express my gratitude also to my father-in-law, Joseph T. Neville, for
his friendship, encouragement, and sacriﬁce in helping me get started on the
analysis and writing of this dissertation. He served faithfully as a sounding
board in exploring alternatives. Similarly, my mother-in-law, Evelyn
Neville, lent her support to my eﬁ'orts.

I gratefully acknowledge my parents, Don Armstrong Jordan and
Louie Mae Gilkey Jordan, who sacriﬁced so much that I might have the
opportunities they never had. Their love of truth set both the tone and the
balance for my future research efforts. I hope that I may yet become half the
teacher that my father was.

Finally, I acknowledge the loving support of my Heavenly Father, who
has blessed me beyond measure in so many ways. From Him I have learned

that the pursuit of truth is an eternal process.

TABLE OF CONTENTS

LIST OF TABLES ..................................................................................... ix

LIST OF FIGURES .................................................................................... xi

CHAPTER I INTRODUCTION ............................................................ 1
CHAPTER II RESEARCH MOTIVATION , BACKGROUND, AND

PRIOR RESEARCH ................................................................................. 8

RESEARCH MOTIVATION ........................................................... 8

BACKGROUND AND PRIOR RESEARCH ................................... 12

Background and Development of Finer Cost Allocations .. 12

Complexity ........................................................................... 16

Competition ........................................................................... 18

Strategic Positioning ............................................................ 23

Managerial Support ............................................................. 28

Cost-Beneﬁt Theory ............................................................. 30

CHAPTER III CONCEPTUAL MODEL AND HYPOTHESES ............ 34

OVERVIEW AND CONCEPTUAL MODEL ................................. 34

COMPLEXITY HYPOTHESIS ....................................................... 36

COMPETITION HYPOTHESIS ..................................................... 41

STRATEGY HYPOTHESIS .......................................................... 43

MANAGERIAL SUPPORT HYPOTHESIS ................................... 45

IMPLEMENTATION OBSTACLES HYPOTHESIS ..................... 46

vii

CHAPTER IV METHODOLOGY ........................................................... 48

APPROACH .................................................................................... 48
SAMPLE SELECTION ................................................................... 50
RESEARCH DESIGN AND DATA MEASURES ........................... 53
CHAPTER V STATISTICAL TESTS AND RESULTS .......................... 73
DESCRIPTIVE ANALYSIS ............................................................ 73
Demographic Data ............................................................... 73
Univariate Descriptive Analysis ......................................... 75
CONSTRUCT DEVELOPMENT ................................................... 84
Manual Reduction of Measures ........................................... 87

Factor Analysis, Item Reliability Testing, and Generation
of Factor Scores .................................................................. 97
MULTIVARIATE REGRESSION ANALYSIS ............................... 103
HYPOTHESES TEST RESULTS ................................................... l 10
DISCUSSION ................................................................................. l 12
Demographic Analysis .......................................................... l 12
Measures, Factors, Item Reliability, and Constructs ......... l 13
Regressions ........................................................................... l 16
CHAPTER VI CONCLUSION ................................................................ 120
CONTRIBUTIONS .......................................................................... 120
LIMITATIONS ................................................................................. 122
DIRECTIONS FOR FUTURE RESEARCH ................................... 122
APPENDIX SURVEY INSTRUMENT ................................................... 124
LIST OF REFERENCES ............................................................................ 13 1

viii

LIST OF TABLES

Table 1. Measures Designed to Assess Complexity ........................................ 56
Table 2. Measures Designed to Assess Competition ....................................... 60
Table 3. Measures Designed to Assess Strategy ............................................. 65
Table 4. Measures Designed to Assess Managerial Support .......................... 68
Table 5. Measures Designed to Assess Implementation Obstacles ................ 70
Table 6. Respondents’ Years in Company and Its Relation to OVERALL 75
Table 7. Industries Represented in Survey and Their Relation to

OVERALL ........................................................................................... 76
Table 8. Descriptive Statistics on 50 Measures and OVERALL .................... 78
Table 9. t-Tests for Mean Diﬁ‘erences Between Plants With Implemented

FCA Methods and Plants Using Less-Fine Allocation Methods ...... 85
Table 10. Pearson Correlation Matrix for All Measures ................................ 88

Table 1 1. Measures Identiﬁed and Dropped While Reducing Number of

Measures ........................................................................................... 96

Table 12. Item Reliability Testing Results for the Four Factors Identiﬁed

in Factor Analysis .......................................................................... 100
Table 13. Factor Loadings and Standardized Scoring Coeﬁcients for the

Constructs ....................................................................................... 102
Table 14. Factor Scores Calculated for Constructs of 10 Observations ....... 103

Table 15. Pearson Correlation Matrix and Simple Statistics of Constructs

and Separate Measures Used in Regression Analyses ................. 106

Table 16. Assessment of Models Used in Regression Analyses ................... 107

Table 17. Assessment of Independent Variables Used in Regression
Models ............................................................................................. 108

LIST OF FIGURES

Figure 1. Basic Model of Hypothesized Relationships ................................... 35
Figure 2. Least Complex Resource Consumption Pattern ............................ 38
Figure 3. Moderately Complex Resource Consumption Patterns .................. 39

Figure 4. Most Complex Resource Consumption Pattern .............................. 40

CHAPTER I

INTRODUCTION

This descriptive study examined whether type and strength of certain
ﬁrm and industry characteristics increase the likelihood of a ﬁrm’s
implementing ﬁner cost allocation (FCA) methods‘. The characteristics
reﬂected 1) market structure and the ﬁrm’s struggle for sales (grouped
together herein as competition), 2) strategies pursued, 3) the degree to which
the ﬁrm's products consume resources in diverse and uncorrelated patterns
(referred to herein as complexity), 4) the extent of managerial support offered
for the implementation, and 5) the strength of implementation obstacles.

Although not directly tested in this study, cost-beneﬁt theory
[Zimmerman (1979) and Jensen and Meckling (1992)] provided the
underlying rationale for implementing FCA methods by maintaining that
ﬁrms expend resources only when increased beneﬁts are expected to produce
a net gain. The study proposed ﬁve hypotheses relating the implementation
of FCA methods to ﬁrm and industry characteristics. The hypotheses drew

on theories and models from strategic planning, economics, systems

 

1 F CA methods include all methods beyond a single plantwide allocation method;
FCA methods include using 1) departmental allocations, 2) multiple cost pools and cost
application bases, or 3) activity-based costing or other processodriven methods.

1

2
management, and managerial accounting. The ﬁve hypotheses suggest that

implementation of FCA methods is increased by 1) greater complexity in
indirect resource consumption, 2) greater competition, 3) following certain
ﬁrm strategies, 4) greater managerial support for FCA methods, and 5) fewer
and weaker obstacles to implementing FCA methods. Empirical survey and
analysis were used to test the hypotheses, although the competition
hypothesis was eventually untested due to a lack of item reliability for the
competition measures proposed.

The results of this study are of interest to ﬁrms considering whether to
implement such FCA methods as departmental application rates, multiple
cost pools, or activity-based costing (ABC). Firms make strategic decisions
about allocating overhead, pricing products, preparing ﬁnancial statements,
calculating taxes, and performing product evaluations; the accuracy of costs
impacts each of these activities. Theory suggests that managers
implementing FCA methods perceive the beneﬁts to outweigh the added costs
of changing systems, training personnel, and gathering and processing the
required information.

The hypotheses build on four theories: theory of complexity, theory of
competition, strategic planning theory, and systems management theory. The
theory of complexity [Gupta (1993); Hwang, Evans, and Hegde (1993); Datar
and Gupta (1994); B6er and Jeter (1993)] examines how various
characteristics of manufacturing inﬂuence whether simple allocation

techniques distort costs. The theory of competition shows the importance of

3

pricing [Varian (1990, 1992)], the ways that industry structure inﬂuences
competition [Scherer and Ross (1990)], and how competitive pressure varies
with time and circumstances [Boston Consulting Group (1968), Porter (1980),
Richardson and Gordon (1980), and Anthony and Ramesh (1992)]. Strategic
planning theory addresses the ways that ﬁrms analyze and manage
competition [Levitt (1965), Porter (1980), and Kotler (1991)]. Systems
management theory [Yellen (1993), Martinsons (1993), Day (1994), Beatty
(1992)] illustrates that widespread managerial support and the authority
and eﬁ'ectiveness of the champion inﬂuence the scope of implementation
within a ﬁrm. Systems management theory also suggests that
implementations occur more frequently when the obstacles to
implementation are fewer and weaker [see Campi (1992), Cooper et al (1992),
and Yourdon (1989)]. These theories are used to explain 1) why some ﬁrms
are willing to incur additional installation, collection, and processing costs in
order to obtain more detailed costs and 2) how detailed costs are used to
guide cost reductions and increase proﬁtability or to help set product prices
at levels that maximize proﬁts and cover both ﬁxed and variable costs.

The study used data from a cross-sectional sample of 124 plants
selected at the 4-digit SIC code level. Data on industry structure and ﬁrm-
SIC-code operating margins were obtained from the Standard and Poor's
Compustat Industrial data base; veriﬁcation of industry coverage in selected

industries was obtained from Manufacturing USA [Darnay (1994)]; and

4

information on ﬁrm beginning year was obtained from Moody’s manuals
[Moody’s (1995)] and the CorpTech Directory of Technology Companies
[Corporate Technology Information Services (1995)]. Other items of
information were obtained through surveys completed by respondents in the
sample plants. The surveys provided respondents’ perceptions on complexity,
competition, strategy, managerial support, and implementation obstacles. In
addition, the surveys captured information on the cost allocation methods
implemented by the plants.

Descriptive statistics were developed on each of the 47 items obtained
from the surveys and on the three calculations based on Compustat data; the
descriptive statistics showed the data to be nonnormal, with signiﬁcant
skewness and kurtosis. As explained in Johnson and Wichern (1992), the
lack of normality affected the usefulness of correlation analysis; the
magnitudes of Pearson correlation coeﬂicients were still valid, but the
signiﬁcance of the correlations become suspect. Furthermore, the lack of
normality dictated the use of principal components factor analysis (rather
than maximum likelihood factor analysis) since principal components factor
analysis is robust against the lack of normality.

In order to use factor analysis meaningfully on the small sample
available, a ratio of 5 observations per measure analyzed necessitated
reducing the number of measures from 50 to 25 or less. Some 27 measures

were dropped after examining either 1) the correlations between measures or

5

2) the correlations between measures and the sums of the measures proposed
for a group of measures, in accordance with Churchill (1979); Churchill
advocated retaining measures which correlated strongly. Two additional
measures were dropped because they measured essentially the same thing
and did not correlate strongly with anything else; two other measures were
dropped because they had so little variation that they were uninformative.

The surviving 19 measures were used in a principal components factor
analysis with varimax rotation; four independent factors were found which
roughly corresponded to the intended factors, although no factor emerged for
competition. Each factor was then tested for item reliability by calculating
Cronbach’s Coeﬁcient Alpha; this testing eliminated another four measures.

Then the data were factor analyzed with principal components factor
analysis with varimax rotation (obtaining the same factors) to develop factor
scores for each of the responding plants. The factor scores and six separate
measures (representing competition, speciﬁc aspects of strategy, and an
additional aspect of complexity), were analyzed in a series of multivariate
regression models to test the proposed hypotheses. (Depending on the
separate measures included in a model, missing data caused the regressions
to be based on 111 or 112 observations.)

Of ﬁve constructs hypothesized as inﬂuencing FCA method
implementation, the constructs for three exhibited signiﬁcance and correct

signs: complexity (99%), implementation obstacles (95%), and managerial

6
support (90%). The data therefore support the three associated hypotheses.
An additional complexity measure (an increased number of labor rates
reﬂecting complexity of human resource skills) was also found to be a
signiﬁcant variable of correct sign.

The lack of a cohesive competition construct precluded directly testing
the hypothesized competition relationship. However, three competition
measures (ﬁrm portion of HHI, perceived level of competition, and lower
proﬁt margins) used as variables in the regressions showed signiﬁcant
relationships but of mixed signs, thus indicating a lack of support for the
competition hypothesis.

A partial construct for strategy was signiﬁcant at the 90% conﬁdence
level and two strategy measures (process innovator and cost-based pricing)
used as variables in the regressions were signiﬁcant at the 95% conﬁdence
level. However, the signs for the three variables were opposite that
hypothesized, thus indicating a lack of support for the strategy hypothesis.

When only derived constructs were used, the regression model
explained only 7% of variation. When the additional measures for
complexity, competition, and strategy were included, the model explained
27 % of the variation.

As the regression results indicate, several constructs and measures

related strongly with FCA method implementation. Since only 7% to 27 % of

7
the variation was explained, the hypothesized inﬂuences did not fully
explain what prompted the plants to implement FCA methods.

The remainder of this paper is organized as follows: Chapter II
examines the research motivation for this study, discusses the background
and development of cost allocations, and presents previous research.
Chapter III presents the conceptual model and develops the hypotheses for
the study. Chapter N presents the methodology, including approach, sample
selection, research design, and the measures used. Chapter V discusses the
statistical tests and their results, including demographics, univariate
descriptive analysis, factor analysis, item reliability testing, and multivariate
regression analysis. Finally, Chapter VI discusses what contributions the
study has made, the limitations on the study’s practical usefulness, and

directions for future research.

CHAPTER H

RESEARCH MOTIVATION, BACKGROUND, AND PRIOR RESEARCH

RESEARCH MOTIVATION

Since the early 1900's, many management accountants have
increasingly argued for ﬁner cost allocations [Johnson and Kaplan (1987),
Previts and Merino (1979), Chatﬁeld (1977)]. The variety of allocation
methods developed and the concern with improper allocations leading to
incorrect product costs expressed by such authors as Shank and
Govindarajan (1988), Johnson and Kaplan (1987), Cooper and Kaplan (1987),
and Cooper (1986) suggest that some academicians favor the use of FCA
methods.

In a survey by Green and Amenkhienan (1992), 45% of respondents
indicated they had implemented activity-based costing (ABC) to some degree
within their ﬁrms. Notwithstanding, some ﬁrms continue to use simple
allocation methods. In a survey taken by Emore and Ness (1991), 74% of
ﬁrms still use labor hours or labor dollars as the allocation base and 30% of

ﬁrms still use plantwide overhead allocation. Perhaps the nature of business

9
in some ﬁrms is not appropriate for more complex allocation methods, such

as ABC.

An underlying assumption of this study is that adopters and
nonadopters of FCA methods are making rational choices; this study sought
to learn more about what characteristics inﬂuence ﬁrms to implement FCA
methods. The study examined three hypotheses about ﬁrm characteristics
(complexity, competition, and strategy) that might inﬂuence the
implementation decision; the study also examined two systems-related
hypotheses that managerial support and implementation obstacles inﬂuence
the implementation decision.

The idea underlying the study was that, since implementing and using
FCA methods consumes additional scarce resources, implementation will not
take place unless the perceived beneﬁts outweigh the perceived costs. One
premise of the study is that, unless resource consumption is diverse and
uncorrelated between products, little additional beneﬁt is perceived. Where
consumption is diverse and uncorrelated, however, it is proposed that
perceived beneﬁts arise as ﬁrms experience increasing levels of competition
and carry out their chosen marketing and pricing strategies. If the perceived
increase in beneﬁts is coupled with adequate managerial support while
facing few and weak implementation obstacles, implementations of FCA
methods increase.

The study problem is important because cost allocations impact several

important stratch decisions. Homgren, Foster, and Datar (1994) maintain

10
that cost allocations affect product pricing, ﬁnancial statements, taxes, and

performance evaluations. In each of these areas, allocations may affect
perceived and/or reported product proﬁtability.

In regular product pricing”, some ﬁrms focus prices on what the
market will bear, whereas other ﬁrms set a price which covers their current
full costs. In order to sustain long-term product proﬁtability, ﬁrms must
eventually price to cover their average long-term full costs [Varian (1992);
Homgren, Foster, and Datar (1994)]. Since products are often identiﬁed for
discontinuance when they are no longer considered suﬁiciently proﬁtable, a
ﬁrm may seek to increase its overall proﬁtability by replacing low-proﬁt
products with high-proﬁt products. Thus, the amount of cost associated with
a product can have long-lasting implications for proﬁtability.

Cooper and Kaplan (1987, 1988) discuss how improper allocations can
occur with less-reﬁned cost allocation methods. Indirect resources consumed
by one product are charged against another product because the allocation
method fails to accurately match resource consumption with the consuming
products. As a result, perceived levels of product proﬁtability can be
inaccurate and result in the retention of actual low-proﬁt products (which

were thought to be high-proﬁt products) and the elimination of actual high-

 

1 Homgren, Foster, and Datar (1994) and Anderson and Sollenberger (1992) discuss
special pricing which covers only out-of-pocket costs, but all agree the circumstances
surrounding the sale are limited. This study focused on normal sales.

11
proﬁt products (which were thought to be low-proﬁt products). When such a

combination occurs, the ﬁrm foregoes additional proﬁt [Cooper (1986)].

As the preceding discussion indicates, using less accurate or less
detailed cost information can result in economic risks. For a variety of
reasons, some rational managers choose not to use FCA methods. Gupta
(1993) found that reducing heterogeneity in cost pools by increasing the
number of cost pools does not always result in more accurate product costs;
Datar and Gupta (1994) showed that incrementally implementing FCA
methods or failing to carefully select the drivers or measurement tools can
lead to less accurate costs. Nevertheless, most studies, including Gupta
(1993) and Datar and Gupta (1994), admit that proper use of FCA methods
can lead to more accurate costs under the correct circumstances, particularly
where varying production complexity occurs. The present study investigated
complexity, competition, ﬁrm strategy, managerial support, and
implementation obstacles as characteristics that inﬂuence managerial

decisions to implement FCA methods.

12
BACKGROUND AND PRIOR RESEARCH

Background and Development of Finer Cost Allocations

Cost allocations spread indirect manufacturing costs (actual or
estimated) over the products produced. While product costing methods vary
in their procedures, the methods primarily differ in how to assign product
costs that do not vary directly with production volume [Chatﬁeld (197 7) and
Previts and Merino (1979)]. For example, one traditional method used by
many ﬁrms allocates non-volume-driven costs to products according to the
direct labor hours consumed by each product. [See Homgren, Foster, and
Datar (1994) and Johnson and Kaplan (1987).] Labor-hour-based allocation
was ﬁrst derived in a time when ﬁrms were highly direct-labor intensive.

With new technologies, many ﬁrms have become less direct-labor
intensive; these ﬁrms use increasing amounts of support labor and capital
equipment. Note, however, that the shift from being labor-intensive to being
machine-intensive is by no means uniform. B6er and Jeter (1993) found that
some industries still are labor-intensive and have maintained constant
proportions of overhead, whereas other industries have seen direct labor
shrink to insigniﬁcance and overhead grow to dominant proportions.

Firms in machine-intensive industries often have overhead amounts
far exceeding direct-labor amounts [Cooper and Kaplan (1987), Pattell

(1987), and Green and Amenkhienan (1992)]. A machine-intensive ﬁrm has

13
large amounts of capital-equipment-related overhead and a relatively small

number of direct labor hours. Allocating the overhead based on direct labor
hours means that accounting will assign a substantial amount of overhead to
a department or product for each labor hour associated with that department
or product. If a manager can ﬁnd ways to reduce the number of direct labor
hours associated with the department or product, the department or product
can avoid being charged with the associated overhead. In effect, the manager
can avoid substantial capital-equipment-related costs arising in running the
department or producing the product by eliminating labor, which is already a
less important resource.

In an effort to avoid misapplication of capital-equipment-related
overhead, FCA methods were developed. Two main allocation methods
evolved. In the ﬁrst method, known as a multiple plantwide allocation
method, plantwide overhead was broken into a few plantwide cost pools,
where the costs going into each pool had a logical cause-and-eﬂ‘ect
relationship with the pool’s cost driver (labor hours, machine hours, number
of personnel, etc.). As a unit of the cost driver is consumed, a predetermined
amount from the associated cost pool is assigned to the consuming
department or product. In the second allocation method, known as a
departmental allocation method, overhead costs incurred by the department
are placed in a pool. As other departments or products consume a unit of the
department’s major cost driver, a predetermined amount from the

departmental cost pool is assigned to the consuming department or product.

14
Both multiple plantwide allocations and departmental allocations are still

primarily volume driven.

Reﬁnements of the multiple-cost-pool approach have been developed
and are referred to as some form of activity-based costing (ABC) or process-
based costing. ABC is explained in An ABC Manager's Primer [Cokins,
Stratton, and Helbling (1993)]. Each activity is examined to determine
which and how much of the ﬁrm's resources it consumes. The costs
associated with the consumed resources are aggregated for the activity. The
activity's capacity is determined, then the costs are spread over the capacity.
Subsequently, the costs are distributed to cost objects (typically parts,
services, ingredients, products, customers, or distribution channels) that use
the activity. Where the cost objects are used internally in support of another
activity, the associated cost is passed along to the supported activity.

Proponents of ABC and other process-based costing methods behave
that such methods produce more accurate product costs by tracing a larger
portion of costs (including manufacturing costs, distribution costs, and other
selling, general, and administrative costs) directly to the activities causing
them to be incurred, then tracing the costs to the organizations or products
which consume the activities. Advocates of activity- or process-based costing
claim that activity-based allocation methods can identify when and where
improper cost allocations occur between products or organizations [Cooper

and Kaplan (1988), Johnson (1988), Shank and Govindarajan (1988),

15
Campbell (1989), and Smith and Leksan (1991)]. In addition, ﬁrms can also

use reﬁned cost data to focus attention on activities or products that can best
beneﬁt from cost-reduction projects [Cooper and Kaplan (199 1), Greenwood
and Reeve (1992), Ostrenga (1990), and Shields and Young (1992)].

The global desirability of using multiple-cost-driver methods has been
challenged by various researchers, including Banker and Potter (1993),
Gupta (1993), and Noreen (1991). While Banker and Potter (1993) upheld
the usefulness of multiple-cost-driver methods in most circumstances, the
analytical model employed showed ﬁrms to be ”strictly better off using a
direct labor single cost driver method” when the demand for an overcosted,
labor-intensive product is growing suﬁciently rapidly in an oligopolistic
situation [Banker and Potter (1993), p. 15]. Gupta (1993) showed where
allocations of aggregated costs can be more accurate than disaggregated costs
because of oﬁ‘setting differences. Noreen (1991) showed that process-based
methods are inappropriate when costs are nonlinear, when nonzero ﬁxed
costs occur at the cost-pool level, or when joint processes are involved.

As the discussion above points out, plants vary widely on the degree of
reﬁnement sought for cost allocations. What characteristics distinguish
plants that implement FCA methods from plants that do not? This study
examined several characteristics which may inﬂuence managers in choosing
which allocation method to implement: complexity, competition, strategy,

managerial support, and implementation obstacles.

16

Complexity

Research shows that, where resource consumption patterns are diverse
and uncorrelated with the allocation base, using a single allocation base
presents a distorted assignment of product costs [Johnson and Kaplan (1987)
and Noreen (1991)]. Products using less of the allocation base are assigned
less costs; nevertheless, the same products may be consuming
disproportionate amounts of support resources. [See Johnson and Kaplan
(1987); Frank, Fisher, and Wilke (1989); and Haka and Marshall (1994).]
This distortion in product costs is referred to as ”improper allocation.”
Homgren, Foster, and Datar (1994) advocate the use of improved cost tracing
and ﬁner cost allocations to avoid improper allocation and to obtain an
increased ability to measure performance and to control operations. Kee
(1995) demonstrates how ABC can be combined with the theory of constraints
to identify bottlenecks and to show the impact that bottlenecks have on
production.

Multiple plantwide allocation methods, departmental allocation
methods, and activity-based allocation methods can reﬂect more closely the
consumption of various resources when diverse and uncorrelated resource
consumption patterns occur [Hwang, Evans, and Hegde (1993)]. Some costs

are spread via direct labor (hours or dollars), others via square footage,

17
others via machine hours, others via number of personnel, etc. [Homgren,

Foster, and Datar (1994) and Green and Amenkhienan (1992)].

One adverse eﬁ‘ect of improper allocations arises with cost-reduction
eﬁ'orts. If a ﬁrm is striving to reduce costs but is improperly allocating costs,
cost-reduction eﬁ'orts could be focused on the wrong areas. Although Fisher
(1991) was not studying allocation methods, Fisher relates that implemented
cost reduction measures often fail to achieve the intended results -- cost
reduction eﬂ'orts often do not cause ﬁrms to meet cost targets or improve
return on investment. Cost-reduction effectiveness might have improved if
the guiding costs truly reﬂected cost occurrence and proﬁtability.

Hwang, Evans, and Hegde (1993) generally agree that multiple cost
pools yield more accurate costs, but they show that care must be given to
selection of appropriate cost drivers. They maintain that the bias of
conventional allocations is a function of the heterogeneity of the production
technology, unit input costs, and the product mix; they also have developed
algorithms to help managers select cost drivers. Gupta (1993) analyzed how
varying heterogeneity (complexity) in products, allocation measures, and
products’ resource usage aﬁected allocated costs at diﬂ‘erent levels of
aggregation (grosser versus more detailed cost allocations). He found
positive correlations between heterogeneity and the magnitude of cost
diﬁ’erences produced by grosser and more detailed allocation methods.
Greater complexity increased the beneﬁts offered by FCA methods in

producing more accurate costs.

18
The preceding discussion suggests that increasing complexity should

increase the need for and implementation of FCA methods. This study used
the following measures in assessing complexity: product diversity, number of
routings, the use of dedicated equipment, the use of multiple labor rates, the
importance of setups, whether the plant is a job shop, the rate of new product
innovation, compressing product life cycles, frequency of production
technology changes, and the number of product lines within the plant. As
explained in the methodology section, a subset of the complexity measures
are used in factor analysis and regression analysis to test whether complexity

increases FCA method implementation.

Competition

Mkt cture

The neoclassical economic theory of competition associates industry
structure with the proﬁt attainable, given the interaction between supply,
demand, and prices. Industry structure ranges from a single supplier
receiving monopoly proﬁts to pure competition with many competing ﬁrms
pricing at marginal cost [Varian (1992)].

One indicator of competition often examined is market concentration
[Scherer and Ross (1990)]. When a ﬁrm holds a very small market share, it

typically experiences more competition. Conversely, when a ﬁrm holds a

19
large (concentrated) market share, it often experiences less competition.

Scherer and Ross (1990) stated that true monopolies do not exist in United
States manufacturing today and that the instance of near-monopolies is
rapidly declining. Conversely, Scherer and Ross maintained that nearly half
of all industries can be characterized as oligopolistic, which they indicate
exists where the four leading ﬁrms control 40% or more of market share.

Competitive pressures usually lead ﬁrms to become more aggressive in
their attempts to gain market share from their competitors. According to
Scherer and Ross, the preferred tool for assessing market structure is the
Herﬁndahl-Hirschman Index (HHI). The HHI considers both ﬁrm numbers
and inequality by squaring market shares and summing them, thus weighing
more heavily the values for large ﬁrms.

As competition increases, neoclassical economics indicates that
proﬁts decrease. Firms trying to improve their proﬁtability can do so by
selling more units (as long as each unit yields a positive return), raising the
prices (which is difﬁcult under increasing competition because competitors
would maintain lower prices and gain market share), or lowering costs. If
ﬁrms can capture market share (especially a large share) through lower
prices, they sometimes can increase total proﬁts even though unit margins
may be lower.

In ﬁrms operating under pure competition, managers need accurate
costs of materials, labor, and overhead for each product in order to deﬁne the

product’s proﬁt-maximizing price. As the proﬁtability of each product

20
becomes clear, managers focus on the more proﬁtable products and redirect

the consumption of resources away from less proﬁtable products and toward
more proﬁtable products.

Neoclassical theory's strength for the current study lies in explaining
the effects of competition. Competition imp acts the prices set by
management. Where prices and proﬁts are high, other ﬁrms try to enter the
market; where entry is successful and no collusion occurs, prices and proﬁts
usually lessen. When prices and proﬁts are very low, potential entry is often
avoided, but the proﬁtability depends on understanding and controlling
costs. Managers attempt to ﬁnd a price low enough to forestall the entry of
competitors yet high enough to make acceptable proﬁts [Varian ( 1990)]. One
claim of FCA advocates is that inaccurate cost allocations can lead a ﬁrm to
price inappropriately, thus ﬁ'ustrating management's attempt to ﬁnd the
proﬁt-maximizing price [Cooper (1986)].

This study used several market-share-based measures to assess
competition: typical product market share, biggest market share of a
product, industry competition as determined with the HHI, and the portion of
the HHI held by the ﬁrm. In addition, the study used respondent subjective
evaluations of the level of competition in the industry, the agressiveness of
major competitors, whether competition is forcing the lowering of prices or
the decreasing of proﬁt margins, the level of foreign competition, and the
level of domestic competition. As explained in the methodology and results

sections, the competition measures failed to form a cohesive factor, so

21
individual measures were regressed separately against FCA method

implementation.

Life-Cycle Stages

Porter (1980) considered the industry or product life cycle, ﬁrst
presented by the Boston Consulting Group (1968), to be a primary result of
competition; it often forms the basis for setting key strategies. BCG’s theory
recognized that many aspects of an industry or product change over time,
going through four stages: introduction, growth, maturity, and decline. The
usefulness of cost information varies over the cycle; the current study used a
guideline that accurate cost information is relatively unimportant in the
introduction stage, slightly important in the growth stage, and very
important in the maturity and decline stages. However, costs are used in the
maturity stage to remain competitive; in the decline stage, they are aimed
more at maximizing proﬁts until exiting hour the industry.

Considerable research and academic discussion has examined how
competition, prices, and proﬁts change as a product goes through the various
stages of the product life cycle. Wernerfelt (1985) found that prices decrease
early in the life cycle, then increase later as market share is given up for
increased proﬁts. Claiming that ﬁrms go through life cycles in a manner
similar to products, only over a longer period of time, Anthony and Ramesh

(1992) used annual dividend as a percentage of income, percent sales growth,

22
and age of ﬁrm to classify ﬁrms in the Compustat data base into growth,

maturity, and declining stages, then used the results to test the association
between accounting performance and stock prices.

The present study used product life cycle (a weighted combination of a
ﬁrm’s products in each stage) and ﬁrm life cycle as measures of competition.
As explained in the methodology and results sections, competition failed to
form any cohesive factor, so competition measures were regressed directly
against FCA method implementation.

Some ﬁrms choose not to follow the general pattern of the entire
product life cycle. Richardson and Gordon ( 1980) point out that some ﬁrms
choose to remain innovators (diﬂ'erentiators), producing only during the
introductory and growth stages while prices and proﬁts are high, then
withdrawing from the market when margins decrease. Others focus on
getting down the learning curve quickly to obtain market share for the late
growth, maturity, and decline stages when their lower costs and high volume
bring reasonably high proﬁts. These ﬁrms choose to become cost leaders.
Hill (1991) and Murray (1988) show that ﬁrms can achieve both product
diﬁ'erentiation and low-cost leadership. Firms have the capability to manage
the product life cycle when viewed across an array of products, so the
management of the product life-cycle affects the level of competition
experienced by the ﬁrm.

This study used two measures in relation to life-cycles: a weighted-

average product life-cycle and a ﬁrm life-cycle (a modiﬁed version of the

23
Anthony and Ramesh model). As explained in the methodology and results

sections, neither life-cycle measure formed a factor with other competition

measures or with the other life-cycle measure.

Strategic Positioning

Structural analysis of industries provides a framework for looking at
the sources and degree of competition within an industry. Having assessed
the level of competition, the ﬁrm may strive to position itself to take
advantage of its own strengths. Porter (1980) discusses several forces
constantly at work on the ﬁrm to reduce product prices or increase product
costs, thus eroding product proﬁtability and increasing competition. Porter
showed that the level of competition depends on many aspects besides
marginal cost.

The strategic aspects of interest to the current study include 1)
whether a ﬁrm focuses on being a product diﬁ'erentiator, a low-cost leader, or
both; 2) how important cost is in the ﬁrm's setting of prices; and 3) how
intense cost reduction eﬂ'orts are. Each strategic aspect is further discussed

below.

24
Product Diﬁ'erentiator vs. Low-Cost Leader

Firms have a strategic choice of being a product differentiator, a low-
cost leader, or both. Porter (1980) explained the ramiﬁcations of this choice.
If a product is undifferentiated from that produced by other ﬁrms, prices may
be lowered in order to attract customers, creating increased competition.
Because new entrants to an industry bring new capacity (and sometimes
substantial resources) and ﬁght for market share, the preexisting industry
may have to lower prices, increase costs (such as redesigning the product,
engaging in more advertising, paying higher sales commissions, etc.), or both
to maintain market share. Two approaches (not mutually exclusive) to
dealing with the increased competition from new entrants are product
differentiation and cost leadership.

Under product diﬂ'erentiation, the seller attempts to make the buyer
perceive the product as being unique by changing the product, by creating
different support services, or by convincing the customer that it is diﬂ'erent
through intensive advertising, special packaging, etc. Under product
diﬂ'erentiation, the buyer sees the seller as the only source of the product
bundle.

Porter ( 1980) advised that, to be successful as a diﬁ'erentiator, the ﬁrm
must ﬁnd ways of diﬂ'erentiating that lead to a price premium greater than
the cost of diﬁ‘erentiating. A diﬁerentiator cannot ignore its cost position,

else it loses the beneﬁt of higher price to the higher costs incurred. The

25
research question suggests that diﬁ‘erentiators become interested in more

accurate cost allocations as the gap between product prices and product costs
narrows or as cost reductions are sought in order to widen the gap.

When ﬁrms pursue cost reductions to widen the proﬁt gap, ﬁrms
sometimes attempt to become low-cost leaders by ﬁnding and exploiting all
possible sources of cost advantage. To be successful as a cost leader,
however, Porter (1980) said the ﬁrm must become the cost leader, not one of
several vying for cost leadership; otherwise, price competition becomes too
ﬁerce and proﬁtability disappears. As the low-cost leader, the ﬁrm has costs
low enough to still make proﬁts, yet has the price low enough to discourage
new entrants that do not yet have such a cost advantage. Still, the cost
leader must achieve parity or proximity in the bases of differentiation
relative to its competitors, or it will not perform well.

Costs are important to both diﬂ'erentiators and low-cost leaders, but
reductions are more extreme with the cost leader, whereas the diﬂ'erentiator
reduces costs only to the point where proﬁts are maximized without losing
diﬁ‘erentiation.

This study used six measures to record aspects of diﬂ'erentiation and
low costs: new product innovation, new process innovation, production of
diﬂ'erentiated products, focus on selling high-proﬁt-margin products,
production of commodity products, and position as industry low-cost leader.

As explained in the methodology and results sections, a subset of these

26
measures were used in factor analysis and regression analysis to test

whether differentiation and low-cost leadership inﬂuence FCA method

implementation.

Importgce of Costs in Setting Prices or Maintam_1_n' ' g Desired
Proﬁtability

A ﬁrm’s choice of pricing process aﬂ'ects its strategic position. Price
generally is determined in three ways: 1) ﬁnding the highest price that the
market will bear (monopoly pricing), 2) adding some markup to costs
incurred, or 3) accepting the competitive market prices. The latter two
methods depend on having accurate costs — either to serve as the base for
calculating price or to determine whether sumcient proﬁts can be obtained at
the market-set price. Thus, pricing methods also inﬂuence the need for FCA
information. Bromwich (1990) points out that prices can be set either at
what the market will bear or at a level covering full long-term product cost
(thus erecting a barrier to entry for competitors). Monopoly pricing is fairly
independent of product cost; mat-based pricing depends directly on costs; and
market-set pricing requires costs be contained at a level insuring adequate
proﬁts. Monopoly costing is typical of less competition, whereas cost-based
pricing and market-set pricing are typical of more competition and are more

likely to generate a need for FCA methods.

27
To examine the relationship of pricing method to FCA method

implementation, this study used a measure of how much the plant uses costs
in setting prices. As explained in the methodology and results sections, using
costs as a basis for setting prices did not form a cohesive factor with other

strategy measures or with the group sum.

Inmnggz' of Cost Reduction Eﬁ‘og

Besides the product diﬂ‘erentiator versus low-cost leader issue and the
choice of pricing method, a ﬁrm’s strategic position also depends on the
intensity with which it pursues cost reductions. According to Porter (1980),
cost reduction eﬂ'orts are more intense among ﬁrms experiencing greater
competition. When a ﬁrm needs cost reductions to survive, it will exert more
eﬁ'ort to determine costs accurately than would a ﬁrm merely trying to widen
proﬁt margins.

This study used two measures to assess cost-reduction eﬂ'orts:
intensity of cost reduction efforts and use of target pricing and target costing
to insure proﬁtability. The cost-reduction measures also failed to form a

factor together or with other measures.

28
Managerial Support

Implementing an FCA method faces many of the same problems as
implementing any system; among the leading problems would be a lack of
managerial support. The systems management literature suggests
implementation is more likely with managerial support, as evidenced 1) by
having an effective champion who can rally managerial and organizational
acceptance, 2) by obtaining top management support, and 3) by having user
support. Since an FCA methods are systems, successful implementation of
FCA methods could be inﬂuenced by managerial support.

Kanter (1983) , Reich and Benbasat (1990), Beatty (1992), and Yellen
(1993) maintain that perhaps the most important antecedent to a successful
information system is a "champion” for the new system. Beatty (1992)
examined successful and unsuccessful implementations of advanced
manufacturing technologies at ten companies; she found that none of the
companies were able to achieve their goals without a champion. .

In clarifying the role of a champion, Beath (1991) pointed out that
champions are diﬁ'erent from ”sponsors,” who provide the authority and
funding for the implementation. Lanford ( 1993) explained that a champion
ensures that management sees the vision of the new system and understands
the beneﬁts. Beatty (1992) saw the role of the champion as facilitating
others' conversion to the new methods, which helps build trust, acceptance,

and commitment in those who are to use the syStem. The present study

29
captured the respondents’ assessments of the effectiveness and

organizational level of the champion.

On a different front, Yellen (1993) said that two of the major
difﬁculties in implementing systems arise when managers do not fully
support the systems and when users are not sufﬁciently committed to them.
Compton (1994) stated that the ﬁrst and most important task in
implementing an ABC system is obtaining a commitment from top
management, but then pointed out that eﬁ'ective use also requires
endorsement and support throughout the user community. Martinsons
(1993) indicated that management endorsement institutionalizes an
innovation and helps persuade others of its viability; such support is deemed
essential to gain resources and reward results.

The present study used eight measures to capture the respondents’
assessments of managerial support. Upper management support was
recorded as the support of the CEO, support of the Controller, and support of
other top management. Support of the users was recorded as support of
middle management, support from Accounting, and support from Marketing.
Assessment of the champion was recorded as the organizational level of the
champion and the effectiveness of the champion. As explained in the
methodology and results sections, a subset of the managerial support

measures were used in factor analysis and regression analysis.

30
Cost-Beneﬁt Theory

The ﬁnal theory of interest in the current study is cost-beneﬁt theory,
which drives the whole issue of whether ﬁrms seek more accurate costs.
Cost-beneﬁt theory is based on the assumption that ﬁrm decision makers
make rational choices. The real world, being full of choices, operates
eﬂiciently in accordance with the information and options available [Cheung
(1992)]. Cost-beneﬁt theory maintains that managers will avoid expending
resources and effort unless they feel a net gain can be obtained.
Implementing an FCA method costs money for software purchase or
programming, retraining of the users, etc. If the perceived beneﬁts of
implementing an FCA method do not suﬂiciently outweigh the costs,
managers should not implement the FCA methods.

Therefore, the ﬁrst question for a ﬁrm to consider before implementing
FCA methods is whether sufﬁcient beneﬁts will .arise from implementing
FCA methods. In some situations no beneﬁts are apparent; in others, strong
beneﬁts occur. Homgren, Foster, and Datar (1994), in their presentation on
plantwide rates versus departmental rates, discuss the situations where FCA
methods provide beneﬁts. They maintain that ﬁner allocation rates are
useful only where resource consumption patterns (by products or by
organizations) are diverse and uncorrelated with the allocation base. Where
the products and organizations each exhibit the same resource consumption

pattern and that pattern is well correlated with the allocation base, the

31
plantwide rate should provide the same beneﬁts as more reﬁned rates. In

the latter situation, the ﬁrm has less incentive to implement more reﬁned
rates.

Given diversity of resource consumption patterns, managers still need
to determine whether the perceived beneﬁts of FCA methods justify the
additional costs. As indicated by Yourdon (1989), all ﬁnancial costs that may
result from implementation will need to be examined, including the costs of
analyzing, planning, building, installing, training, etc.

Acquisition and installation costs include the actual costs of building
or obtaining the hardware, software, and training needed for the new
techniques. These costs, together with normal operating costs, are
traditionally considered in ﬁnancial analyses and capital budgets;
managerial accountants, MIS staﬁ‘, or top management frequently base their
decisions about implementing systems on the outputs from ﬁnancial analyses
and capital budgets. Campi (1992) points out that focusing too intently on
short-term ﬁnancial measures can inﬂuence a ﬁrm’s management to delay or
avoid implementation.

In addition to these costs, however, the ﬁrm may incur other fairly
signiﬁcant costs. As implied by Jensen and Meckling (1992), the obstacles
also include the efforts required in order to learn, adjust to, and use the new
techniques. Time and effort would be spent negotiating within the ﬁrm to
have the new information used, measuring the costs themselves on an

ongoing basis, dealing with the changes in prestige of products as a result of

32
allocation changes (if improper allocation is occurring), monitoring the

performance of managers using the information, organizing the ﬁrm to use
the information (including any changes in policies and practices to make the
appropriate level of detail available), and so on. Thus, the total cost of
implementing FCA methods is likely to be signiﬁcantly higher than the
acquisition and installation cost.

While little or no research has been performed directly on the costs of
FCA methods implementations, FCA methods are really a subset of
information systems. Jensen and Meckling (1992) focus on the cost-beneﬁt
aspects of obtaining, processing, and using information; their ﬁndings are
applicable to FCA methods.

Jensen and Meckling examine the role that knowledge plays in
organizational decision making, with special emphasis on the "bounded
rationality” caused by the physical limitations speciﬁc to each individual.
Because humans have limited mental capability, gathering, storing,
processing, transmitting, and receiving knowledge are costly activities which
use the time and eﬁ'ort of the individuals involved. Because of this cost,
decision-makers do not seek out information unless it is pertinent and
valuable.

Under cost-beneﬁt theory, if ﬁrms do not implement ﬁner allocations,
then the total costs associated with FCA methods must exceed the perceived

beneﬁts which would accrue from their implementation.

33
The current study captured information about obstacles, both ﬁnancial

and nonﬁnancial, which managers in the plant might consider as negating
the beneﬁts which could derive from FCA methods. Measures of ﬁnancial
obstacles included cost of adequate planning, cost of bringing in consultants,
cost of changing the information system, cost of training, cost of running a
dual cost system for some period of time, and the cost of gathering and
processing more detailed data. Measures of nonﬁnancial obstacles included
managerial resistance to making the change, lack of high-level sponsor, focus
on short-term ﬁnancial measures, diﬂiculty of quantifying costs and beneﬁts
of the change, and the difﬁculty of communicating the results of planning

throughout the organization.

CHAPTER HI

CONCEPTUAL MODEL AND HYPOTHESES

OVERVIEW AND CONCEPTUAL MODEL

This study examined why ﬁrm managers choose to implement FCA
methods; the study measured several ﬁrm characteristics that might
inﬂuence the implementation decision. The initial model, as shown by the
solid arrows in Figure I, assumed that ﬁve factors (complexity, competition,
strategy, managerial support, and implementation obstacles) independently
inﬂuence the implementation choice. Two possible interactions were
foreseen as occurring between competition and strategy and between
implementation obstacles and managerial support. (Although a few of the
measures showed interacting correlations, the use of fewer measures in an
orthogonal varimax rotation factor analysis precluded the possible
interactions in the main analysis.)

Each of the ﬁve hypotheses related an area of ﬁrm characteristics to
the actual level of FCA method implementation. Firms may avoid the
expense of fully implementing FCA methods because collecting and

processing data (as well as changing organizational practices) is not costless.

34

35

 

F'Igu'e 1. BasicModelof Hypdhesized Fblationships

 

 

[add I W
muslin-c b

.11 ‘..xLL.L_L_.'_

 

 

mu
m
an

-1- , o -
-AM“_..1*~_

\‘~\ an m l M»
are: am --- We

 

 

 

 

 

 

 

Avoidance is more likely to occur where one or more of the following
conditions are met: 1) resource consumption is less complex (correlated with
the allocation base and not diverse), 2) competition is low, 3) speciﬁc
strategies (high product differentiation, non-cost-based pricing, and little
interest in cost reductions) are followed, 4) little managerial support for FCA
methods exists, and 5) major or costly obstacles hinder implementing the
FCA methods. The prediction underlying the study was that, once perceived
beneﬁts associated with FCA methods outweigh the perceived costs, FCA
methods will be implemented. If a champion arises in a complex, competitive

plant which follows speciﬁed strategies, FCA is more likely to be needed; if

36 .
suﬂicient managerial support exists and implementation obstacles can be

overcome, FCA methods are more likely to be implemented. Each of the ﬁve

hypotheses is further examined below.
COMPLEXITY HYPOTHESIS

Before implementing any FCA method, ﬁrm managers consider the
beneﬁts arising from such an implementation. In some situations, no
beneﬁts are apparent; in others, strong beneﬁts occur. Homgren, Foster, and
Datar (1994), in their presentation on plantwide rates versus departmental
rates, discuss favorable conditions for FCA methods. Finer allocation rates
are useful where resource consumption patterns (by products or by
organizations) are diverse and uncorrelated with the allocation base—that is,
where complexity occurs. Complexity may be indicated in producing a
variety of substantially diﬂ'erent products, using many routings (either to
make a single product or to make numerous simple products), requiring long
and expensive setups, requiring many types of labor (with different hourly
rates), frequently incorporating production technology changes, etc. In this
study, ten measures were used to record attributes of complexity for analysis.

Where the products and organizations in a plant each exhibit the same
resource consumption pattern and that pattern is well correlated with the
existing allocation base, the plantwide rate (a gross allocation method) may

provide the same accuracy in cost allocations as do departmental rates (an

37
FCA method). In the latter situation, the plant has little incentive, in terms

of cost accuracy, to implement departmental rates (although departmental
rates or other FCA methods may offer additional valuable insights about the
plant’s cost drivers).

The ﬁrst hypothesis relates the complexity of resource consumption by
products with the likelihood of implementing FCA methods. Three levels of
complexity, represented by four different cost allocation methods, are
considered. The lowest level, yielding only a gross allocation, is represented
by the pattern shown in Figure 2, where consumption patterns are not
diverse and are correlated with a single driver. The allocation method
suggested at the least complex level is the single-cost-driver allocation
method.

Two consumption patterns describe the intermediate complexity level.
The ﬁrst occurs when a product coming through a department consumes an
equal percentage of each departmental resource but the pattern in one
department is not the same as the pattern in other departments. (See
example in Figure 3's upper panel.) The second occurs when consumption
between departments is not diﬂ‘erent, but the consumption patterns of
diverse types of resources diﬁ‘er. (See example in Figure 3's lower panel.)
Two allocation methods suggested for the intermediate complexity level are
the departmental allocation method and multiple plantwide allocation

method.

38

 

Figure 2. Least Complex Resource Consumption Pattern
All Resources Used in Same Portions as a Single Cost Driver

oure1 re3

 

* Any driver could work equally well in this situation.

 

 

 

The highest level of complexity occurs when each product coming
through a department consumes diﬁ'erent percentages of the resources
associated with diverse activities regardless of the departments where they
are located. (See Figure 4 for an illustration of this situation.) The allocation
method suggested for the highest complexity level is the activity- or process-
based allocation method.

A ﬁrm may have diﬂ'erent plants operating at different levels of
complexity, so more than one resource consumption pattern may be pertinent
to a ﬁrm or even to a single plant. Conceivably, one portion of the
organization may experience a more complex pattern, generating a need for
FCA data within just that portion. The organization as a whole, however,
may exhibit less complex resource consumption patterns and not need more

complex FCA methods.

39

 

 

ﬁgmeii. Moderately Corrpiex Resorm Consurmtion Patterns
MMhSmanhth-mﬁrﬂmsssswcostm

Product

 

Each depatment uses one driver. may be different between depatments.

mammmwhmm
.SldeWM-sMedWhGup

 

Fbsource group throughout plant uses single driver, different groups use
different drivers.

 

 

40

At complexity levels higher than that appropriate to a single plantwide
allocation method, the inaccuracies of cost allocation might cause products to

be assigned incorrect costs. Where the threat of improper allocation is high,

 

Figure 4
Most Complex Resource Consumption Pattern
Each Activity/Process in Each Department Used In Different Portions

Product ent X

65. 4

45%

C
ota '5; _ 1 :2: . .-;
" ' a

Driver‘ Cur Drilled Bent Primed Cut

 

' Each activity uses only one driver, which is speciﬁc to that activity. The drivers
shown are only examples.

 

 

 

ﬁrms are more likely to perceive the beneﬁts of FCA methods as outweighing
the implementation costs. In view of the preceding discussion, Hypothesis 1

relates complexity to implementation of FCA methods:

H1: As consumption of resources by various products becomes increasingly
complex, plants are more likely to implement FCA methods. That is,

greater complexity increases FCA method implementation.

41
COMPETITION HYPOTHESIS

FCA information may be especially useful to managers who need
precise cost information in order to be competitive. Knowing total costs may
become important to managers as their ﬁrms’ margins decrease under
increasing competition; this need for knowledge may lead to a desire for more
accurate costs. The second hypothesis concerns the relationship between
competition and the implementation of FCA methods. Three aspects of
competition were investigated: 1) product or ﬁrm life-cycle stage, 2) market
share, and 3) respondents’ assessments of competition.

Product and industry life-cycle theory recognizes that many aspects of
an industry or product change over time, going through four stages:
introduction, growth, maturity, and decline. Typically, competition is low
during the introduction stage, increases progressively during the growth and
maturity stages, then declines somewhat during the decline stage. In this
study, ﬁrms and their products were classiﬁed into life-cycle stages (two
separate measures); the measures were assigned competition values
consistent with theory and were regressed individually against
implementation of FCA methods.

As competition increases, market share typically decreases. In this

study, four market-share-based measures were used to record the market-

42
share aspect of competition and were regressed individually against

implementation of FCA methods.

The pe0ple within a plant or ﬁrm often have assessed the level of
competition in their market. In this study, ﬁve measures were used to record
the respondents’ subjective opinions on competition and were regressed
individually against implementation of FCA methods.

Hypothesis 2 relates competition to implementation of FCA methods:

H2: Greater competition increases FCA method implementation.

Hypothesis 2 is broken down to match the three approaches to assessing

competition:

Hm: When competition is greater in the product or plant life cycle, FCA

method implementation increases.

sz: As market share decreases, FCA method implementation increases.

H20: As the perception of competition increases, FCA method

implementation increases.

43
STRATEGY HYPOTHESIS

Almost every viable ﬁrm coordinates its activities according to a
conscious strategy. The third hypothesis concerns how the ﬁrm’s and plant’s
strategies inﬂuence the implementation of FCA methods. The strategic
aspects considered include 1) whether a ﬁrm focuses on being a product
diﬁ‘erentiator or a low-cost leader, 2) how important cost is in the ﬁrm's
setting of prices, 3) and the intensity of cost reduction efforts.

As explained by Porter (1980), a ﬁrm strategy of being a product
diﬁ'erentiator leads to little emphasis on costs, other than to make sure price
is higher than the total cost and to enhance proﬁts -- survival is not an issue.
If, however, the ﬁrm strategy is to attain low-cost leadership, great
importance is placed on having accurate costs -- inaccurate costs could result
in ﬁrm failure. While both product diﬁ'erentiators and low-cost leaders could
decide to implement FCA methods, the low-cost leader is more likely to
implement such methods. In this study, six measures were used to record
aspects of differentiation and low-cost leadership (thus allowing for a plant to
be pursuing both differentiation and low-cost leadership); the measures were
used in further analysis of the relationship to implementation of FCA
methods.

Pricing methods also inﬂuence demand for FCA. Bromwich (1990)

points out that prices can be set either at what the market can bear or at a

44
level covering full long-term product cost (thus erecting a barrier to entry for

competitors). Pricing at what the market can bear is more typical of less
competition, whereas cost-based pricing (or even market-set pricing) is
typical of more competition and is more likely to result in implementation of
FCA methods. In this study, the degree to which prices were based on costs
were recorded in a measure which was individually regressed against
implementation of FCA methods. I

According to Porter (1980), cost reduction efforts are more intense
among ﬁrms experiencing greater competition. In this study, the intensity of
cost reduction eﬁorts and the degree which target pricing is being used
(which often forces cost reductions) were recorded and individually regressed
against implementation of FCA methods.

In view of the preceding discussion, Hypothesis 3 relates ﬁrm

strategies to expected use of FCA methods:

H3: As ﬁrms 1) move away from product diﬁ'erentiation, 2) use costs as the
basis for product pricing, and 3) increasingly pursue cost reductions,
those ﬁrms are more likely to implement FCA methods. That is,

certain types of ﬁrm strategy increase FCA method implementation.

Because Hypothesis 3 actually represents multiple strategies, it is

further separated into three subhypotheses:

45

Ha: Low differentiation in product and process increases FCA method

implementation.

Hsb: Basing prices on costs increases FCA method implementation.

Hsc: Intense cost reduction eﬁ'orts increases FCA method implementation.

MANAGERIAL SUPPORT HYPOTHESIS

The fourth hypothesis concerns the relationship between managerial
support for implementing FCA methods and the actual implementation of
FCA methods. The study examined whether widespread support for
implementing FCA methods increases the likelihood of implementation; to
record managerial support, eight measures were used. The measures cover
the respondents’ perceptions of support at various levels of management and
among certain users. The measures also identify the level and effectiveness
of the champion. Hypothesis 4, therefore, relates managerial support to the

implementation of FCA methods:

46
H4: As managerial support for FCA methods increases, FCA

implementations become more likely. That is, managerial support

increases FCA method implementation.

IMPLEMENTATION OBSTACLES HYPOTHESIS

The ﬁfth hypothesis concerns the relationship between the strength of
implementation obstacles and the implementation of FCA methods. The
higher the incurred implementation costs (ﬁnancial or nonﬁnancial), the less
likely that FCA methods implementation will be completed. Compared to the
ﬁrst four hypotheses that revolve around increasing beneﬁts, this hypothesis
focuses on decreasing overall costs.

If decision makers view the beneﬁts of FCA knowledge obtained from a
speciﬁed FCA method as outweighing the costs associated with gathering,
processing, assimilating, and using the data, the FCA method will be
implemented. Otherwise, it will not be implemented. The current study
suggests that individuals in ﬁrms will make the effort to implement FCA
methods only if that effort is viewed as the least costly means of providing
desired beneﬁts. To record implementation obstacles, the study used 1 1
measures. Six measures recorded ﬁnancial obstacles, whereas the other ﬁve

measures investigated nonﬁnancial obstacles.

47
Hypothesis 5 relates these obstacles to the likelihood of FCA method

implementation:

H5: As implementation obstacles become more numerous and stronger,
managers will be less likely to implement FCA methods. That is,

implementation obstacles decrease FCA methods implementation.

The ﬁve hypotheses presented above explain the overall characteristics
this study examined in relation to FCA methods implementation. The
details of what measures were used and how the research was conducted are

related in the next section.

CHAPTER IV

METHODOLOGY

APPROACH

Research was directed along three fronts: ascertaining the level of
competition from externally reported data; ascertaining the levels of
complexity, competition, strategic measures, managerial support, and
implementation obstacles from plant surveys; and determining from survey
responses the degree to which ﬁner cost allocation techniques had been
implemented.

External information on competition was gathered from the Standard
and Poor's Compustat Industry data base, Moody's manuals [Moody's
Investors Services (1995)], CorpTech Directory of Technology Companies
[Corporate Technology Information Services (1995)], and Manufacturing
USA [Darnay (1994)]. Compustat data were used to form the pool for ﬁrm
selection. Whether the ﬁrms from Compustat adequately represented the

industry as to number and market share was veriﬁed using

48

49
Manufacturing USA. The data obtained were used to identify and classify

industries and ﬁrms as to the levels of competition being experienced

Following ﬁrm selection, plants within the ﬁrms were surveyed about
ﬁrm and industry characteristics and practices. Information gathered
included data on complexity, competition, strategic measures, managerial
support, implementation obstacles, and cost allocation schemes employed
Surveys of the ﬁrst mailing were addressed to plant managers. Since the 29
surveys received from the ﬁrst mailing mainly came from controllers, surveys
of the second mailing were addressed to controllers or head accountants.

The data gathered were analyzed using descriptive statistics, manual
examination of correlations to reduce the number of measures, principal
components factor analysis to establish independent factors, Cronbach’s
Alpha to determine item reliability (which further reduced the number of
measures), factor analysis to obtain factor scores, and multivariate regression
analysis to test the hypotheses. These analyses tested whether survey and
calculated items adequately represented the ﬁve independent variables and
whether signiﬁcant relationships existed between the dependent and

independent variables.

50
SAMPLE SELECTION

The current study used a sample drawn from a number of industries
for which market structures indicated various levels of competition (degrees
of monopoly power, oligopolistic power, or pure competition.) Sample
selection began with a search of the Standard and Poor's Compustat Industry
data base using a selected range (3000-3999) of primary Standard Industrial
Classiﬁcation (SIC) codes. While ﬁrms in one category often contain
business from other SIC codes, treating the entire ﬁrm by its primary SIC
code offers the best approach to using the data that are publicly reported and
is a common practice (see Manufacturing USA [Darnay (1994)], the 1987
Census of Manufactures [US Department of Commerce (1992)], and
Anthony and Ramesh [1992] .) Using the primary SIC code identiﬁes each
ﬁrm with one and only one SIC code. Working at the 4-digit level provides a
balance between underdeﬁning and overdeﬁning industries with comparable
ﬁrms. The ﬁrst pass through the data base selected 2,091 ﬁrms with 1992
sales data.

Next, ﬁrms were eliminated if necessary data were not available for
the time period in question. The desired data included the following from the
Compustat Industry annual data base: item 117 (Sales (Restated)) for the
years 1986-1992 and items 18 (Income Before Extraordinary Items) and 21

(Dividends - Common) for the years 1987-1992, as well as item 12 (Sales

7 51
(Net)) for 1992. In addition, the ﬁrm's ﬁrst year of operations (BYEAR) was

taken from Moody's manuals [Moody's Investors Services (1995)] or CorpTech
Directory of Technology Companies [Corporate Technology Information
Services (1995)] to allow a calculation of ﬁrm Age. Failure to ﬁnd a BYEAR
resulted in deleting 272 ﬁrms, while other data missing from Compustat for
one or more years resulted in deleting 609 ﬁrms. Thus, the calculations
requiring Compustat data and BYEAR were based on 1,209 ﬁrms, from
which the survey ﬁrms and their plants were also drawn.

The study examined industries with various structures, as identiﬁed
through HHI values calculated from the 1992 values for net sales from
Manufacturing USA [Darnay ( 1994)]. Some industries selected for survey
had high HHIs, whereas others had low HHIs; in addition, selected
industries had to contain three or more ﬁrms previously identiﬁed on the list
from Compustat. Besides the HHI itself, the ﬁrm’s portion of its industry
HHI was calculated to indicate the level of competitive pressure experienced
by a ﬁrm within its SIC code.

In selecting ﬁrms, the attempt was made to insure that the industry
was adequately represented as far as the number of ﬁrms and their sizes
since both aspects aﬂ'ect the HHI calculated. (Note that Manufacturing USA
indicates some ﬁrms as having slightly diﬁ'erent primary SIC codes than does
Compustat; in these instances, the Manufacturing USA codes were used.)

Since diﬁ‘erent plants within a ﬁrm could use diﬁ‘erent allocation methods,

52
surveys were directed at 1,071 speciﬁc plants chosen ﬁom Dun’s Industrial

Guide: The Metalworking DirectoryTM 1994/95 (Dun & Bradstreet 1994).
For ﬁrms with numerous plants, no more than eight plants were chosen;
when such selections were necessary, the plants were chosen at random. The
number of plants to which surveys were sent was limited by the budget
available for printing and mailing. Names of individual plant managers and
plant addresses were taken ﬁ'om Dun’s Industrial Guide. Although Dun’s
listed only plants with metalworking, plants across most manufacturing SIC
codes are represented due to the prominence of at least some metalworking
in most plants.

Following ﬁrm selection, ﬁrms were surveyed about ﬁrm and industry
characteristics and practices. To encourage response, three incentives were
offered: ﬁrst, the respondent could request the general results of the survey;
second, ﬁrms submitting multiple surveys were offered the results coming
from each plant in the ﬁrm, provided that each plant agreed; and third, two
$75 certiﬁed checks were sent to two respondents selected by a random
drawing.

The ﬁrst-mailing survey was addressed to the plant manager; the
cover letter requested that the survey be completed by an appropriate
individual, such as the plant manager or the controller. The respondent
needed a good understanding of the plant’s products, processes, and costs, as

well as of its overhead allocation practices. Since 25 of 38 ﬁrst-mailing

53
responses were from controllers or head accountants, the second mailing one

month later was addressed to such individuals, which resulted in 87
additional responses, one of which was ﬁ'om a plant which had answered the
ﬁrst mailing while the second mailing was in transit.

Of the 1,071 plants on the mailing list, 116 had invalid addresses and
22 responded that they had no real manufacturing activities, reducing the
potential pool of respondents to 933. Of the 125 completed surveys, 15 were
found to contain self-contradictory information about the dependent variable.
Since one of these was the duplicate response, it was dropped from the
sample. The remaining 14 respondents were contacted and the discrepancies
resolved; thus these plants were retained in the sample. The 124 responding
plants served as the study's sample, representing a response rate of 13.3
percent. An additional 2 1 plants responded that it was against their ﬁrm’s
policy to answer surveys; if these plants are removed, the response rate

becomes 13.6 percent.

RESEARCH DESIGN AND DATA MEASURES

The cross-sectional research design of this study used several
measures to record characteristics hypothesized as inﬂuencing FCA method
implementation. A few of the measures were calculated from Compustat
data, but most were gathered from the ﬁrm survey (see the Appendix for a

copy of the survey). This section discusses the individual measures, their

54
ranges, and the source from which they were obtained. For ﬁve of the

measures, explanations of how they were calculated are also given.

After initial survey development, the survey was reviewed by nine
former Executive MBA participants, six of whom had constructive comments
to improve it. Following that, individuals in two local plants site-tested the
survey and provided additional feedback. The suggestions improved both the
wording and content of the survey.

Several demographic items of information were gathered for each
responding plant; two items proved to be of interest: 4-digit SIC code and
years in the company. Since the survey was developed to represent many
industries, the 4-digit SIC code was gathered for each responding plant. If
Porter (1980) is correct about the competitive forces acting on industries, it
was possible that some industries would show a greater tendency to
implement FCA methods than other industries. The number of years the
respondent had been in the company should have little to do with FCA
method implementation, but it was highly correlated with FCA method
implementation. Both of these items are discussed further in the results
section.

The dependent variable, OVERALL, was an 1 l-point Likert scale
wherein the respondent indicated the plant’s level of implementation of more
detailed cost allocation methods. A ranking of “ 1” indicated the overall level
of implementation to be that of single plantwide allocation methods; a

ranking of “1 1” indicated the overall level of implementation to be that of

55
very detailed methods. Since a plant could have implemented multiple

methods in various portions of the plant or for diﬂ'erent purposes, the overall
implementation level could fall anywhere from 1 to 1 1.

Tables 1 through 5 present the measures associated with complexity,
competition, strategy, managerial support, and implementation obstacles,
respectively. Each table shows the measure name, a description of the
measure, the source of the item (survey item number or calculation source),
whether the measure is normal (shown as recorded by the respondent) or
reversed (making low responses high and high responses low), and the range
of values the measure may take. The item numbers indicated refer to the
survey (see the Appendix) and show the major item number and, where
needed, a letter to indicate which question under that item. Thus, item 4d is
the fourth question under item 4.

The meanings of most items shown in the tables are clear since they
represent either 7 -point Likert scales, Likert reﬂections (8 minus the value),
or simple calculations (described in the tables). Seven calculations (Number
of Product Lines, Typical Product Market Share, Largest Product Market
Share, Industry Competition Level, Firm Portion of the Industry HHI, Life-
Cycle Stage, and Plant Life-Cycle Stage) are more involved and are discussed
further.

Table 1 shows the ten measures used to record complexity information;
the measures record manufacturing attributes which show consumption

patterns to be diverse and uncorrelated with the allocation base. The

 

8833a

Am 83H

4888.388 888 .888 8.83
8888.388 8 888 8.“: 838.3 2: .8 883 on» 833
3 .838 2: 38838 :SmeQEoD 3.80 855 838.8%

883880

 

6A

8882

an as:

88 .8 83.89838 a .88 3383. 8.88 888.58
8.8 $883 38 888 .888 €888.58 8.8 $883
38 833 8 38 85 38838 comugoeﬁ 8.382% 882

88883

 

Nan

8888M

at. as:

.328 new a 23 888
888 .883 68a 8883 a 3 8838 88 328 no” a 8
8.8.88 833 25 8.83 3 838 26 38838 88% cox.

gaming

 

352

on 883

.338 8 888 8 83.83 8888““ 888 8%38338 8

828 885 8288.8 2: $8838 25S 88% \o :StoQEm

358m

 

8882

an 89:

.88 8.8.8 888 888 .888 ”A888 888
8 8838 83 38 888 2: 8 888 8 882 «8.53 833
3 838 25 3883.8 £80 883 BERN \o aﬁsuobtb

8an

 

b;

aﬁoz

on 83H

88: 8:83 88 8 838886 .8 3983 .88on 888
88.8 88: 838.3 8888 a 3 80.88% 8 888388
833 3 8.8% 85 38838 8888233 8838Q

amlumm

 

b;

8882

an 89:

$88888 8.88 8888 .883 88888888 .885
8 8888 88.8% 8: 833 a 885 £88 $883 8.8%
833 3 8.886 2a. 388.38 b.8835 828% 88.83.

88883

 

64

35.2

S as:

.8836 888 888 .883 8.888.888

.885 8 8:88 88 823 a 88:3 888 3883 88m 833
8 8.83. 23 388.38 3.88sz 82.5 835$ 8883‘

838.83

 

mmqam
88E

 

 

c8889“
8 8882

 

888m

 

838.889

 

882
8882

 

8883885 883. 3 88889 8.8882 A 8389

 

 

57

 

.383 $5.838.“ .888 2: 3 88.23
8038.3 8.83 85 8:: 888 2% .8585: on”. .883

 

 

 

 

 

 

 

8 89: 89a 2: £83 888 8» 8 82823 «883 8.8 88: 838.3
3 A 8882 88—4880 88.88% >88 Boa 38838 88.5 8389510 88832 885
.gnosumd 888 .88 @9888 888 .8me
$8888 88 883883 88883 833 :83 88:wa
bé @8825“ B 88.: 85 $8838 mmwmﬁoeéomR coﬁoznogm E 8830 8880
8mg 88.82% 282
883 8 8882 88am .833ng 8882

 

8.888 A 383.

 

58
measure Product is included because the more product lines a plant makes,

the more likely it is that not all products will consume resources in the same
proportions. The measure Lines quantiﬁes the number of product lines
produced in the plant and is included because the more product lines a plant
makes, the more likely it is that not all products will consume resources in
the same proportions. Similarly, Routings is included because each routing
speciﬁes the production resources, sequence of use, and duration of use that
are needed to make the product; greater diversity of routings may result in
disproportionate use of support resources. The measure Setups is included
because lengthy setups increase the overhead cost, which is then spread over
the units made using the setups; depending on the size of the run, the
consumption of resources may be uncorrelated with the allocation base. The
measure Labor indicates whether the number of labor rates is small or large;
many rates indicate that products are consuming diﬁerent resources, which
may be occurring in disproportionate amounts between products. Dedicated
equipment (measure Ded_Eq) reduces the need for setups and simpliﬁes
associating cost with the consuming product, whereas a job shop (measure
Job_Shop) uses the same equipment for many products, thus increasing the
likelihood of having resource consumption be uncorrelated with the
allocation base. The remaining measures (Innovate, Compress, and Change)
show the rate of change in the manufacturing environment; change

complicates the ability to capture costs and intensiﬁes the eﬁ'ects of such

59
measures as Product, Routings, and Setups. Product, Routings, Setups, and

Labor turn out to be of particular in interest in subsequent analysis.

As mentioned previously, Lines is a calculated value and was obtained
as follows. The survey responses were ordered based on the number of
product lines, were ranked in ascending order, and were divided into ten
groups. Those in the ﬁrst group (fewest product lines) were assigned “1”;
those in the next group were assigned “2”; and so on until those in the last
group (most product lines) were assigned “10.”

Table 2 shows the 1 1 measures used to record competition information.
As discussed in Chapter H, four measures (Typ_Shar, Big_Shar, Ind_Comp,
and Firm_HHl) were market-share-based measures, two measures (Plan_Cyc
and Life_Cyc) related product and plant life-cycle stages, and the remaining
ﬁve measures ( Per_Comp, Agress, Lower, Foreign, and Domestic) were
respondents’ perceptions of competitive pressures in their industries. The
calculations for the market-share based measures and the life-cycle measures
are discussed below.

'l‘yp_Shar and Big_Shar related to the typical and biggest market
shares of a plant’s products; both measures were calculated following a
common method. Each measure was recorded by the respondent as a
percentage and entered as a two-decimal-place fraction of 1. Each was
multiplied by ten (to yield a number between 0 and 10), then subtracted from
10 to reﬂect the suggestion that higher market share might be associated

with less competition.

 

.835888 8.88 888 8ng 383 85
8 388 8838.8 .8.“ 8mm 88. .3 38: 988 emu—88 8.888

 

 

 

 

 

 

 

 

2-.. 89.28 8 8: 8:8 2: a. 8888 a a 23m :82 889: 3385 3845.
3.. 88 on 5m 883388 8.88 888 8:3; 8888 0888.5
.8” m8o: 88m m88> 8 $0388 @283 23 883 88888 mﬁ 88:?
5A 88.82 888080 .8 8388388 .88: a $8888 88% 380.83 :8: 80..qu
8388988 988 888 8&3
88388“ 838.8% 58 8888888 88888 maﬁa.“
KL 8m88m an 88: 8.85 2: 38888 88.38800 c.2888 .3 85§30 8.8889
88.8888 988 888 8&3
”893838 858.8% :8.» 883888 888m 988
b; @8825 an 88.: 8.8.6 25 38888 88.38880 888k .8 88330 888m
8388988 988 888
883 888.88 838388 8 8.8.88 598 8 88.8
.8..-“ @8885“ on 89: 98888 8.8m 85 38888 5982 EEK .8 83¢ 833 833
4888988
8.88 888 883 ”883.8 2: 8 883888 .3 833.88
88838888 .8 8%.: 25 .8 83888 8.88888
8A 88.82 am 803 85 388.88 3888in 838580 .8 88a 888ng 25. 883‘
8888988 8.88 888 8&3 9.5868 one
8 «8880 833888 .8 823 23 .8 839888 988888
b; 888 Z am 89: on: 388888 8388880 \0 88a 88.88ng SE. n8oOI8m
om8m c888m 282
88m 8 888 Z 888w 8.58880 8882

 

 

 

 

 

8.88985 8882 3 88889 88882 .N 838,—.

 

 

61

 

 

 

 

 

 

88=n8oo 8888888 8.88 888 8888 “8
88m 8 8 888 888.88 883 .8 888... a 8 8.8 8» 8 8888
v-8 88.8 Z 888080 888888 .8 ~88 88 38888 88% 380-88 23. 96.8.84
888888 888“
<mD ”.82 88 88888 8:888 888» 8 88 8 838—88 Saw 8888 88
8-8 89.28 8888 a. 883 p.88 2: a Em 8.338 3: 8 =88 3...: 2:. H88
«Vanda 898 8888988 8.88 .888 888 WEE 8.8888
b; 8888 888o80 8» .8 8888 a 8 888 8 88% 8.888800 89885 89 888088
4888888 8.88 888 888 388 88 8
888 8888 a 8w 8.8 88 .3 38 8.88 88.88 88888.
3 -c 888>8m Am 88.: 8888 8» .8 88888." a 8 83m 888: 888ch 8883 8.81mi
88m 8888 . 88 Z
88% 8 8882 88m 8388an 8882

 

 

 

 

 

@289 .N 886

 

62
Industry Competition Level (Ind_Comp) was based on the market HHI

for that 4-digit SIC code, which can range from 10,000 (indicating no
competition) to near zero (indicating intense competition in an atomistic
industry); Ind_Comp was calculated from the Manufacturing USA [Darnay
(1994)] data. Ind_Comp was the same for each ﬁrm in the 4-digit SIC code
and reﬂects a comparison across industries. The HHIs for the various SIC
codes were numerically ranked such that the values representing greatest
competitive pressure (the smallest HHIs) were at the top, then the ranking
was divided into seven groups, with the ﬁrms in the top group (most
competitive industries) being assigned "7" and the ﬁrms in the bottom group
(least competitive industries) being assigned ”1.” The categories were used to
avoid having outliers; some ﬁrms had very large HHI values, whereas many
other ﬁrms had near-zero HHI values.

In calculating Firm Portion of the Industry HHI (Firm_HHI), the
ﬁrm's market share (expressed as a percentage) was squared, then divided
by its industry HHI and multiplied by 10. The product was subtracted from
10 so that larger values represent greater competitive pressure. Values
range from ”0" to ”10.” Here again, the artiﬁcial categories were used to
permit distinctions to be made between plants with low portions of their
industry’s HHI.

Life-Cycle Stage (measure Life_Cyc) was based on the levels of

pressure generally theorized as existing within each life-cycle stage. The

63
approach used here generally followed that of Anthony and Ramesh (1992)

which used sales growth (SG), dividends as a percentage of income (DP), and
Age to divide ﬁrms into ﬁve life-cycle stages (LCS)3. Four exceptions to the
approach taken by Anthony and Ramesh were as follows: First, SG was
based on restated sales in case acquisitions, discontinued operations, or
accounting changes had occurred during the six years. Second, calculations
were performed only once for each ﬁrm since the present study's focus was on
the pressures currently felt by the ﬁrm as it chooses which FCA methods to
use. Third, new ﬁrms with at least 1992 data available were classed as
introductory ﬁrms. Fourth, the interpretation placed on the resulting groups
was slightly diﬁ'erent, with the ﬁrst two groups treated as
introductory/growth stages and the last three as maturity/decline stages, as
explained in the next paragraph.

When splitting the ﬁrms into the ﬁve groups for life-cycle stages, two

groups were assigned to the maturity stage since previous research had

 

3 Anthony and Ramesh began by calculating annual values for 86, DP, and AGE, as
follows:

DPt = (DIVt/IBEDt) x 100,

SGt = ((SALESt - SALESt-1)/SALESI_1) x 100,

AGE = Current Year - BYEAR.

After repeating this process for each of the last ﬁve years, the means of SG (MSG)

and DP (MDP) were calculated. Then all ﬁrms were ranked on each variable individually,
divided into three groups, and each ﬁrm was assigned a score as follows:

DP Score: 1 = Low, 2 = Medium, 3 = High,
86 Score: 1 = High, 2 = Medium, 3 = Low,
AGE Score: 1 = Young, 2 = Adult, 3 = Old.

Subsequently, the three scores for each ﬁrm were combined in TOTAL, then
sorted in ascending order and divided into ﬁve life-cycle groups. The ﬁrms in each group
were then assigned a value (Life_Cyc) for the life-cycle stage in which they fell.

64

found the majority of ﬁrms falling in this stage. Then, since the purpose of
identifying life cycles was to associate the competitive pressure theorized for
each stage with ﬁrms in that stage and their choice of FCA methods, values
for Life_Cyc were assigned as follows: " l" for the introductory ﬁrms (group
1), ”3" for the growth ﬁrms (group 2), ”7" for the maturity ﬁrms (groups 3 and
4), and ”5" for the declining ﬁrms (group 5). These assignments were in
accordance with the pattern of competition over the life cycle, as described by
Porter (1980) and other researchers.

Plant Life-Cycle Stage (measure Plan_Cyc) represents the plant's
overall product life-cycle stage as perceived by the respondent. It was
calculated as a weighted average by using the percentages of products
reported for each life-cycle stage by the respondent weighted by values for
the life-cycle stages. The ﬁnal weighted values range from "1" to ”7," with
products receiving the following weights for the various stages: ” 1"
(introductory stage), "3" (growth stage), ”7" (maturity stage), and "5" (decline
stage).

Table 3 shows the ten measures used to record plant strategy
information. As discussed in Chapter II, six measures (Prod_1nn, Proc_Inn,
Differen, Hi_Prof, Commod, and Low_Cost) were used to record respondents’
perceptions about the plant’s strategies to differentiate itself and/or become a
low-cost leader. The plant’s ability to differentiate itself from its competitors

by having unique products (Diﬂ’eren), by innovating new products

 

5A

coﬂgom

a. at:

.3395» 8358988 988 a 888 8:3: ”883.8
2: 8 .8682 88-33 on» 8 839 05 8.83 3 838 85
.8 :eﬂaoeuea 983.888 85 388888 8863 88.0-33

880.584

 

5A

89885

3. a8

.mgab» 3338988 988

a 888 8.33 ”3838.8 388888 8888888888:
88388 283 on... 8.83 3 838 mg... .8 839.8808
983.8988 23 38888 8.888% 3388880

88880

 

5A

88.82

3. 80.:

Engage»

8388888 988 a 888 8:2: Home» 5988 2.88
AME » 88 8 883 23 833 3 838 one .8 8888.89
@8888»? 83 388.82 83% 2.8.83 ”58¢ dmmm

mania

 

5A

88.82

a. as:

.5385» 833388 988 a 888 8% ”8338888 .8
82% 3 8.8988 8:? 888: 8 898388.36 >33: 88
98 $2688 @838 05 833 3 838 2: .8 83.80.88
@8889»? on» 38888 asggeexeta “882%

6.8.88

 

5A

aﬁoz

9.. 89:

.595» e>ﬂﬂoq88 988 2888 non—was
88.8888 889:— a 8: 8 839 83 £033 3 .838 2:
.8 ueﬂneouon 32888888 25 38888 88885 889$

Salsa

 

5A

88.82

3. so:

.3395» 335388 988 2808 quWE
838888 8388 a 88 8 .883 05 A833 3 83m» 2: .8
8388qu 982.88»? 23 $888.88 88825 835$

Gnu—108nm

 

 

mam
88h

 

©$Ho>om
8 88.82

 

388m.

 

838.889

 

08» 2
98982

 

unseen—m »»»»»< 3 @8389 8.8802 .m 0389

 

 

 

.me88» 8238988 988
a 888 88.8 ”888» 88.88 9588.83 8.8889
8 8889 839 8: 833 £93 38938 on”. .8 839889

 

 

 

 

 

v-9 888 Z .8 839 93888988 85 398898 888m. 828830 8 88k 888m
.8398...» 88.8988
988 a 888 883 ”888988 88 88.89 889
85 83? £8, 38938 05 .8 839889 983.8988
9.9 889,39 8 839 85 388.88 833* 88.88% 880 8 b.8835 8835
@395» 83330988 988 a 888
88.3 88.3 25 388 3 388 89.8» 3 88 85 88
838.8 39 .89 88 883 m »8» 839 we. 83? 3 838
b; 8888 Av 839 85 .8 839889 ».83:89»8 88 388898 8880 8835 889.
.395» 888988 988 a 888 8AMEJ.8888 38
8 88.8 838.8 39 88: 839 one 833 3 838 one
5.9 689,8 8 83m .8 839889 983.8988 8... 388.88 88¢ 885.880 88880
8mm 9.8823“ 882
88.9 8 88.82 888m 83988Q 8882

 

 

 

 

 

8.289 .m 8399.

 

67
(Prod_Inn), or by innovating production processes (Proc_Inn) turned out to

be of particular interest in subsequent analysis. One measure (Cost_Bas)
records the degree to which the plant prices its products based on cost. Two
measures (Target and Intense) show the plant’s focus on cost-reduction
eﬁ'orts, either through direct cost-reduction campaigns or through using
target pricing to set a price and reduce costs to allow the desired proﬁts. The
last measure (Service) was added at the suggestion of a survey reviewer
because providing high-quality service is an important strategy for many
ﬁrms; the Service measure could also help the plant to differentiate itself.

Table 4 shows the eight measures used to record managerial support
for FCA method implementation. As discussed in Chapter 11, three measures
(CEO, Control, and Top) record respondent perceptions of upper management
support, three measures (Middle, Account, and Market) record the user
support, and two measures (Level and Effect) record the organizational level
and perceived effectiveness of the FCA method champion. While all of these
measures group together well, Control, Top, and Level are of particular
interest in subsequent analysis.

Table 5 shows the l 1 measures used to record implementation
obstacles which would impede FCA method implementation. As was
discussed in Chapter II, ﬁve measures (Resist, Sponsor, Sht_Term,
Quant_CB, and Commun) record nonﬁnancial obstacles to implementation,

whereas the remaining six measures (C_Plan, C_Consul, C_IS_Chg, C_Train,

68

 

v-9

88.82

cm a»:

5.8998 988 282: Honda

”Sum 25 93...? 82988 <09 w888399: 3.8998
8893 mﬁnzoooa 05 938.5» Be: .8 8390889
3.868989 oﬁ 388.992 toQQam. 8.32889.‘ 8 ~2me

9989i

 

5.9

88.82

cm 98.:

6.8998 988 @988 8&3 mg

2: 93:3 8388 <09 ”88989989 «.8998 98888
8:88 05 938.5» Boa .8 8.59889 889.8989

05 398398 989.93% Essences: £38.39 .8 8&3

3252

 

6.9

8882

om 8:

9.8998 0.88 m8mﬁ 89$: mg

on» 93$? muoﬁoﬁ <09 m888£9§ «.8998 98888
93 .898 2: 938.98 3o: .8 8598.89 589.8989

25 388.992 “885m. 9828938: Qoﬁ 8 m2on

89.

 

352

am 98.:

#8998 8.88

388 88.3 ”and mg... 9398 82988 <09 m888£9§
8.8998 8:95.98 23 958.98 Bo: .8 8398.89
£98883 29» 388.993 toQQSW 822280 8 8:93

880

 

5.9

88.82

am so:

#8998 0.88 2808 momma “Sum 05 88 «8:88 <09
m888£9§ $8998 GHQ 06 938.5» Bo: .8 89998.89

389.8989 23 398298 838% CEO 8 8.293

Omo

 

“mam
83,9

 

 

83339
8 88.82

 

8.98m

 

8399.883

 

082
38.on

 

f899am 89.8882 885‘ 3 8885 8.5882 .9. 389.

 

 

69

 

98998 8.88 8888 .8983 883838 88 88a .89 88:
68 8888888 93 .8 98998 8.5 88838 8 88:
8: 88:88 8:... 8388mm >8: .8 839889 8.888988

 

 

 

 

v.9 8882 S 889 85 398898 8888th 8.239830 8 8:3 850
98998 8.88 8888 .883 885838
88 89a .89 8888 88 .3 689 98858 .8 889 8: .8
v-9 8882 m 889 8398.89 8.888988 85 $88.88 289830 8 ~83 884
8.8998 8.88
8.88 .8ng ”88a 8:... 9.38, 88288 <09 88883989
38998 88859 u88488 88 988.58 Be: .8 8398.89
5.9 8882 .«w 8889 8.888988 88 888898 82330 8.9883: 8 8ngQ 8832
can 8888M 88a 2
88.9 8 8882 888m 8398880 8882

 

 

 

 

 

$5800 .9. £89.

 

70

 

8882

«Na 98.:

.3888 98988 a 89888 88918389883983

89: 8.93.3: 9389883983 8388 <09 .89 ”39939
88988 333 88888 888 85 833 3 838 3:

.8 93998.89 8.8388989 23 388.998 M98939 8 890

939: 0

 

88.82

.8 839

.3888 98988 a 8988

.883 ”9389883983 8388 <09 8.9383 M98939

.8 8398.9 33 9383988 3: 9888.95 u38399888
.8 8.58 88 833 3 9938 on... .8 93998.89 8.8398989
8%. 3988.998 8.998: 8 83me 9383938890

998880

 

>3

35.32

ca 839

.3888 .8888 a 8988 .8383 ”9388883983

83.9.3 8388 <09 8 3 M98988 .8 83885 898 888
u3989a9a .8 9.898 88 833 3 9938 88 .8 93998.89
8.888988 88 888.998 888% 88 8890 3398:9930

m0|a890

 

9.3

3982

8 89

.3888 8m988 a 8988 .883 ”9393883883
8388 <09 8.93.93 8.9888 88899 889.988

8 883 a 833 3 9938 85 .8 93998.89 8.9939898“
85 388.998 8.988: 2898.99 8.89-230 8 88.9

8.89.13m

 

Razz

33 839

.3888 9mm898 a 8988 .883 3939
85 3 88898 89 883 38:88 <09 833 3 9938 3:
.8 839889 893.8989 35 388.998 .989ch 8 83

88890

 

88.82

«a so:

.3888 888.98 a 8988 83MB ”3.2988 <09
3 u989a8 $38.9 9888888 98988 Bo: .8 93998.89
8.8398989 3: 388.9989 828.98% ~38mc99§~

888

 

 

 

88.882
8 88.8 2

 

8.98m

 

9393.880

 

88 2
8.9882

 

8388.5 9388883989 m88< 3 89389 8.8832 .n 389.

 

 

71

 

b;

88.82

5 38

.2888 nonnohn a £82: 8&3 ”8338838:
8:88 «Gm «86:3 3.8 amass 988 wﬁmmcooa

98 ”38:8” 5MB ugaﬁomma $80 05 853 3 “:38
2: .8 838089 £86898." 2% 38882 635 \o 800

5.5.0

 

5A

8882

..2 am:

.323on wwwaobm a 2808 gunmE ”83388038 8:88
<Om 986:3 $83 .8 £3.89 a .88 @8393 88 856
9383 £33 cgaﬁomma $80 25 A033 3 83.8 2: .8
833889 $868.83 2G $8883 889$. :35 \o .80

ESQIO

 

8.52

m2 80.:

.323ch Homage» a

£888 883 ”83388395 8:88 «mom 983:: 98¢:
as 93:95 :33 88838 $80 2: 833 3 «:38 05
.8 839889 5:38.83 23 $8883 MESS \o 800

 

88.82

:2 :83

.293on amuse“? a £808 uaﬁﬂacuﬁnmaoaﬁw

8.: 989:: 8888 «RE 8 .8 85388295 25
mag 83?? noun—5&5 on“. 9:988 :33 38888
380 on”. 833 3 “:38 c5 .8 838089 3:38an
2: 35883 93:35 «8393 eozuztoxs \o 880

«.6..de

 

PA

88.82

uﬁ so:

.088ch acmqobm a 888 8:3: ”833.883qu 85
2283 nouﬁaoﬁoiaﬁ 8:88 «Gm 828 38:38“.
E mﬁmﬁﬁ £85 88888 $80 2: 833 3 £888 0% .8
8398.89 £868.88 25 3888...: 3293.390 8 800

ﬁgmﬂcolo

 

 

madam
8.5m

 

838%
.8 88.82

 

8.58m

 

852.889

 

@852
3882

 

8.....89 .m 039—.

 

‘72
C_Dual, and C_Data) recorded ﬁnancial obstacles. While all of these

measures grouped together well, Sponsor, Quant_CB, C_Plan, C_IS_Chg,

and C_Data were of particular interest in subsequent analysis.

CHAPTER V

STATISTICAL TESTS AND THEIR RESULTS

DESCRIPTIVE ANALYSIS

In the descriptive analysis, two interesting demographic variables are
discussed, then univariate descriptive statistics (number of observations,
minimum, maximum, range, median, mean, variance, standard deviation,
standard error, skewness, and kurtosis) are presented for each of the 50
measures, as well as for the dependent variable’s measure, OVERALL.
These statistics provided information which inﬂuenced the choice of factor

analysis rotation to be used in subsequent analysis.

Demographic Data

While the demographic data were not used for testing the hypotheses,
they were used to look for biases in survey responses. A number of
demographic characteristics were collected, including information on the
respondent’s position, area of expertise, background, educational level, years

in the company, and years in the industry, as well as the mailing answered,

73

74
the degree of choice the plant has in choosing its allocation method, and the

ﬁrm’s 4-digit SIC code. Each characteristic was analyzed by performing a
MANOVA, using OVERALL as the dependent and the characteristic as the
class variable. Only two characteristics — the respondent’s years with the
company and the ﬁrm’s 4-Digit SIC code —— showed any signiﬁcant
relationship with the dependent variable.

As shown in Table 6, a signiﬁcant relationship (at the 99% level) eldsts
between the respondent’s years with the comp any and the level of FCA
method implementation. The longer the respondent had been with the
company, the higher the level of implementation reported. While the
relationship is interesting, it scarcely seems to offer a cause-and-eﬂ'ect
relationship. The number of years a respondent has been with the company
is not likely to affect the FCA method implemented. The fact the relationship
exists does not offer any reason why the survey should be biased.

The ﬁrm’s 4—digit SIC code also showed a signiﬁcant relationship (at
the 95% level) with OVERALL. In all, 34 4-digit SIC codes were represented
in the survey; Table 7 shows the industry breakdown of responding plants by
2-digit SIC codes, the MANOVA results on the 4-digit SIC codes, and a few of
the 4-digit SIC code industries with the highest or lowest means for
OVERALL. Given 1) that ﬁrms were selected to represent industries with
diﬁ‘erent market structures and levels of competition and 2) that different

industries vary in their dynamics, industry bias is not an unexpected ﬁnding.

75

 

Table 6. Respondents’ Years in Company and Its Relation to OVERALL

 

 

 

Years Experience Number Group Mean
in Company of Respondents for OVERALL
1-3 22 4.1
4-6 18 6.8
7 -9 13 4.5
10-12 24 5.8
13- 15 l l 6. 7
16-20 16 6.3
21-30 14 6.5
over 30 4 8.0
Unreported __2_ --
Total 124 5.8
Average Years 1 1.6
MANOVA Results:
Multiple R2: 0.154
F-Ratio: 2.975

Two-Tailed Probability: 0.007

t

In fact, industry bias supports the rationale behind the study and should not

adversely aﬁ'ect the study.

Univariate Descriptive Analysis

In order to survey responses, univariate descriptive analysis was

performed on each of the measures; the results are shown in Table 8.

76

 

Table 7. Industries Represented in Survey and Their Relation to OVERALL

Breakdown of Responding Plants by 2-Digit SIC Code Industry:

 

2-Digit SIC Number of
Code Group Indust_ry Description Respondents
33 Primary Metals 19
34 Metal Fabrication 1 l
35 Machinery (Except Electrical) 29
36 Electrical & Electronic Machinery 19
37 Transportation Equipment 14
38 Measuring Instruments (Including Medical) 2 1
39 Miscellaneous Manufacturing _1
Total Plants Responding 124

MANOVA Results for 4-Digit SIC Codes:

Multiple R2: 0.368
F-Ratio: 1.589
Two-Tailed Probability: 0.045

4-Digit SIC Code Industries With Highest Or Lowest Means for OVERALL:

4-Digit SIC Number of Group Mean
Code Group Indu§t_ry Description Respondents for OVERALL
3523 Farm Machinery 3 10.3
3546 Power Driven Tools 1 9.0
3651 Household Audio/Video Eq. 1 9.0
371 1 Motor Vehicles & Car Bodies 3 7.5
3452 Bolts, Nuts, Rivets, Washers 3 7.3
341 1 Metal Cans 4 3.5
3561 Pumps & Pumping Equip. 7 3.3
3634 Electric Housewares & Fans 1 2.0
37 16 Motor Homes 2 1.5
3572 Computer Storage Devices 1 1.0
375 1 Motorcycles, Bicycles, Parts 1 1.0
Total Plants Responding 124 5.8

77
Table 8 shows how many responses (N) there were for each measure.

For example, for the measure Eﬂ'ect (effectiveness of the champion), only 94
of the possible 124 respondents provided the information, so approximately
25% of the responses are missing. Missing responses indicate (among other
possibilities) either that respondents do not like disclosing this information,
that they do not know the information, or that they do not understand the
question.

Table 8 also reveals the minimum and maximum values of the
measure (MIN and MAX), the ranges of responses, and the medians and
means of the measures. The table further reveals important information
about the variation experienced in the measures and shows this variation in
the columns for variance (VAR), standard deviation (SD), standard error
(S.E.), skewness, and kurtosis.

The measurement for skewness (SKEW) is adjacent to the z-value (Z)
for skewness, and the measurement for kurtosis (KURT) is adjacent to the
z-value (Z) for kurtosis. In both instances, the z-value indicates whether or
not the measurement is signiﬁcantly different from normal. Only seven of
the 50 measures are reasonably normal; the others exhibit too much
skewness and/or kurtosis to be considered normal. As explained in Johnson
and Wichern (1992), the lack of normality aﬁ‘ects the usefulness of
correlation analysis; the magnitudes of Pearson correlation coeﬁcients are
still valid, but the signiﬁcance of the correlations become suspect. The lack

of normality, therefore, has major implications as far as choosing a method

8
7

.000? Eocos> ucu camccoa m_m>_u:u 05 Suu__u>:_ .o: 300 2:055: 00 x00. 0.30:0 362 25:09:00 .3853
.o 230: .252 05 2 2.0 5:260: won—4.020 $5 .92 a 3:07. 03:3. :0 30:39.0 80:: gag-N 200 6qu

 

 

 

 

000.0. 00:.- 000- {0.0- 000.0 000.0 000.0 000.0 0.0 0.0 0.0? 0.? 0: 3::
00.20- 0.00- 000 000.0 03.0 000.? 000.0 000.0 0.v 0.0 0.0 0.? v0_. 00:20
000.0- E:- 300 000.0 0070 08... V000 0004.. 0+ 0.0 0.0 0.? V0_. 32an0
30.0. 00.?- 000.0 000.0 0.0 000.? 000.0 000.0 0.v 0.0 0.0 0.? V0? 2055.
000.0- 00?? 000.0 03.0 00.0 000.0 00.0 30.0 0.0 0.0 0.0 0.? v0? aocwlnow
2.0.0- 000.? 000.0. 000.0- 003 000.? {0.0 0006 0.v 0.0 0.0 0... v9 3300
000.0. 000.7 03.0. K00- 0000 000.0 000.0 00:. 0.0 0.0 0.0 0... 00v .003
000.0- 50.7 000.7 000.0- 050 000... 000.0 3.3. 0.0 0.0 0.0 0._. V0_. awlnmo
009.0. 000...- 000.0 000.0 05.0 00;. 03.0 000.0 0.v 0.0 0.0 0.? V0_. 00:03”.
000.0- 00.7 000.0- 09.0- 000.0 000... 00v.0 000.». 0.0 0.0 0.0 0.? 00_. 8:02.”.
00.50522 508.955,
N .EDV. N >>mx0 .00 0.0 m<> z<m=2 2300.2 0025. x<s_ 2:2 2 _ $505.2
202:3. 08:39.0 coast; 223.030 2050 _

 

 

 

 

 

4386 can 8582 on 8 80380 89588 .0 sea.

 

 

.500: E0505 0:0 53:02 2020:: 05 0:00:95 :0: 0000 3.3-:0: :0 x00. .mﬁbacm .200.— mEoSQES 390:3
__0 230: 3:00.. 05 2 0:0 3:058: 303.02: .05 .05. a 2::03 030:3. :0 30:39.0 80:: mo:_0>-~ 0.00 ”202

 

 

 

 

 

3.0.: :00.0 ﬁlm?- 00: T 00:.0 v00: 000.0 mmvh 0.0 0.0 0.0 0.: :0: o>0l0:_-_
2.0.0 .00.: 00‘». 000. T 000.0 000.0 000.0 000.0 0.0 0.0 0.0: 0.0 00: _IIIE.._u_
W 30.0. #:0. T vnnd- :0:.o- 00:.0 :00: 000.0 000% o.v 0.0 0.x- o.: 00: QEoO-l-UE
0.3.0- 9.0. T 000.0- 000.0- 000.0 00:.0 000.0 00000 0.0 0.0 0.0 0.0 mo: 5:5 05
v: : .0 000.0 03”.? KNO- w :0.0 000.0 50.0 000.0 :N 0.0 0.0 0.0 0:: :mcleE.
0.0. T :000- 000.0- 00 : .0. m0 : .o 000.: 000.: v00.v 0.? 0.0 0.0 v: 00: o>0lcm_n_
00: 0. 000.0- 000.0- 000.0- 00 : .o 000.: 000.0 000.0 0.0 0.0 0.0. 0.: :0: 0:09:00
0:0.0- 00:. T 00:.T $0.0- 00:.o 000.: 000.0 00:.v o.v 0.0 0.0 0.: v0: c990“.
000. T 0:50. 0010- 000.0- mvto 05.: 0:00 000.». 0.0 0.0 0.0 0.: V0: :95.—
0:0.0 000.0 he”? :0. T 000.0 00:: 0:0: 00:.0 0.0 0.0 0.0 0.0 V0: 000:9».
03.: 00:0 000....- 000. T 000.0 000.: 000.: 0:0.0 0.0 0.0 0.0 0.0 V0: 0800!:00
0050005. :25an00
N 0.me N >>w¥w m.w .D.w m<> Z<m_>_ Z<_Dm_>_ m02<m X52 2:2 2 _ mmamﬁi
300:3. «00:39.0 cozatu> 220:0on £000

 

 

 

 

 

8.309 .0 03.0.0

.000... £05.? 0:0 :00::02 0.03.0:0 05 0:00:95 :0: 0000 3:0.E0: :0 :00. 0.0205 .200: 95:00:50 00.05::
:0 0:30: .030: 0:: 0: 0:0 3:0:5: 0000.090 :05 .90. 0 3E0... 0.00:3. :0 30:39.0 00:: 00:_0>.~ 0.00 “0:02

 

000.: 000.0 30.0 .V::.: V::.o 000.: 000.: :0:.0 0.0 0.0 0.0 0.: 00: 00.300
w v00: 0:0.0 0: :0. 00:.T :0:.o :0v.: 3:.0 000.0 0.0 0.0 0.0 0.: 00: 00:95
000.0- .000- 0v0. T 00v.0- 0V: .0 000.: 000.0 :00 0.0 0.0 0.:. 0.: 00: .090:

0:0.0- 0v0.T 00:0 000.0 00:0 :05: 00:0 {0.0 0.0 0.0 o...- 0.: v0: 000M000
000.0- :0:.:- 000.0 0:0.0 0.1.0 0:0: 000.0 {0.0 0.0 0.0 0.0 0.: V0: .000 30..

 

000. T 000.0- 0:0.0 :000 :0:.0 000.: 000.0 000.0 0.0 0.0 0.0 0.: 00: 002.050
:00.:- 000.0- 03.0- 000.? :0:.0 000.: 000.0 03.0 o.v 0.0 0.0 0.: 00: :0..n.I_I
000. T 000.0- 000.0 mmvd 03.0 000.: 0:00 :000 0.0 0.0 0.0 0.: v0: :0..0::.0
000. T 000.0. 000.0 00v.o 00:0 000.: 000.0 000.0 0.0 0.0 0.0 0.: v0 : cc_loo._n.
000.0. .00.:- :V:.: :000 :0:.o 0.00.: 000.0 000.0 0.0 0.0 0.0 0.: v0: cc_luo:n.

 

0050005. 32052

N kmDv. N >>mv.0 .m.0 0.0 m<> Z<w_>_ Z<_Dm_>_ m02<m x32 2:2 2 _mmzwﬁs
0.00:3. 30:39.0 :0.:0.:0> 0.80.5000 0.00m _

 

 

 

 

 

 

 

6.00000 .0 030.0

.500: E0023 0:0 0005.03 0.00.05 05 0:00:03. :0: 0000 2:00:00 :0 :00. 00205 8:00: 02000088 .000000

:0 05:0: :0000: 0:: 0: 000 5:00:00 0000.020 :05 .0>0. 0 3.00.0 0.0003. :0 «00:39.0 00:: 00:_0>-N 0.00 ”0:02

 

 

 

 

 

W000- Nmmd- 00:.0- 000.0- 00:.0 000.: 000.0 000.0 0.0 0.0 0.0 0.: 00 80:0
000:- 000.0- 03.0- 0000. 00:0 0:00 000.0 ::0.v 0.0 0.0 0.0 0.: :N: .05..
000.0 000.0 «00.0 000.0 «:0 00:: :00: 000.0 0.0 0.0 0.0 0.: 00: .9005.
.000: 000.0 000.0- 08.:- 00:.0 000: 000.0 000.0 0.0 0.0 0.0 0.: :N: :0:08<
«00.: 000.0 000:- 09.0- N::.0 com: 03.: 000.:‘ 0.0 0.0 0.0 0.: 0:: 0.00:2
0::0- :000- 000:- 0000- 00:0 ~00: 000.: «00:. 0.0 0.0 0.0 0.: 0:: 00...
00:0 000.0 000.0- 000:- 00:0 3.0.: 000.0 0:0.0 0.0 0.0 0.0 0.: «N: .2800
«00.0- 0000. 000:- :000. «0:0 000.: ::0.N 000.0 0.0 0.0 0.0 0.: 00: 000
005000: 9.00000 0:000:02
N 0.09. N 39.0 0.0 0.0 m<> z<m=2 2500.2 0025. x52 2:... z _ 00:05:
0.00:3. 30:39.0 :0.:0...0> 0.20.3000 0.00m _

 

 

 

 

 

 

00.00000 .0 0309

.58: E050; 0cm came—.0... 0.05.05 05 202.02... no: 300 53.2.20: .0 0.00. airman .208 35:09:00 30.00....
.0 220: «032 05 2 030 3.08.0: 303.020 .05 .90. a 3.5.0 0.00:3. .0 30:30.0 .ouc: «029$ Rom ”202

 

 

 

 

 

 

05v. _.- 0v0.0. mm _. . .3 3N0- ~0VNO 35d N005 _005.0 0.0 0.0? 0. 3 0. r vmw 44<¢m>0
VOYT 0N0.0. F50.N- 00.10. _00...0 005... 50d 0004. 0.0 0.0 0.5 0.? 0m? 200.10
000.N- 500. T 50v.0. 00rd. 050 N00... 50v.0 Nmré 0.v 0.0 0.5 0... 0N_. _NJOIO
03d- 000.0. 500.0 00 _. .0 00 r .0 000. F 0VN.0 000.0 0.v 0.0 0.5 0. F 0Nr £90.10
000. ..- .000. 0N5“. 05v.o. 00rd 0.06 000.0 505% 0.0 0.0 0.5 0.? 0N.‘ 0:0I0_Io
0N0.N. 00m. T 0Nv.0- 000.0- 00rd 050.N 00N.v 007v 0.0. 0.0 0.5 0.? 0.... _amcoolo
:0. T va.0. 0mm... V5N.0 325.0 0v0é 000.N 53.0 0.0 0.0 0.5 0... ON? cmElo
09. T 0000. 0.50 00.20 0320 000... .‘NmN 000.0 0.v 0.0 0.5 0... 0.... czEEoo
050.N- 0N0.0- 0w 5.0. 0N0.0. 0070 0056 0.0.0 Nv0.0 0.v 0.0 0.5 0.? 0N_. mOJcmac
.000- 000. T 00N.0. 90.0. 505.0 :09 00N.0 000.0 0.v 0.0 0.5 0. w 5 I. E..o...le
0000- 03. T 000... 050.0 0570 000... N0v.0 00N.0 0.0 0.0 0.5 0.? ON? 82000
95. ..- N050. 000.? 03.0 505.0 vw5é F50N 000.0 0.0 0.0 0.5 0... our Emmy.
02:03:. 3.0830 535580.05.

N Hmnv. N >>m¥0 .w0 .00 m<> z<m_2 Z<_0m.2 m02<m XS). 2:2 2 _ mmamﬁﬁ

0.00:3. 30:39.0 :o_uu_..u> 229.630 0.30 J

 

 

 

 

 

6.3000 .0 0309

 

 

83
of factor analysis. Maximum likelihood factor analysis assumes normality of

the measures and will not work with nonnormal data. Fortunately, the
desired method, principal component factor analysis, is robust enough to
work with nonnormal data [Johnson and Wichern (1992)].

Clearly, the measures Per_Comp and Agress are the worst skewed and
have the greatest kurtosis. Further examination reveals that nearly all
respondents scored both these measures with very high values (the
maximum value possible was a “7”). Of the 124 responses for Per_Comp, only
eight were less than “5,” and 99 were “6” or “7.” Similarly, of the 124
responses for Agrees, only 15 were less than “5,” and 104 were “6” or “7.”
With so little variation and such extreme skewness and kurtosis, Per_Comp
and Agress become uninformative and may contribute to an inadequate
measure of competition for the analysis.

Descriptive statistics also were used to examine whether plants with
less-ﬁne cost allocation methods exhibit different characteristics than do
plants which have implemented FCA methods. The plants were ranked in
descending order on OVERALL and divided into rough thirds. The means of
the upper and lower thirds were tested to see whether the means were
diﬁ‘erent, as follows. The pooled estimates for the standard deviation for both
populations were calculated and adjusted for their pooled sizes. The result
was divided into the diﬁerence of their means, producing the t-value. Since
the comparison being tested was directional (whether the FCA group was

higher than the less-ﬁne allocation methods group), a one-tailed test was

84
appropriate. Using 120 d.f., t-values greater than 2.358, 1.658, and 1.289

indicated signiﬁcance at 99%, 95%, and 90% conﬁdence levels, respectively.
As shown in Table 9, four measures (Routings, Labor, Change, and Proc_Inn)
were signiﬁcantly different at the 99% conﬁdence level. Eight measures
(Product, Lower, Foreign, Cost__Bas, Top, Quant_CB, Commun, and C_Plan)
were signiﬁcantly diﬁ‘erent at the 95% conﬁdence level. Another ten
measures (Setups, Job_Shop, Compress, Lines, Plan_Cyc, FirmJII-II, Middle,
Eﬁ‘ect, C_IS_Chg, and C_Train) were signiﬁcantly different at the 90%

conﬁdence level.

CONSTRUCT DEVELOPMENT

The approach proposed for testing the hypotheses was to use factor
analysis to form constructs, to test the constructs for item reliability via
Cronbach’s Coefﬁcient Alpha, then to perform regression analysis using the
ﬁnal constructs. This approach needed to be modiﬁed somewhat because of
the small sample size obtained.

In order for factor analysis to work correctly, 3 “large” sample and
“high” observations-to-measures ratio are needed. Gorsuch (1974) maintains
1) that at least 100 observations are needed and 2) that a ratio of at least ﬁve
observations per variable included in the factor analysis is required. Other
authors, such as Nunnally (1978) and Lawrence and Yeh (1996), prefer to

have 10 observations per variable included in the factor analysis. All agree

85

 

Table 9. t-Tests for Mean Diﬂ'erences Between Plants With Implemented
FCA Methods and Plants Using Less-Fine Allocation Methods

 

Calculating Signiﬁcance of Difference in Means

 

 

 

 

 

 

 

 

Measure

Difference SD of 1 Jailed Confidence

of Means Difference t-value Level
Product 0.913 0.493 1.853 95%
Routings 1.216 0.497 2.445 99%
Ded_Eq 0.047 0.549 0.086 -
Labor 2.175 0.589 3.695 99%
Setups 0.768 0.482 1 .593 90%
Job_Shop -0.841 0.584 -1 .440 90%
Innovate 0.259 0.464 0.559 -
Compress 0.640 0.487 1.314 90%
Change 1.401 0.365 3.841 99%
Lines 1.009 0.754 1 .338 90%
Per_Comp -0.283 0.278 -1.018 -
Agrees 0.108 0.278 0.388 -
Lower 0.758 0.437 1 .733 95%
Foreign 0.889 0.523 1 .698 95%
Domestic 0.017 0.427 0.040 —
Plan_Cyc -0.521 0.371 -1 .403 90%
Typ_Shar -0.095 0.596 -0.159 -
Big_Shar 0.737 0.831 0.887 -
Ind_Comp -0.1 18 0.501 -0.235 -
Finn_HHl -1 .095 0.755 -1 .449 90%
Life_Cyc 0.085 0.620 0.137 -
Prod__lnn -0. 526 0.436 -1 .207 —
Proc_Inn -0.960 0.405 -2.368 99%
Differen -0.474 0.439 -1 .081 -
Hi_Prof -0.474 0.465 -1.018 -
Commod -0.026 0.538 -0.048 -
Low_Cost 0.071 0.421 0.169 -
Cost_Bas -0.737 0.438 -1 .684 95%
Target 0.054 0.398 0.136 -
Intense -0.071 0.375 -0.190 -
Service 0.010 0.338 0.030 -

 

 

 

86

Table 8. (Cont’d)

 

Calculating Significance of Difference in Means

 

 

 

 

 

 

 

 

 

Measure
Difference of SD of 1-Tailed Confidence

Means Difference t-value Level
CEO 0.371 0.467 0.794 -
Control 0.52 0.412 1.266 —
Top 0.706 0.353 2.003 95%
Middle 0.452 0.326 1 .386 90%
Account 0.41 1 0.409 1.005 —
Market 0.1 12 0.341 0.328 -
Level 0.319 0.474 0.674 -
Effect 0.800 0.510 1.569 90%
Resist 0.125 0.437 0.286 -
Sponsor -0.472 0.460 -1 .026 —
Sht_Ten'n 0.381 0.495 0.769 -
Quant_CB -0.980 0.428 -2.292 95%
Commun 0647 0.385 -1 .679 95%
C_Plan -0.650 0.369 -1 .764 95%
C_Consul -0.534 0.553 0965 --
C_IS_Chg -0.647 0.487 -1 .329 90%
C_Train ' -0.625 0.437 -1.430 90%
C_Dual -0.509 0.471 -1 .081 -
C_Data -0.491 0.431 -1.139 -
OVERALL 7.437 0.244 30.419 99.9%

 

 

87
that the 5:1 ratio is a minimum. Meeting the minimum ratio necessitated

reducing the number of measures to less than 25 prior to performing the

factor analysis.

Manual Reduction of Measures

Churchill (1979) discusses reducing a larger number of measures to a
smaller number by examining 1) the correlations between measures or 2) the
correlations between measures and some aggregate measure (such as the
sum of the individual measures) proposed for a group of measures. When
using either approach, Churchill advocated retaining measures which
correlated strongly. Manual examination of l) the correlations between the
measures, 2) the correlation between management support measures and the
sum of the management support measures, and 3) the correlation between
implementation obstacles and the sum of the implementation obstacles
measures resulted in dropping 27 measures. Table 10 shows the correlations
which were used in the reduction process.

First, 18 measures were dropped which did not correlate with at least
one other measure at a magnitude of 0.35. The measures dropped in this
manner are shown in the upper part of Table 11. Note that Labor is dropped
because it did not correlate with any other measure, but it is brought back at
the regression analysis stage because it had the highest correlation with

OVERALL (0.340) of any of the 50 measures.

 

 

 

 

 

 

000.0. 500.0. 500.0. N ..N.0. 000.0 000.0 50 .. .0 000.0 .0 w .0 0v .. .0 m_0>o|lm.__..
00 ...0 0 50 V00. 0. 000.0 0 ..0.0 VN00. 5.0. 0. 000.0 0v .. .0. 00 r .0. .1... $0.“.
00...0 ..50.0 000. 0. 050.0 3.0. 0. 0N...0 000. 0 000.0 5N00 V000 QEooluc.
N0 ...0. 000. 0.. 0 ..0. 0 500.0 000. 0. 550. 0. 50.. .0. N50. 0 0000. 00.0. .520 0.0
80.0 08. o. 80.0 83 to o 08 o :3. 8» o :00. ~80 .2045
N000. 5N _. .0. 0N.. 0. 0..N.0. 0N0.0. 0.0. 0 000.0. NNO. 0. 005.0. 5.. 5.0. 0.. 25.1
500.0 N.. ...0 000 0. 0: .0 000.0. v00. 0. 53.0. 0N .. .0 ..v v .0 00N0 0.30800
0NNO N30 05. 0 000. 0 03.0 03. .0 00.0 53.0 00.0 090 c920“.
000.0. 00N0 0NNO 0N0. 0 000.0 .000 0N w .0. 000.0 0 .00 0 ..0.0 526..
00 .. .0. 00 F .0 000.0 00N0 00.. .0. 0.0.0 0000. NNO. 0. 500.0. 000.0 mmwnm<
0v... 0 0VNO 000.0 0NNO 000. 0 N000 050.0 000. 0 N30 0 .. F0 9:00 .0....
000... 000. 0 v0 50. 00...0 50 .. .0 ..0..0 N5N.0 2‘0. 0 00N0 0N00 8c...
000... 00.10 0..v.0 0.. ...0. 03.0 00 .. .0 000. 0 00N0 5v .. .0 00:20
000... 0N... 0 00...0. .000 0.1.0. 000. 0 050.0 .000 mmmLQEoo
000... 50N. 0. 000.0. 000. 0 V5 .. .0 00 P .0 N5 5 .0 29105..
000. .. 00...0 .VNO. 0 00...0 00.0 000.0 A.020 new
000. .. 50 .. .0 N0 .. .0 00N0 000.0 003.00
000. .. 000.0. ..v .. .0 0v .. .0 Ewan.—
83 83 02.0 25m 80
000. .. 0 .50 005.33...
000.? 8:09.".

.85.. @0155 320500 290cc. ao..0lao.. 33.00 503 0.3001000 005501 .0320

 

0350002 =< .80 5.502 5330.000 30.800 .3 0309

 

 

 

 

 

 

 

 

00 F .0. 05.0 50.0 0N00. 000.0. 00 .. .0 «5.0. 5.0. 500.0 N500 .0000
0 5.0 50.0 00 .. .0 0 5.0 050.0. 000.0 05 F .0. 50.0 000.0 00 _. .0 05..
0NN.0 000.0 0 5.0. 000.0 05.0 50.0. 050.0. 050.0. 000.0 03.0 .0002.
50.0 5.0. 0N0.0 000.0 0 5.0. 050.0. 05.0. 050.0. 5 5.0 000.0 .5084.
N000. 000.0 000.0 50.0 0.. ...0. 50.0. 0 5.0. 00 .. .0. 50.0 .50. 202.2
50.0 000.0 0. _. .0 00 v .0 00N.0. . 5.0. 050.0 050.0. 000.0 50.0 00..
.500 0 5.0. 000.0 000.0 5N0. 55.0 50.0 000.0. 0N . .0 .5 r .0 00:00
.50 3 _..0. 000.0 000.0. 50.0. 5 F .0 0 5.0 55.0. 50.0 0 5.0. 000
05.0 000.0 00 F .0 00 F .0. 0 5.0 000.0. 0N _. .0. 50.0 05.0 0 5.0 00.200
0N0.0 050.0. 00 _. .0. 000.0. 000.0. 00 F .0 05.0 000.0 0 5.0 5 F .0 0000.5
00 .. .0 50 _. .0. .. 5.0. N000 00 .. .0 0 5.0. 00 F .0 050.0 00.. .0. v 5.0. 0.0.0...
N000 5.0. 000.0. No. F .0. 05 .. .0 N000 N50 0N P .0 000.0. N50. 000 0000
50.0 05.0 050.0. 50.0. 05.0. m 5.0. 00 .. .0. 050.0 50.0 0v F .0. .000 304
5 F .0. 0N0.0. 50.0. 5 .. .0. 00 P .0. 000.0. N0 _. .0. 50.0 50.0. 00N.0. 0022.00
000.0. N000. 000.0. .5N.0. 500.0 5 5.0 N50. 55.0. 000.0. 00 F .0. .20 _I
00 _. .0. 00N.0. 00 P .0. NON.0. 0 5.0 50.0 50.0. 000.0. 50 P .0. 050.0. conga
000.0. 05.0. 00 _. .0. v 5.0. 00N.0 50.0 500.0. 03.0. .500. 00 P .0. ::_I020
N50. 5N0. 50.0. 50.0. 000.0 05.0 000.0 000.0. 0000. N000. :0. 020
_00:_._ 000020 0020200 0.98:. 00:000.. 30.00 0000.. 0.300000 000.500 .0520

 

6.500. .0. 0309

 

 

 

 

 

 

 

 

 

000.. 005- v5.0 05.0- 00N.0. 000.0 55.0 000.0 0N5- 50.0 00. .0- 0.05:0...
000.. 000.0 05.0 05.0. 55 N. 5. 5N0- 50.0- 0.00. .000- .1: :50
000. . N... .0- v5.0. 0. 5 000.0 5. 5- 000.0- 50.0 50.0 0.000100.
000.. .000 500.0- 000.0- 03.0- 005 N50 05.0. 00:00.0
80.. 00.0 03.0 .30 .000 N. ..0 000.0 .2010».
000.. 000.0- 050. 55. 000.0. 050.0 04 .000
000.. 0... .0 5N0 .0N.0 0. 5 2.00800
000.. 055 000.0. 05.0- 00.200
000.. 050.0 050.0 0030..
000.. 000.0 0029..
000.. 0500.00

20500.... _III.E_0 02000:. 00:00.0 00:005.. 0._l.:0_0 00.000.003.20... 0030.. 002% 0200l..00
00 .0 5N0 5 . .0 050 0N. .0. 0v . .0 05.0 N000 050 5N0 ._..<mm>o
v5.0 050. 000.0 55.0. 500.0 000.0 50.0 0000 v. .0 N0 .0 0.00H0
000.0. 000.0 000.0 . 5.0 050.0. 5N . .0. 05.0 0000 000.0. 0N . .0. .030I0
405.0 050.0 00.0 v. .0. 50.0 500.0. 00.0 000.0. 50.0. 500.0. EWFIO
500.0 000.0 .000 .500. 3. . .0 000.0 50.0 0. .0 000.0 50.0. 0:0 0_I0
N000 500.0. 500.0 5. .0. N 5.0. 0N . .0. 00 .0 05.0 0000. 000.0. _00:00..0
0. .0. ON . .0 0NN.0 000.0. 000.0 0N00. 05.0 N0. .0 000.0. 000.0. :90 0
100 . .0. 000.0 00N.0 00. .0. v 5.0 500.0 0000. 5.0.0 5. .0. 000.0. 2.0.2200
00 . .0. N000. 05.0 050.0. 5. .0 00 .0 50.0. 000.0 00 . .0 .000 00 .:mac
5 5.0. 000.0 50 . .0 N0 . .0. 000.0 N0 .0 000.0 05 .0 5N. .0 0.. . .0 E00... .:0
.0000 0. .0 50. .0 000.0. 000.0 05.0. 000.0 000.0 000.0. NN . .0. 0000000
N000 00 .0 N0 . .0 N50. .N00 0N. .0 000.0. 00 . .0 000.0 000.0 .2000

500:... «000:0 0020.000 0.055. 00:000.. 002.00 0000.. 0500000 005000 .0520

 

 

8.500. .0. 0300.

 

 

91

 

 

 

 

 

 

500.0 03.0- 0000. 0N. .0- .N. .0- 03.0- 0000- 300 0.0.0- 0N . .0 00. .0- «Dorm

53.0 0. . .0 000.0- 0N0.0- 0 .0.0 00 . .0 N0. .0- 000.0 N .0.0 000.0 000.0- .90..
000.0- 000.0 VN0.0 .000 000.0 #000 .0 . .0.. 000.0 000.0- 0 .00- 000.0- «900.2
000.0 000.0- 0N. .0- 000.0- Nv0.0- NNN .0 0N. .0- N0 . .0 mm . .0- 000.0 0.0.0- «csooo<
‘0. ..0- 000.0- 000.0 000.0 VN0.0 N0 . .0 0NN.0- 0N . .0 000.0- 000.0 .000 0.022
0v . .0- 000.0 0000- 000.0 000.0- 000.0 NNN.0- .0. .0 000.0- 000.0 0N0.0- 00h
00. .0 000.0 #000- v00.0- 000.0- 0N . .0 0N. .0- 0N . .0 0. ..0- 500.0 000.0- .0.-«coo
_000.0 000.0- 000.0- 58.0- 0v0.0- N000.- N.N.0- 00 . .0 000.0- N0 . .0 0N. .0- Owo
000.0 0N. .0 0. ..0- m .0.0 0N0.0 000.0- 00 . .0 NN0.0- 000.0- Nmod 000.0- 00.200
N00.0- 00 . .0- 0000- 000.0 000.0. 000.0- 000.0- . . . .0 .NN.0 v0 . .0 N000- 00:25
<N. .0 0N . .0- v. ..0- 000.0- 000.0 N0 . .0- 03.0 00N.0 .000 v0 . .0 N000. 0.0-0.3.0
100 . .0 00 . .0 #000- vuod- 000.0- N00.0- N0 . .0- 0.0.0- hN0.0- 00. .0- 0NN.0- mum M000
Nv0.0- h .0.0 00 . .0 N000 ..0.0- Nv. .0 mNNd- 00 . .0 000.0 .v. .0. 000.0- 0000 26-.
0000.0 000.0- 0N..0- N0 . .0 .N. .0 N .0.0 00 . .0. .0 . .0 .000 0000- ..0.0- DOES-m0
00. .0 N00.0- 0.00.0- .0N.0 00N.0 NNN.0 0 .0.0 30.0- 00 . .0 0 .00- 000.0 “—90 _I
0N0.0- 03.0- onNd- 3N0 m0 . .0 00. .0 0.N.0- #000 000.0- 000.0- 0 . N.0- cam-0E0
0N0.0 N000 N000 m .0.0 00. .0 VN . .0 N000 000.0- 00. .0- 00. .0- N0 . .0- 0.0—look“.
30.0 N .0.0- 0N0.0 000.0 00. .0 000.0 000.0- 30.0 000.0- 00N.0- NVN.0- 0:. 02n—
Tﬁxoloﬁ _IIIEE mEooluE 8.0% 8:010»... 0.3020 00.3800 81.200 0030.. 009.04.. 9501.00

 

00.20000 .0. 0308

 

92

 

 

 

 

 

 

 

 

 

oooé ooto- «Rd- oood ooto Fwto owto w_-d oood woud ooEoo
ooo... ooto omod wuto owud ooto o_. to :to- oood 3:25
ooo.? Food. F F to. mood mood oood- owod- m? to- “moon-_-
oooe o _- to oood mm to oo v .o mood- wood mom Ioooo
ooo... wwod Pond owod nwod- omto “moo 26..
ooo... mo _- .o nwwd oood ow F .o ooEEIoo
ooo.? wrod wnmd toad 35 _I
ooo. _. wood mmod conga
ooo. _- omod cEIooi
ooo.? cc. not
83.5w 032:. «No.3 ommloooo 30045.. 35:60 behalf c2920 c5185 CEIUEQ

Food Not? omod- onod Sod- wood. oood- mnto owto owod tod- -_._<mm>o
«wod mmod. Food oood. omod owod- Rto o8d- omto- oood Food- SmoHo
wood omod owod mmod. wuod o3d- wood wood owod m8d- mood- _uzolo
:mFto Food. Rod moto- wmto- unto- wwod- wood owod omto- mood- cry-Flo
mood wwod owod. Sod- E-od owod owod osto mood onod- omod- ago 910
mmod. m_.od- oood mood omod urto owod- Food. Sod. moto- oSd- Sacco-.0
tad oSd- oood moto- Nrto- owod. mood- Cod hood- Bod- oood- cos. 0
wmto Std oood- Food «Nod- oBd- ooto- wood owod Foto- noto- cﬂEEoo
wmod Etc- wmod oood mmod oood- wood owod «Ed oood- mood- mo Emao
Std oood- Sod- wood oood _.wod- oood ooto Fuod oood- moto- Eo... Em
wwod owod- owod oood- R P d- wmod- owod wr to oood- F P to- Rod .omcoaw
o..o.o- mmod owto mmod. Foto- oood Foto owod oood swod- umod- £31

53018... _IIIEE mEoOIus .9.on .acmleﬁ 345$ 8383 c920“. .26.. 32% decodon.

 

 

8.9.39 .3 039—.

 

 

 

93

 

 

 

 

 

ﬂood. oood oood mo F .o. oood w Fod. wood. oood. o Fwd. Food. ._..<mm>o
oo F .o wood. mood oood. mo F .o. omod. oood- o 3.9 mmod No to Boo-Id
:ow F .o mwod. wood F F to Food. oood- mwod. Fwod. o Fod oood. _oaolo
wm to wood mwod Food m Fod. oood. oood. o Fod om F .o no F .o amt-Io
oo F .o oood m 5.? wood wwod- oood. Fwod wood F Fwd oood one o_lo
wood mwod. w Fo.o omod o Fod. oood. F Fod oo F .o om F .o oo to _aocOOIo
NF to wF to. mwod. Fmod oood- F F to. owto- owod oood omod coE o
om to mood. on F .0. wood mood. omod. oood ww F .o omod oo F .o cDEEoo
oood owod. Nood- F Fod. F F to. mm to. oood oood oomd oo F .o mo Elana
«o F .o oo to o F F .0- Food. oood. wood. oood. 9 to oo F .o m Fwd Em... Eo
om F .o wo to. owod. oood owod. mood. Fo F .o. owod- oF Fd omod boocoao
3 to Food. owod. Fo F .o. wood oo F .o. omod. owod. o Fwd on F .o 3:31
o F to. wmod. Food. wood. oo F .o. oood oFNd. mwod. Fm F .o. oood- 695
Food. oo F .o oood. mood oood omod. oood mwod Food. w 5d. .05..
mood. mo F .o owod Fm to mwod wood owod oood. oood. owod «9:22
o Fod. oood mmod omod N Fo.o mwod. oood. oood omod. wood. anoo<
mood. 3 to F Fod oo F .o. oood wood oood. oood oood. mood. 22::
oo F .o. wwwd oood oood. wood oood on F .o. wo to oo to. omod. no...
oood. omto mood. omod m Fo.o oood. oood. oood oood. owod. .2200
fommd. wood wood mwod o Fod oood. oood owod omod. Nood- Owo
woos-am 3:25 693. oooloooo 500450.. 35:80 3.53: 22220 cc_loo.n_ cc _looi

 

 

Goo—SUV .oF 03oF

 

 

94

.965 9: c. 3539: $50
5.3 32o... 922.8 __o no.5; .8539: tendon _otooocoE *0 .883: 9.: 9.632 .8 non: a. £5. domq on.
$28 mos-ooo.: 05 .0 :5» on. 5.3 5:28 on. .6 a2 on. on 9:88.. on. .6 5.6.9.8 on. @505 Sam 2.86 .202

 

 

 

 

 

.83 .88 omod mood 88 83 o8.o 83 ..._<mw>o
.83. 80.0. wad. 83. 93- 8nd. mood. at? «some
83. 20.0. omto. mome- ~:.o. 83. 80.0. mood. _aaono
33. 8.? 8m? 83. 83. :3. o8...- NoFd- smﬁno
.83. .88 23- med. 83. 80.0. oood. SF? 80 o...o
wond- wmwd. 8N9 owod. 83. 83. oood. F8? 328-0
83 8d. mmod. 2.8. 83. 83. 83. 33. cm... 0
9&9 oood wowe- 83. 83. 83. 83 08.9 5558
www.c- owto. mmod. two. 83. 8N? Swo- m8.9 mo .68
wowe- oood mmod. 80.0. mood- mmod- 89o. :3. 5.3 so
55.? 83. mwod. 83. 8nd- o8¢ 83. 8.8. .888
mm;- owod. 9N9 8nd. woo? 8m? wow? 83. 3.3m
«8... 8m... 8m... «8... «8... 2»... .8... 2m... 55 82o
.83 ~88 wag 83 23 63 mmod 8.3 58m
.83 omod :8 83 88 too 63 .96..
o8.F 8.8 8.3 mwwd 8.8 mmwd .982.
o8.F omod 88 83 83 .882
83 mm; «8.0 omod 28.5.
83 88 28 8F
O83 omod .928
08.3 8.0

85 .96.. Evans. .882 28.2 8F .280 80

 

 

 

 

9558 do ﬂown.

.0306 0.... 0. 0050008 .050 5.?
22oz 0.0.0.000 =0 50...? 80.30008 00.03000 c0..o.coE0.QE_ .6 .0053: 0... 3.0300. .0. 000: 0. 0.:... down 05
000.00 005000.... 05 .0 E30 05 5.3 053.00 05 .0 00. 0.... .0 05000:. 0:. .0 0030.00.00 05 03050 Edo 030.0 ”0.02

 

 

 

030. 3:0. BS. 83. 02.0- 03.0- 00.0- 02.0- 80.0 30.0. 80.0 ....<mm>o
38 80.0 02.0 02.0 20.0 02.0 80.0 02.0 08.0 2.0.0 :00 50o 0:20
83 000.0 00w.0 08.0 80.0 08.0 30.0 08.0 mmod 08.0 08.0 smouu
08.F 80.0 :80 8w.0 0mw.0 :20 80.0 .30 000.0 03.0 80.0

08.F 20.0 08.0 08.0 08.0 8.3 80.0 5.0 000.0 smmu0

08.F 8.8 000.0 08.0 mww.0 08.0 F30 20.0 80 9.0

83 mw00 «3.0 0.3.0 F000 80.0 02.0 880.0

83 08.0 00w0 88.0 80 80.0 8.0 0

08.F 080 3.3 000.0 80.0 00.20.00

83 m0w.0 F30 080 00 .080

83 03.0 F30 ES 20

08.3 80.0 800%

08.F 8.80

 

 

@0010 .3010 0.3.0 80:03 .0.-.8010

005.10 58800 mOIano _EoHIEo .omcoaw 8.00m

 

 

8.308 d. 030,—.

 

 

Table l 1. Measures Identiﬁed and Dropped While Reducing Number of
Measures

18 Measures Which Did Not Have a Correlation of at Least 0.35 With Any
Other Measure:

Ded_Eq Domestic Low_Cost
Labor’ Plan_Cyc Cost_Bas
J ob_Shop Ind_Comp Target
Lines Firm_HHI Intense
Lower Life_Cyc Service
Foreign Hi_Prof Eﬁ‘ect

* Note that Labor did not correlate with any other measures, but it had the
highest correlation (0. 340) with OVERALL of any measure. It is later added
as a separate variable in the regression model.

9 Measures in Managerial Support and Implementation Obstacles Which
Correlated Least Strongly With the Appropriate Group Sum:

CEO Market Sht_Term
Middle Resist C_Consul
Account Commun C_Dual

2 Nearly Identical Measures Which Did Not Correlate With Other Measures:

TYILShar Big_Shar

2 Measures Correlated Only With Each Other:

Per_Comp Agress

4 Measures Dropped Later During Item Reliability Testing With Cronbach’s
Coeﬂicient Alpha:

Commod Innovate
Compress Change

97

Since seven of the eight managerial support measures and all the
implementation obstacle measures survived this examination but more
measures needed to be dropped, correlations of these two groups of measures
with their aggregates were examined. By dropping some of the measures
with the weakest correlations with the aggregates, the nine measures shown
in the lower part of Table l l were identiﬁed and dropped.

Two additional measures (Typ_Shar and Big_Shar) were dropped
because they measured essentially the same thing and did not correlate
strongly with anything else; two other measures (Per_Comp and Ag'ress)
were dropped because they correlated only with each other.

As will be explained below, the remaining 19 measures were used in a
principal components factor analysis with varimax rotation; four factors were
found which roughly corresponded to the intended factors, although no factor
emerged for competition. Each factor was then tested for item reliability by
calculating Cronbach’s Coeﬂicient Alpha; this test resulted in the elimination

of another four measures (Commod, Innovate, Compress, and Change).

Factor Analysis, Item Reliability Testing, and Generation of Factor Scores

Hypothesis testing had been designed to use ﬁve constructs:

COMPLEX, COMPET, STRATEGY, SUPPORT. and OBSTACLES. The

primary measures contributing to each construct were to be identiﬁed

98
through factor analysis and proven reliable by calculating Cronbach’s

Coefﬁcient Alpha. Values of the constructs for regression analysis were to be
factor scores generated by multiplying standardized scoring coefﬁcients (not
factor loadings) by the associated raw scores of the measures used in the
factor analysis to produce an optimally weighted linear combination of model
measures.

The approach described above had to be modiﬁed somewhat because of
the small sample size obtained. As discussed previously, proper functioning
of factor analysis required that an observations-to-measures ratio of at least
5:1 be maintained. When the number of variables had been reduced to 19
(which met the 5: 1 ratio requirement), analysis proceeded in accordance with
the approach described above.

The data were factor analyzed using principal components factor
analysis with varimax rotation. Factor analysis was used to verify that
measures within a construct belonged together for measuring an underlying
cause; the principal components approach was used because it is robust in
dealing with nonnormal data [Johnson and Wichern (1992)]. Varimax
rotation of the factors was used to produce an orthogonal solution in which
the constructs are independent, and the nature of the hypotheses assumes
independence of the ﬁve constructs. Due to missing data, the factor analysis
(as well as the regression analysis) was based on 1 12 observations.

The factor analysis yielded four factors. One factor matched well with

complexity and contained Product, Routings, Setups, and Commod. A second

99
factor had measures from both the complexity and strategy groups and

contained Innovate, Compress, Change, Prod_Inn, Proc_Inn and Diﬁ'eren. A
third factor matched well with managerial support and contained Control,
Top, and Level. The fourth factor matched well with Implementation
Obstacles and contained Sponsor, Quant_CB, C}lan, C_IS_Chg, C_Train,
and C_Data. No factor emerged for competition; the measures did not
correlate well enough.

With the individual measures grouped through factor analysis, the
measures in each factor were suggested as measuring one latent variable.
The item reliability of the measures in each factor was examined by
calculating Cronbach’s Coefﬁcient Alpha to see if the measures in the factor
formed reliable constructs. For a construct to be reliable, an alpha of
approximately 0.7 or higher was desirable. The factors for managerial
support and implementation obstacles both had alphas above the threshold,
but the ﬁrst two factors did not. By eliminating measures identiﬁed as
correlating least strongly with other measures in the group, the ﬁrst two
factors also became reliable constructs. Results of item reliability analysis
for the four constructs are shown in Table 12. Table 12 contains the
construct overall alpha, how well each measure correlates with the construct
total, and the value to which the construct alpha would change if that
measure were dropped.

With the constructs identiﬁed and tested, the data on each plant were

processed in a factor analysis package to calculate scoring coefﬁcients and to

100

 

Table 12. Item Reliability Testing Results for the Four Factors Identiﬁed in
Factor Analysis

Conplexity Construct

Cronbach Coefficient Alpha for STANDARDIZED variables: 0.710122

 

 

 

Deleted Variable Correlation with Total Alpha if dropped
PRODUCT 0.671636 0.432258
ROUTINGS 0.603312 0.524274
SETUPS 0.340447 0.835553
CONNOD (DROPPED) (DROPPED)

Strategy Construct

Cronbach Coefficient Alpha for STANDARDIZED variables: 0.715454

 

 

 

Deleted Variable Correlation with Total Alpha if dropped
PROD_INN 0.597127 0.547533
PROC_INN 0.601591 0.541777
DIFFEREN 0.415911 0.764974
INNOVATE (DROPPED) (DROPPED)
COMPRESS (DROPPED) (DROPPED)
CHANGE (DROPPED) (DROPPED)

Nanagerial Support Construct

Cronbach Coefficient Alpha for STANDARDIZED variables: 0.845865

 

 

 

Deleted Variable Correlation with Total Alpha if dropped
CONTROL 0.822227 0.636045
TOP 0.710469 0.787365
LEVEL 0.600577 0.888956

Iapleaentation Obstacles Construct

Cronbach Coefficient Alpha for STANDARDIZED variables: 0.846922

 

 

 

Deleted Variable Correlation with Total Alpha if dropped
SPONSOR 0.565542 0.833610
OUANT_CB 0.607573 0.825656
C_PLAN 0.638262 0.819759
C_IS_CHG 0.643148 0.818814
C_TRAIN 0.721229 0.803435
C_DATA 0.594967 0.828057

101
generate factor scores for each observation. Table 13 shows the factor

loadings associated with each measure and the standardized scoring
coefﬁcient for each measure as generated for each construct. Scoring
coeﬂicients related measures to the constructs and represented optimal
weights on a plant’s data to produce linear composites known as estimated
factor scores. In this study, estimated factor scores were to be calculated for
the constructs COMPLEX, COMPET, STRATEGY, SUPPORT, and
OBSTACLE, although the COMPET construct failed to emerge from the
analysis and was not calculated. Factor scores were computer-generated

using the following equation:

F’: = EbiiVi,

where
F’,- = the estimated factor score for factorj
bi: = the scoring coefﬁcient for measure i of factor j

v,- = the respondent’s score for measure i.

The estimated factor scores are the values of the independent variables
in the multivariate regression analyses used to test the hypotheses. Table 14
shows the estimated factor scores for a few of the responding plants. Note
that, where any values of the individual measures composing a factor are
missing, the whole observation is dropped from analysis. Twelve
observations were dropped, leaving 1 12 observations to be used for the

analysis.

102

 

Table 13. Factor Loadings and Standardized Scoring Coeﬂicients for the

Constructs
Rotated Factor Pattern

OBSTACLE SUPPORT COMPLEX STRATEGY

PRODUCT -4 8 82 * -16
ROUTINGS 11 3 78 * -10
SETUPS 1 2 4O * 8
PROD_INN 7 -7 3 69 *
PROC_INN 13 -8 6 72 *
DIFFEREN -2 12 -19 50 *
CONTROL -10 89 * 1O -5
TOP -15 82 * -1 -5
LEVEL -7 65 * 4 4
SPONSOR 60 * -28 -8 -2
OUANT_CB 63 * -17 14 10
C_PLAN 7O * -9 -15 -3
C_IS_CHG 71 * 6 11 9
C_TRAIN 80 * 1 -8 5
C_DATA 62 * -8 17 9

NOTE: Printed values are lultiplied by 100 and rounded to the nearest integer.
Values greater than 0.3 have been flagged by an '*'.

Standardized Scoring Coefficients

OBSTACLE SUPPORT COMPLEX STRATEGY

PRODUCT -0.02368 -0.01608 0.49030 ~0.07536
ROUTINGS 0.02753 -0.02805 0.36191 -0.01265
SETUPS -0.00232 0.01066 0.10069 0.04823
PROD_INN -0.02904 -0.00056 0.03984 0.36475
PROC_INN 0.00365 -0.00637 0.05987 0.42270
DIFFEREN -0.01387 0.03439 -0.02525 0.19961
CONTROL 0.06401 0.55606 0.02582 -0.01657
TOP -0.00810 0.29648 -0.06418 0.00585
LEVEL 0.01084 0.14808 0.00185 0.04320
SPONSOR 0.14821 -0.04824 -0.03708 -0.04914
0UANT_CB 0.15651 -0.04153 0.07060 0.03273
C_PLAN 0.22676 0.04805 -0.08884 -0.06181
C_IS_CHG 0.20896 0.07167 0.05240 0.01919
C_TRAIN 0.32022 0.05567 -0.06569 -0.02728
C_DATA 0.15531 0.01687 0.04525 0.01761

103

 

Table 14. Factor Scores Calculated for Constructs cf 10 Observations

OBSERVATION OBSTACLE SUPPORT COMPLEX STRATEGY
18 0.61185 0.53898 0.45502 -0.38267
19 0.48885 -0.26316 1.23050 -0.72356
20 -0.27707 0.46768 1.20343 0.26033
21 0.39591 0.67268 -0.32788 ~0.73173
22 (Missing Data)
23 -0.25657 0.80836 0.93625 -0.91673
24 -1.30105 0.86135 -0.15359 -0.87234
25 -0.38575 1.18918 1.31309 1.15749
26 -0.87283 0.18857 -1.06576 1.46196
27 0.34228 -2.39622 -0.43336 -0.89093

Note 3

where any aeasures needed for the factor analysis were lissing, no factor score
was calculated for that observation. Missing data occurred 12 tiles, leaving 112
observations for analysis.

MULTIVARIATE REGRESSION ANALYSIS

Various regression models were used to test the data. These models
varied in the number of constructs used and in the inclusion of some
combination of Labor, Proc_Inn, Cost_Bas, Per_Comp, Lower, and Firm_HHI
as separate independent variables. Since the STRATEGY construct was
usually only marginally signiﬁcant, the individual measures were also tried
in models to see if they performed better than the construct. Proc_Inn

correlated highly with the construct STRATEGY and was substituted for that

104
construct; Proc_Inn was never used in a model including STRATEGY.

Cost_Bas, on the other hand, did not correlate highly with STRATEGY and
was used in addition to that construct. Since no cohesive factor was found for
the competition construct, the individual measures were included in models
to add the inﬂuence of competition, which appeared to be nonmonolithic.
Three measures were found to improve the model: Per_Comp, Lower, and
Firm_HHI. Labor was used separately because it seemed to represent a
diﬁ‘erent type of complexity and correlated very highly with OVERALL; since
Labor did not correlate with COMPLEX, it was used in addition to the
COMPLEX construct. The models took one of two major forms, depending on
whether or not Proc_Inn was substituted for STRATEGY. Within either
major model, some of the independent variables could be omitted. The two

major models were as follows:

OVERALLi = [30 + BICOMPLEX: + B,STRATEGY: + [3,,SUPPORTi
+ B,OBSTACLE, + [3,,Labori + [36Cost_Basi
+ [3.,Firm_I-IHIi + [3,,Per_Compi + [3,,Loweri + 5103:

and

OVERALLi = [30 + BlCOMPLEX: + BzProanna + [3,,SUPPORTi
+ [34OBSTACLEi + [3,,Labori + BGCosLBasi
+ B.,Firm_HHIi + [3,,Per_Compi + 89Loweri + Blot:i

105
As indicated earlier, regressions based solely on the constructs used

112 observations. The other models used either 111 or 112 observations;
models with Firm_HHI had one fewer observation.

The constructs and separate measures used as variables were
independent, as illustrated in Table 15, which shows the Pearson Correlation
Matrix of the constructs, the separate measures, and OVERALL. Note that
Proc_Inn and STRATEGY had a very high correlation (0.876). This single
high correlation posed no problem because the two variables were not used in
the same model; rather, Proc_Inn was substituted for STRATEGY. Table 15
also shows the simple statistics (including the number of observations (N),
means, standard deviations, sums of variation (Sum), minimum value and
maximum value) for the constructs, separate measures, and OVERALL. As
the zero means and sums of variation show, the constructs OBSTACLE,
SUPPORT, COMPLEX, and STRATEGY were all standardized, whereas the
separate measures and OVERALL were not standardized.

Table 16 shows the results for the eight regression models, including
the F-value, Prob > F, the adjusted R2, and the test of ﬁrst and second
moment speciﬁcation’s Prob > Chi-Square (White’s Test); Table 16 also shows
the conﬁdence level of each independent variable in the model (“--” indicates
that the construct or separate measure was not included in the model). Table
17 shows more detail on assessing the models, including the variable

coeﬂicients, the standard errors of the coefﬁcients, the t-values of the

106

 

Table 15. Pearson Correlation Matrix and Simple Statistics of Constructs
and Separate Measures Used in Regression Analyses

Pearson Correlation Matrix

 

 

 

OBSTACLE SUPPORT COMPLEX STRATEGY LABOR COST_BAS PROC_INN
OBSTACLE 1.000
SUPPORT -0.036 1.000
COMPLEX 0.014 0.021 1.000
STRATEGY 0.037 -0.012 -0.041 1.000
LABOR 0.077 0.036 0.157 0.019 1.000
COST_BAS 0.007 0.016 -0.030 -0.024 0.042 1.000
PROC_INN 0.146 -0.089 0.063 0.876 -0.015 -0.104 1.000
FIRM_HHI -0.067 0.092 -0.200 0.034 -0.047 0.108 0.078
PER_COMP -0.034 -0.062 0.123 -0.211 0.070 -0.225 -0.137
LOWER 0.027 -0.085 -0.012 -0.081 -0.123 -0.027 -0.156
OVERALL -0.150 0.137 0.225 -0.103 0.340 -0.187 -0.209
FIRM_HHI PER_COMP LOWER OVERALL
FIRM_HHI 1.000
PER_COMP -0.051 1.000
LOWER -0.051 0.070 1.000
OVERALL -0.192 -0.077 0.148 1.000
Simple Statistics
Variable N Mean Std Dev Sum Minimum Maximum
OBSTACLE 112 0 0.919 0 -2.18 2.16
SUPPORT 112 0 0.924 0 -2.98 1.23
COMPLEX 112 0 0.884 0 -1.90 1.56
STRATEGY 112 0 0.827 0 -1.55 2.38
LABOR 123 4.707 2.307 579 1.00 7.00
PROC_INN 124 3.395 1.508 421 1.00 7.00
COST_BAS 124 3.871 1.771 480 1.00 7.00
PER_COMP 124 6.016 1.097 746 2.00 7.00
LOWER 124 4.766 1.648 591 1.00 7.00
F1RM_HHI 123 8.207 2.959 1009 0.17 10.00
OVERALL 124 5.798 2.744 719 1.00 11.00

107

 

Table 16. Assessment of Models Used in Regression Analyses

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Model1Model2 Model3 Model4 Model5 Model6 Model? Model8
Model Assessment:
F-Value 2.905 3.580 3.999 3.404 5.984 5.024 5.399 5.527
Prob > F 0.025 0.016 0.010 0.012 0.000 0.000 0.000 0.000
Adj. R-Square 0.064 0.065 0.075 0.080 0.152 0.153 0.265 0.270
White's Prob > Chi-Square 0.122 0.180 0.154 0.266 0.138 0.095 0.675 0.736
Signiﬁcance of Variables:
OBSTACLE 94% 95% 92% 91% 97% 97% 99% 98%
SUPPORT 91% 91% - none 90% 90% 94% 91%
COMPLEX 99% 99% 99% 99% 97% 97% 99% 98%
STRATEGY none - — — - none 94% -
Labor — - - — 99.9% 99.9% 99.9% 99.9%
Proc_Inn - — 96% 95% - - — 96%
Cost_Bas — - — - - - 98% 98%
Per_Comp - - - - -- - 98% 99%
Lower - - - -— - — 97% 97%
Firm_HHI — - — — - - 97% 96%

 

 

coefﬁcients, and the one-tailed probabilities for the variables to be used in
testing hypotheses.

As shown in Table 16, the Prob > F indicated that each of the eight
models was signiﬁcant at the 97% or higher conﬁdence level. Since the
White’s Prob > Chi-Square was greater than 0.05 for each model, White’s test
identiﬁed no problems with heteroskedasticity or model speciﬁcation for any
of the models [White (1980)].

When the constructs were taken alone (models 1 and 2), STRATEGY

was not signiﬁcant; OBSTACLE, SUPPORT, and COMPLEX were signiﬁcant

108

 

Table 17 . Assessment of Independent Variables Used in Regression Models

Parameter Standard T for H0: 1-Tai1ed
Model Variable Estimate Error Parameter=0 Prob > |T|
Model 1
INTERCEP 5.688 0.249 22.885 0.0000
OBSTACLE -0.431 0.272 -1.583 0.0582
SUPPORT 0.372 0.271 1.374 0.0861
COMPLEX 0.680 0.283 2.405 0.0090
STRATEGY -0.285 0.302 -0.943 0.1740
Model 2
INTERCEP 5.688 0.248 22.897 0.0000
OBSTACLE -0.441 0.272 -1.621 0.0540
SUPPORT 0.374 0.270 1.385 0.0845
COMPLEX 0.691 0.282 2.447 0.0080
Model 3
INTERCEP 6.693 0.624 10.725 0.0000
OBSTACLE -0.384 0.273 -1.407 0.0812
COMPLEX 0.729 0.281 2.594 0.0054
PROC_INN -0.292 0.166 -1.755 0.0411
Model 4
INTERCEP 6.631 0.624 10.619 0.0000
OBSTACLE -0.377 0.273 -1.382 0.0850
SUPPORT 0.336 0.269 1.248 0.1074
COMPLEX 0.720 0.281 2.566 0.0059
PROC_INN -0.274 0.166 -1.645 0.0515
Model 5
INTERCEP 4.015 0.536 7.492 0.0000
OBSTACLE -0.510 0.260 -1.964 0.0261
SUPPORT 0.342 0.258 1.327 0.0937
COMPLEX 0.544 0.272 1.999 0.0241
LABOR 0.359 0.103 3.477 0.0004

109
Table 17. (Cont’d)

Parameter Standard T for H0: 1-Tailed
Model Variable Estimate Error Parameter=0 Prob > |T|
Model 6
INTERCEP 4.003 0.536 7.474 0.0000
OBSTACLE -0.500 0.260 -1.926 0.0284
SUPPORT 0.339 0.257 1.316 0.0956
COMPLEX 0.531 0.272 1.950 0.0269
STRATEGY -0.308 0.288 -1.072 0.1430
LABOR 0.362 0.103 3.504 0.0004
Model 7
INTERCEP 8.722 1.761 4.951 0.0000
OBSTACLE -0.573 0.246 -2.331 0.0109
SUPPORT 0.378 0.243 1.553 0.0618
COMPLEX 0.495 0.263 1.886 0.0311
STRATEGY -0.441 0.276 -1.595 0.0569
LABOR 0.408 0.098 4.165 0.0000
COST_BAS -0.280 0.132 -2.123 0.0181
FIRM_HHI ~0.159 0.081 ~1.954 0.0267
PER_COMP -0.634 0.210 -3.016 0.0016
LOWER 0.269 0.136 1.983 0.0250
Model 8
INTERCEP 9.622 1.889 5.094 0.0000
OBSTACLE -0.517 0.248 -2.087 0.0131
SUPPORT 0.339 0.244 1.389 0.0840
COMPLEX 0.552 0.263 2.099 0.0192
LABOR 0.400 0.098 4.096 0.0000
PROC_INN -0.280 0.153 -1.827 0.0353
COST_BAS -0.297 0.132 -2.250 0.0133
FIRM_HHI -0.144 0.082 -1.760 0.0407
PER_COMP -0.615 0.207 -2.976 0.0018
LOWER 0.255 0.136 1.880 0.0315

110
but adjusted R2 was very low (6% to 6.5%), indicating that the constructs

accounted for only about 6% of the variation. Substituting Proc_Inn for
STRATEGY (see models 3 and 4) made a slight improvement in the adjusted
R2 (to 8%), but adding Labor (see models 5 and 6) caused a notable increase
in the adjusted R2 (to 15%). When Cost_Bas was added (not shown), adjusted
R2 increased to nearly 19%. When the competition measures Per_Comp,
Lower, and Firm_HHI were added, the adjusted R2 made another notable
increase to 27%. Adding the additional measures also helped stabilize the

original constructs’ signiﬁcance, since all constructs were signiﬁcant at a 9 1%
or higher conﬁdence level. In Model 7, the four constructs were signiﬁcant
at the 94% or higher conﬁdence level; the separate measures were each

signiﬁcant at the 97 % or higher conﬁdence level.

HYPOTHESES TEST RESULTS

Testing each of the hypotheses depended on two indicators. The ﬁrst
was whether the sign of the parameter estimate agreed with that
hypothesized. The second was whether the parameter was signiﬁcant.

The parameter signs consistently supported the hypotheses for
complexity (H1), managerial support (H4), and implementation obstacles (Ha)
since each had the hypothesized sign (positive for H1 and H4 and negative for

H5). In addition, the sign for Labor (a complemty measure) was always

111
positive, as hypothesized The negative parameter for strategy (Ha),

however, was consistently opposite of that hypothesized. Similarly, the signs
for Proc_Inn and Cost_Bas (strategy measures) were consistently negative.
N 0 construct emerged for competition (H4), but the three competition
measures used in regressions, Firm _HHI, Per_Comp, and Lower, had mixed
results, with Firm_HHI and Per_Comp having opposite signs from that
expected and Lower having the anticipated positive sign.

The complexity hypothesis (I'll) is supported by COMPLEX’s
signiﬁcance at the 97% to 99% conﬁdence level in all models. The
competition hypothesis (H2) cannot be supported directly since COMPET
failed to emerge, but the three separate measures were signiﬁcant at the 96%
to 99% conﬁdence level. The strategy hypothesis (He) was not supported in
the constructs-only model but was signiﬁcant at the 94% conﬁdence level
when the competition measures were added to the model. The managerial
support hypothesis (H4) was weakly supported at the 90% to 94% conﬁdence
level. The implementation obstacles hypothesis (H5) was supported at the
92% to 99% conﬁdence level.

In view of the sign and signiﬁcance indicators, the data strongly
supported the complexity hypothesis (1'11) and the implementation obstacles
hypothesis (H5). The data weakly supported the managerial support (H4)
hypothesis. Unless the separate competition measures were included in the

model, the data did not support the strategy hypothesis (Ha); with the

112
separate competition measures, the strategy hypothesis was weakly

supported. Considered with or without the separate competition measures,
the strategy hypothesis is supported only at the 82% to 94% conﬁdence level.
The data could not directly support the competition hypothesis (H2) since no
cohesive construct emerged for competition. However, separate competition
measures were highly signiﬁcant but mixed in sign, which indicated that the

competition hypothesis was not supported.

DISCUSSION

Demographic Analysis

Examination of demographic data in comparison to the dependent
measure by using MANOVA procedures revealed only two demographic data
items with signiﬁcant diﬁ'erences. The ﬁrst demographic data item,
respondents’ years in company, may be a result of the positions to which the
surveys were sent, but there was no reason to suspect that this possibility
biased the results. The second demographic data item was 4-digit SIC code.
Diﬂ‘erences between industries were expected; there was no reason to think

industry differences adversely affected the study.

l 13
Measures, Factors, Item Reliability, and Constructs

Review of the descriptive statistics revealed major problems of
skewness and kurtosis in all but seven measures. Nearly all the measures
had set minimum and maximum values, eliminating the possibility of
outliers. However, the skewness and kurtosis showed the measures to be
nonnormal, precluding the use of maximum likelihood factor analysis since it
required normal data. Principal components factor analysis, however, was
robust in regards to nonnormal data, so that method was used.

The small sample size complicated the factor analysis. Too few
observations in comparison to the number of measures analyzed would result
in unstable factors, where the measures might load incorrectly. It was
necessary to reduce the number of measures to a 5:1 ratio of observations to
measures. Since only 124 observations were received and since some
observations were missing data, the reduction process sought to limit the
number of measures to 20 or less. Manual observation of correlations
between measures or between a measure and its group sum reduced the
number of measures to 23. Four additional measures were dropped because
they related only to one measure and factors should have at least three
measures. Since the measures dropped did not correlate with the other
measures, the resulting input set represented the best set of measures to be

used under the circumstances. Since most of the measures used in the study

114
had not been previously developed and tested, it was not unusual that so

many of the measures did not measure a common latent variable. The lack of
cohesiveness among the measures within a group suggests that the survey
had strong measurement error. Better pretesting of the survey could have
avoided this measurement problem in the actual survey.

Principal components factor analysis was performed using varimax
rotation to insure independence of the factors. At least one cohesive group of
three measures was found for each of the ﬁve areas hypothesized except
competition. The resulting factors were analyzed for item reliability using
Cronbach’s Coeﬂicient Alpha; the reliability process resulted in eliminating
four more measures so that alphas of at least 0.70 were obtained for each
factor. Thus, complexity was represented by Product, Routings, and Setups,
which are consistent with the search for a resource consumption pattern
where consumption is diverse and uncorrelated with the base. No factor
emerged for competition, so latter analysis was forced to use separate
measures in an attempt to cover competition. The lack of a cohesive group
raises doubts as to the validity of the underlying assumption for Hypothesis
2 that competition is a cohesive latent variable. Strategy was represented by
Prod_Inn, Proc_Inn, and Diﬂ‘eren, which all related to Hypothesis 3a; no
factors emerged for hypotheses 3b and 30. Any conclusions drawn about
Hypothesis 3 relate only to Hypothesis 3a. This lack of overall cohesiveness

in the group of measures also raises doubts about assuming strategy to be a

115
cohesive latent variable. Managerial support was represented by Control,

Top, and Level. Given that nearly all the measures suggested for this group
had high correlations and were reduced primarily based on which correlated
most strongly with the group sum, this factor could well have had more
measures if they had been included in the factor analysis. The
implementation obstacles group was represented by Sponsor, Quant_CB,
C_Plan, C_IS_Chg, C_Train, and C_Data. The measures in this group
correlated so strongly together and with the sum of the group that it was
diﬁcult to eliminate more measures.

After the item reliability analysis, the measures remaining were again
processed through the factor analysis procedure to generate factor scores,
which became the values representing the constructs for each observation.

In addition to the constructs, several measures were identiﬁed for use
in the regression analysis. Labor was retained because it represented a
different type of complexity where many labor rates (representing diﬂ'erent
labor resources) were required. The measure Labor did not correlate well
with any other measures, but it correlated with OVERALL higher than did
any other measure. Where the strategy group had proven mostly
insigniﬁcant in hypothesis testing, two measures (Proc_Inn and Cost_Bas)
were examined as substitutes for STRATEGY. Finally, three measures
(Per_Comp, Lower, and Firm_HHD were retained to represent competition,

which had failed to form a cohesive group.

116

Regressions

Several regressions used the data to test the hypotheses in regards to
parameter signs and parameter signiﬁcance. The data consistently
suggested Hypothesis 1 (complexity) to be valid and supported. The
complexity of product and process, resulting in resource consumption
patterns which were diverse and uncorrelated to the allocation base, was
related to the implementation of FCA methods, with increasing complexity
being associated with implementation of ﬁner allocation methods. The data
suggested that Hypothesis 2 (competition) was invalid in that competition
was not a cohesive latent force. Some of the measures suggested for
competition were signiﬁcant, but the direction of the sign was inconsistent.
No signiﬁcant association was found for the two life-cycle measures for
testing Hypothesis 2a; therefore, Hz. was not supported. Firm_HHI, an
individual measure for testing Hypothesis 2b was signiﬁcant, but the sign
was reversed; therefore, 1121. was not supported. Lower and Per_Comp, two
individual measures for testing Hypothesis 2c, were signiﬁcantly associated
with implementation of FCA methods, but the two measures took opposite
signs, so I'IZc was not supported. The data suggested that Hypothesis 3
(strategy) was not supported in the area of differentiation in product and

process (Hypothesis 3a). There was an association between diﬁerentiation in

117
product and process and the implementation of FCA methods, but the sign

was opposite that expected. No constructs emerged for testing Hsb or H3c; nor
did signiﬁcant associations emerge for individual measures to test those
hypotheses. The data also suggested that Hypothesis 4 (managerial support)
was valid; the association between managerial support and FCA method
implementation was signiﬁcant at a lesser conﬁdence level and in the
anticipated direction. The data strongly suggested that Hypothesis 5
(implementation obstacles) was valid; the signiﬁcance and sign of the
association were as expected.

This study suffered from ﬁve problems. First, there was a large
amount of measurement error, as evidenced by the large portion of measures
which could be eliminated during the reduction of measures based simply on
the failure to correlate with any other measure at a 0.35 level. Better
pretesting and reworking of the survey instrument could have alleviated this
problem.

The second problem was the small sample size. A much larger sample
size would have eliminated the need to reduce the number of variables. A
large sample size also increases the likelihood of measures loading correctly
onto the factors.

The third and fourth problems resulted ﬁ'om improper research design.
The third problem was that hypotheses assumed cohesive underlying forces

where none existed. Hypotheses 2 and 3 each assumed a cohesive latent

118
force; data did not support this assumption. Splitting the hypotheses into

subhypotheses helped correct this deﬁciency. The fourth problem was that
the conjectured relationships were incorrect; this problem showed itself
where signs were opposite of what was expected. More thorough thinking in
regards to how measures relate to the underlying variable might have
prevented some incorrect speciﬁcation. The fourth problem, however, is also
what the study was intended to test.

In addition to the four problems discussed above, a ﬁfth problem is
that other forces which aﬁ‘ect the FCA method implementation decision were
not incorporated in the model. Literature discussing other events or courses
of action taking place in manufacturing plants suggests some other paths
managers are following which may affect the decision to implement FCA
methods. Consider two examples. First, a plant committed to continuous
improvement may choose to focus on process improvement rather than costs,
assuming that costs would eventually come under control. This is the
philosophy behind statistical process control. Second, a plant embarking on
implementing the theory of constraints focuses simply on reducing total
operating costs rather than on allocating the costs to individual products.
Such a plant would not need an FCA method to ﬁnd total operating costs.

Although this study has had a number of shortcomings, the study still
has produced results supporting three of the ﬁve hypotheses. For the other

two hypotheses, relationships between at least some measures and FCA

1 l9
methods implementation were established, but the hypotheses were not

supported due to inconsistent signs.

CHAPTER VI
CONCLUSION

This chapter addresses the contributions of this research and discusses
the limitations of the research and analysis. Finally, it discusses possible

directions for further research.

CONTRIBUTIONS

The literature has suggested that ﬁner cost allocation methods are
appropriate in situations where the resource consumption patterns are
diverse and uncorrelated with the allocation base (referred to in this study as
complexity). Previous to this study, no study had empirically tested whether
FCA method implementations actually support this suggestion. As a result
of survey analysis, this study has shown a signiﬁcant positive relationship
between complexity and FCA method implementation. In addition, a reliable
item has been developed for measuring complexity in an FCA methods
implementation setting. This item considers product diversity, routings

diversity, and extent of setups in assessing complexity.

120

121
The literature also suggests that project implementations are more

likely to occur when managerial support is present and implementation
obstacles are missing. N 0 previous studies have related these suggestions to
the implementation of FCA methods. This study has provided support for
these two suggestions in regards to implementing FCA methods. As a result
of survey analysis, a positive signiﬁcant relationship (at the 90% conﬁdence
level) was found between managerial support and FCA methods
implementation. In addition, a reliable item has been developed for
measuring managerial support in an FCA methods implementation setting.
This item considers support by the controller, support by other top
management (excluding the CEO), and the level in the organization of the
champion for FCA methods in assessing managerial support. As a result of
survey analysis, a negative signiﬁcant relationship was found between
implementation obstacles and FCA methods implementation. In addition, a
reliable item has been developed for measuring implementation obstacles in
an FCA methods implementation setting. This item considers lack of a
sponsor; the difﬁculty of assessing costs and beneﬁts; and the costs of
planning, changing the information system, training the users, and collecting
and processing additional data in assessing implementation obstacles.

The results of this study, therefore, support the movement occurring in
the manufacturing community toward implementing ﬁner cost allocations in

situations where products and processes become complex and differ in their

122

consumption of resources and where managerial support is present and

implementation obstacles are missing.

LIMITATIONS

The results of this study are useful as an informative study but have
two limitations. First, when considering only the constructs developed in the
study, only 6% of the variation is explained. By considering additional
measures, however, 27 % of the variation is explained. The low explanatory
power exhibited indicates that other factors not included in this study exert
great inﬂuence in affecting which FCA methods are implemented. Thus, the
signiﬁcant relationships found in this study relate to only a portion of the
decision-making inﬂuences. Second, the small sample size raises the
question of whether the results are speciﬁc to the sample or are generalizable
to the main population. Even if the results are limited to the sample, the
results may be of interest to managers deciding whether or not to implement

FCA methods.

DIRECTIONS FOR FUTURE RESEARCH

This study has limited explanatory power. Two other inﬂuences

(focusing on process rather than cost and tracking only total operating costs

instead of reﬁning product cost allocations) were suggested during the

123
interpretation of results. Future research could address these two inﬂuences

and incorporate the appropriate measures into a common model.

In the survey, data were obtained which assessed the plants’
performance relative to that of their competitors. The relationship of
performance to FCA method implemented could be examined. Similarly,
data were obtained about the importance of cost allocations in making
various decisions. The relationship of cost usage to FCA methods
implementation could also be examined.

In this study, some measures did not show a signiﬁcant relationship to
FCA methods implementation using regression analysis but some
relationship may exist. To explore this possibility, LOGIT techniques could
be used to group responses by characteristics, then to ﬁnd whether the
distinction helps improve the ability to predict implementation of FCA
methods. For example, each observation could be classed as a high or low
FCA methods implementor; then the characterization could be tested.

In conclusion, future research could focus on methods to add missing
inﬂuences or to ﬁlter out the noise which presently obscures the relationship

between speciﬁed plant/ﬁrm characteristics and the use of FCA methods.

APPENDIX

SURVEY INSTRUMENT

Michigan State University
Firm Characteristics and Allocation Practices Study

Baslclnetructlons

Thenkyouforperticipatinglnthisproject. Thereseerdiersaremminingmcertamﬂrmclmisms are
assccldedmmtheﬂnn'scholceofoverheadellocetionmsﬂiods.You indicateycurvclmtaryagreementto

particlpstebyccmpletingandremrnlngthisquesticmaire.

Yourperliclpdicnlsinwtanttcthesucceesctthissmdy. Toshmvurappnciaticn.h~creepcndsrits’mnsswlll
umummmeﬂsmdnckMﬂchmybemednduhdunhubmmm
merriberstchnch. Pleasinhdeycummeenthesurveyhcaeeycuaredrawn. Yourmmewllnotbereleaesdtc
ctherphntscrtlrms.

Pleeeetlleuycugmmbttnbeetcimabﬂly.

mmmmmmmm. Oniyeggregatediawlbereleeesdtcthepwlic.
manhunmeﬂrmwlllNOThemedeeveﬂabletcmﬂnn. www.mmnmm

mmwuwamwmwuumm;mammmmmam
ftm'snspuﬁigplmbumessaphntindcatesldoesnctmwbemned.

mmmmmmmmeWMmmm
recelvhgll.

missuveylsbelngmeﬂsdon .)

ltycuweudiueaccpyctthe MMWM.MMmmmc-Uvﬂchm
ycudeei'eJhendiachtliecudtctheMdthesuvey.

mdeWbmehmﬂnwlmmum.

ltycuheveanqustioneccncerningthisprclect.pleeseccnlact
ananGJouland (611)363-8642
ltMr.Judenlsunvelhbh.thencmtactanycwieiolouhgd(517)$5-7a6:

Dr.SuemF.l-laka
Dr.chddMarshall
Dr.FredricH.Jacobs

Plseeerehirntheccrnpletedquesticrvielreto:

WlnG.Jordan
DepemnsntofAccwnting
EllBrcedGradusteSchoolofMuugement
MichiganSteieUnlvu'slty
N270NorthBusineechmpleu
EestLansing.Michlwn 6824-1121

124

125

MIchlganStateUnlverslty
FlunCharacterisﬂcsandAlocetlonPracticesShndy

Telusabcuyourself. (Yurnum.000,mdtdephaummbuwilbehsptccnﬂdaiid,emwﬁhyurﬂrm.) The
WWWbWhmhmﬂnﬂdeuﬂWhmmm
mmwcmldbeWMsbneiconvenﬂon.

 

 

 

 

 

Ycuneme: Yourtitte:
Yourplant Yourtelephanemrnber.
MMMMMWM:
Yearswlhthisccmpenr— Yearslnmlslndustm
Yurprhnryheclvemdcrareactupertise:
D Accommg 0 Finance
0 WW 0 W
0 EM 0 W
D Other.pleeseepecity

 

mumbmmmmmmmmmmnm
WMMWhWwaMsMW. Aplantiscnecrmcrehtﬂdlngedaslte
whichjotﬂyprcducecneormoreprcductlnes. Ckclethecorrectreepcnee.

M w
Ayes m Obey.
TheiMybontymcaW ............... 1 2 3 4 5 6 7
Ournﬂcrcanpdlcrsarenotputicdartysggreeelve ....... 1 2 3 4 5 6 7
Miamigmyﬂrmtcsevsreiyreducepnces
endlerprcllmerunsmlsprcducts .................. 1 2 3 4 5 6 7

Myfsrnlscoenctﬂengedhlhenuhstpiece
wmmmummw

mmﬂy.liodailly.technologicaledventage.orprice....1 2 3 4 5 6 7
mumboilsnchdsngedhthenuhstplace
bydcrmeticcompstitorswlhedlfferentisledverngebeesd
onquallty.1lodbllliy,technclcdceledvantage.orprice....1 2 3 4 5 6 7

thﬂWcﬁdeJMMMMMMamdd-dm
Mheechproﬂctlh-cyclestage. (Seemuiesttumotthemstnsm)

1mm sup: '5 Maturity Stage: 3
GM Stage: 1 Decine Stage: ‘5

 

 

126

mmmmmmmwmmwmmmbwumm
tempt-It. Circletl'ieccrrectresponee.

w 860:1eg

Ayes Natural Disease
Myﬂnnenrcebetnswwcductbnovatlon ................ 1 2 3 4 5 6 7
Myﬂnnoicelsatnewprcceesimovwon ................ 1 2 3 4 5 6 7
wmmmmm . ...... 1 2 3 4 5 6 7
Mytimselsprkwtlyhlgh-proﬂ-marginproducts .......... 1 2 3 4 5 6 7
Myﬂrmprnducesmdlfier'entlated(cornrnodty)pmducts ..... 1 2 3 4 5 6 7
Myﬂrmismhdushylav—ccetlseder ................... 1 2 3 4 5 8 7
Myllrmbeeesprlcesmlymcceu .................. 1 2 3 4 5 6 7
whammmmmmm .. 1 2 3 4 5 6 7
Myflnnisheeviylrmhedlnccetreductloneﬂorts ........ 1 2 3 4 5 6 7
Myttmfocmeeetrorvyonmeervice ............ 1 2 3 4 5 6 7

Mytlnndmthepluncselectlscwndoceticnnm..1 2 3 4 5 6 7

Immmmmmwmmmmmmmam
mdcoethgmtlnn’sproductswlthycuplarl. Ckcletheccrrectreepmee.

w W
MYcurlet: Ayes Natal Discuss
ProductlmsareNOTverydlverse ..................... 1 2 3 4 5 6 7
Preductlhesarebulltceesenthiythemrcwng ....... 1 2 3 4 5 6 7
EWisdedcdedtceslnglepmductllne ............ 1 2 3 4 5 6 7
Alldirectldiorlsccsteddthesamestumm . ....... 1 2 3 4 5 6 7
Setipstcctu'igehetweenpmmctsmecnlymtine ..1 2 3 4 5 6 7
Theplmtisajcbahcp .............................. 1 2 3 4 5 6 7
Thereiectnewproductkncvlionlsveryelcw ............ 1 2 3 4 5 6 7
Theluwiothmliecychhasmw ..1 2 3 4 5 6 7
mmmmm .......... .1 2 3 4 5 6 7

EMMWanthu‘mue-sz
mmmmmmmmmmmmmmm ......... s
Thetotelmanuiectulng ovsanuomnmramm ...............
“remaprcducthes .............................................

 

 

‘I'henunbsroiroutlngsmed .............................................
Themnberoienglneerlngchangeordsrspsryeer ............................
mmammm(mmmmmm)mmmm

Formtyplcdpmduct.wmmduredoycuwnkycutkmhdds7 .......... 5

Whidoywﬂhkbﬂnhrgsstdomsehcmuhstdmhddbymﬂmferapmduct
manufacundhyourplerl? ............................................. i

 

 

 

10.

11.

127

MmWsWWbmWfaﬂnMMhmmmm
followingdimsnsions. Circletheappropriatereepcnee.

Much About Much

Greater the” Less
Overelpreductquelly ........................................ 1 2 3 4 5 6 7
Above-everagecostreducticnscverthelastttu'eeyears .............. 1 2 3 4 5 6 7
Siesgmlhintheleﬂtineyears ............................. 1 2 3 4 5 6 7
mmmmmwmm ....................... 1 2 3 4 5 6 7
GrewlhinReturnonAasds(ROA)hthelastttneyeus ............ 1 2 3 4 5 6 7
Overﬂcornpetlivepeeltionmey .............................. 1 2 3 4 5 6 7
Overallcustcrnereervice ...................................... 1 2 3 4 5 6 7

thMdethmmmmWMnha
WWuavay-baeedccdhg)hyurtkm.crdehmmm.

Don't Stung ﬂung

Know Mice mm W
CEOsmpcrt ...................... 0 1 2 3 4 5 6 7
Complerswport .................. 0 1 2 3 4 5 6 7
Othsrtopmanagementsuppcrt ........ 0 1 2 3 4 5 6 7
Mddlerrismgsment ........... 0 1 2 3 4 5 6 7
Accountingeuppcrt ................... 0 1 2 3 4 5 6 7
Markethgswpcrt ................... 0 1 2 3 4 5 6 7

thWMthiﬂﬂnWWthcﬁdw-ﬂm
Whyeulim.0heckthe¢prcprietereepcnee.

0000000
3
8

Immwamdmmhmmmmmmmm
mammmmm.mmm response.

Don't Very Very
Know EM" W W
Clumpion‘s ell‘ectlvenees O 7 6 5 4 3 2 1

lndicetebelouﬂnknpatanceciﬂurdeccstdlcceﬂanphyhﬂnspedﬂedm. Circletheqiprcprﬂe

response
Little Moderate Extreme
W W W
Financielrepcrtlng .................................. 1 2 3 4 5 6 7
Costreductlondecisions ............................ 1 2 3 4 5 6 7
Continuouslrnpmvement ............................. 1 2 3 4 5 6 7
Productmbrdecisions ............................... 1 2 3 4 5 6 7
......................................... 1 2 3 4 5 6 7
Reecucedocliondecisicns ......................... 1 2 3 4 5 6 7

12.

“mmmmmmdmmmmm.mmm
thmmummmm? Pleesecirclethempreprm
response.
He Moderate Strung

Obshcle Obstacle Obshcle
Whimsistancetomigthechange .............. 1 2 3 4 5 6 7
Lackciahlgh-levelspcnsor .......................... 1 2 3 4 5 6 7
Focuscnshort-terrnfinsnclslmessues .................. 1 2 3 4 5 6 7
Quiltyingcostsendheneﬂtsotthechenge ............. 2 3 4 5 6 7
Cormmnicsthgresulsciplenningtlucughcutthecrganwcn1 2 3 4 5 6 7
Costotadequaephnnlng ............................ 2 3 4 5 6 7
Costolbrhginghconsim ......................... 1 2 3 4 5 6 7
Costotchmghgthehicrmdimsystem ................. 1 2 3 4 5 6 7
Costcftraining .................................... 1 2 3 4 5 6 7
Coddnrﬂuadrﬂcostsystemfcrscmepenodotm...1 2 3 4 5 6 7
CaddMNMigmdsteilsddﬂa ........ 1 2 3 4 5 6 7

mmdﬂummdmeMmmhﬁnWthaﬂdm

timssunlasthepmcsesbywilchthsywerew MWdehW
qusdlcns. mmwmmmunsdilswm

”WWW M‘WMWMWIWM

driver. LEAST DETAILED mdhod.

DepertmerialchPlsntwlde Alccstssalplmtmfachnigcostsbyushgeltherasim

Alocdlchsthcds mperdepsrmcrmwlwldem-s. MODERATELY
DETAILEDmethed.

Activly-chmcese—BasedAlocstlm Assignscodstcindvlduelactlvltiesorpreceseeshthertim

Mdhod Wittnntncescoshtomersctlheectlvlies. MOST

13.

14.

DETAILED mdhod.

Funchdﬂnﬂmtypuddecsﬁonhdinimnsmmnmﬁndmmmwm2)the
thﬁchlmW.m3)MMdesmmmmwmh
ereeslnwhichllsimplemerned. (Advenemcfmeusntmeyusemmenmetechmmopercenugssdc
nothevstoewdioox.)

hum—W—

 

 

Yes No Year deIgOcsts
SinglePlsntwldeAlccstionMsthod .................... 0 Ci ‘5
WaWWAm-uonm.... 0 0 is
Activity-«WWW ............ D D 96

 

lndicdethenilmerofhdnctcostpools medhywrplantunderthevaricuselccdionmethcds. (Slt'pthe
methodltlisncthiplementedmircle'rlnormorepoolssleused.)

OnlyTwo Sevsnwlcn
OostPoois OeuPeob
WorMpIePlantwide 2 3 4 5 6 7

Actlvly-orPrccess-Based 234567

15.

16.

17.

16.

19.

129

Wmmmsmwmm,mmmsmawmdm
deteiedcostelocdiormiethodselongthecontinuumbelow. Clrcietheappropriﬂereeponse.

Only Only
shyl- Imus WV
Plentwlde «Diner-n! Deﬁned
Allocation Aleutian Methods
Methods Methods (1e, ABC)
W W

.0—
-b—
5r

l
I

1 1 1
l l I
1 2 3 4 5 6 7 8 9 10 11

Domopummheepmmmnuidbmﬂanmmm?
D Yes D No

Immmmmmmamwmmuamu.
Pieeeeprovldethemproprinereeponeee. (Sltbthemdhodiilwesneverwa

 

1W“ tar-mu»:
YOU YOU W
W W Yes No
SinglePlsntwldeAlocetionMethod D 0
WerWtiplePlerﬂwlde
................. D D
Acuvlty-orProoeee-Bued
AlocdlonMdhod ................. D D

 

lndcﬂthdocdonWMmpﬂipr-liigbmﬁemhhm”. Pleeeecheck

theepproprideleeponee.
Drop Nearenge W Exp-nuke
SkyePlemwldeAlocetionMethod ......... D Ci 0 CI
DepemnenteerPhntwide
Method ...................... D D D D
Actley-orProcees-BeeedAlocdlon
Method .............................. D D D D

nmmmmmm.mmmnmmemmm7 (Agenyou’inputwill
NOTbereleeeedtootherfinns.)

D Yes. Otherplu'sshmylirmmeyreceivemyhpm
D No. Donoteheremymwlthmyotherpluisorm.

Myoulorywttm.energy,endW1ees. Yuaeeistancelsge‘lyeppreclded. We!

mneponeeswllbeprotected.

130
Warrants
m
Cost-Reduction EM-Stmctundeﬂatdeelgnedtolduﬂtymdndmemotuceesmcm.
CoetCuler—AnWmﬁorproductiorMichooetsmcolected.

CoetPool—Apooloimoney(especisllyindirectcoets) icrmedfrcmcoetevmlcherevlewedaseherhgacommonooet

(river. ThuuuyhecodmdMHMWMwmcodoﬁech(umam)wmmwubmchm
consurnethecoetaiver.

chm—Ammmowuammmwmmwmmmuw. The
mmmmmmummmawmmm.mmmmmwmm.
spunormsweﬂesgemratingsuppodanmgoﬂnrmembusofﬂnagenlzaﬂon.

mmMW—WMMWMWMMWMW.

m-mmeMNMNeMMbW;M,MWdhmp-w
reeponsihleiortheplent’slndustry.

W

mmsm-mmummammhhmmMumsmm.
mmbmmwwmmmuuamummmuuubmmmm;
mmmmeoemmmumemmhmwmmmmum
prontotions.

mm—mmmummaammdm‘mmmm. Thlepheeels
eleoephomhedbm mm;mdmm;mmmmamdu
scebuoduchmlhu;huvymmudmmhdwhgovuﬁm;meddngmwnnddsbﬂnmh.

MehsrltyStage—Occuswheneelesvohmcontinueetoincreeee.Misdeaeesingmand,m-Iy,levehoﬂor
declinessllgﬂly. Uriteabsmeyﬂuchieiemnmenngeotphnormonepucemm. Thisstagelseleo
WWW mmnmamuwmwmdmwmmmm

Worprooeseeshneedofheevyrepelncoet-pncem; Wotnewmmunewm
moddsendsizesbmdepechledesinducumntsacmceeebnstocim

mm—mmmmmmummmaemmmmmw. Otherm
«mmmmmmsmomaummwmmmmwmmmumwhm
mMm;WWbM;R&DWbM;MWWBeﬂd

LIST OF REFERENCES

LIST OF REFERENCES

Anthony, Joseph H. and Ramesh, K. "Association between accounting
performance measures and stock prices: A test of the life cycle
hypothesis," Journal of Accounting and Economics, Vol. 15, 1992,
pp.203-227.

Anderson, Lane K. and Sollenberger, Harold M. Managerial Accounting,
Eighth Edition, South-Western Publishing, Cincinnati, 1992, Chapters
3, 5, and 1 1.

Banker, Rajiv D. and Potter, Gordon. "Economic Implications of Single Cost
Driver Systems," Journal of Management Accounting Research, Vol. 5,
Fall 1993, pp. 15-31.

Beath, Cynthia Mathis. “Supporting the Information Technology Champion,”
MIS Quarterly, Vol. 15, No. 3, 1991, pp. 355-372.

Beatty, Carol A. "Implementing Advanced Manufacturing Technologies:
Rules of the Road,” Sloan Management Review, Vol. 33, No. 4,
Summer 1992, pp. 49-60.

Boer, Germain and Jeter, Debra. ”What's New About Modern
Manufacturing? Empirical Evidence on Manufacturing Cost
Changes,” Journal of Management Accounting Research, Vol. 5, Fall
1993, pp. 61-83.

Boston Consulting Group. Perspectives on Experience, Boston Consulting
Group, 1968.

Bromwich, Michael. ”The Case for Strategic Management Accounting: The
Role of Accounting Information for Strategy in Competitive Markets,"
Accounting, Organizations and Society, Vol. 15, No. 1/2, pp. 27-46,
1990.

Campbell, Robert J. “Pricing Strategy in the Automotive Glass Industry,”
Management Accounting, July 1989, pp. 26-31 and 34.

Campi, John P. “It’s Not as Easy as ABC,” Journal of Cost Management, Vol.
6, No. 2, Summer 1992, pp. 5-11.

131

132

Chatﬁeld, Michael. A History of Accounting Thought, Revised Edition,
Robert E. Krieger Publishing Company, Huntington, New York, 197 7,
pp. 104-168.

Cheung, Steven N. S. "On the New Institutional Economics” from Contract
Economics by Werin, Lars and Wijkander, Hans, 1992, Blackwell
Publishers, Oxford, pp. 48-65.

Churchill, Gilbert A., Jr. “A Paradigm for Developing Better Measures of
Marketing Constructs,” Journal of Marketing Research, Vol. XVI,
February 1979, pp. 64-73.

Cokins, Gary, Stratton, Alan, and Helbing, Jack. An ABC Manager '8 Primer,
Institute of Management Accountants, Montvale, New Jersey, 1993.

Compton, Ted R. "Using Activity-based Costing In Your Organization - Part
1,” Journal of Systems Management, Vol. 45, No. 3, March 1994, pp.
32-39.

Cooper, Robin, "Schrader Bellows Cases," #9-186-272, Harvard Business
School, 1986.

Cooper, Robin and Kaplan, Robert S. ”How Cost Accounting Systematically
Distorts Product Costs,” from Accounting & Management: Field Study
Perspectives, ed. by Bruns, William J. Jr. and Kaplan, Robert 8.,
Harvard Business School Press, Boston, 1987.

Cooper, Robin and Kaplan, Robert S. ”Measure Cost Right: Make the Right
Decisions," Harvard Business Review, September-October 1988, pp. 96-
103.

Cooper, Robin and Kaplan, Robert S. "Proﬁt Priorities from Activity-Based
Costing," Harvard Business Review, May-June 1991, pp. 130-135.

Cooper, Robin; Kaplan, Robert S.; Maisel, Lawrence S.; Morrisey, Eileen; and
Oehm, Ronald M. Implementing Activity-Based Cost Management:
Moving from Analysis to Action, 1992, Institute of Management
Accountants, Montvale, New Jersey.

Corporate Technology Information Services. Corp Tech Directory of
Technology Companies, 1995.

Darnay, Arsen J. (Ed.) Manufacturing USA: Industry Analyses, Statistics,
and Leading Companies, Fourth Edition, Gale Research, Detroit, 1994.

133

Datar, Srikant and Gupta, Mahendra. “Aggregation, Speciﬁcation and
Measurement Errors in Product Costing,” The Accounting Review, Vol.
69, No. 4, October 1994, pp. 567-591.

Day, Diana L. "Raising Radicals: Diﬁ'erent Processes for Championing
Innovative Corporate Ventures,” Organization Science, Vol. 5, No. 2,
May 1994, pp. 148-172.

Dun and Bradstreet. Dun’s Industrial Guide Guide: The Metalworking
Directory?M 1994/ 95, 1994.

Emore, James R. and Ness, Joseph A "The Slow Pace of Meaningful Change
in Cost Systems," Journal of Cost Management for the Manufacturing
Industry, Vol. 4, No. 4, Winter 1991, pp. 36-45.

Fisher, Anne. “Morale Crisis,” Fortune, Vol. 124, No. 12, Nov. 18, 1991, pp.
70-72, 76, and 80.

Frank, Gary B., Fisher, Steven A. and Wilkie, Allen R. “Linking Cost to
Price and Proﬁt,” Management Accounting, June 1989, pp. 22-26.

Gorsuch, R. L. Factor Analysis, 1974, W. B. Saunders, Philadelphia.

Green, Forrest B. and Amenkhienan, Felix E. "Accounting Innovations: A
Cross-Sectional Survey of Manufacturing Firms," Journal of Cost
Management, Vol. 6, No. 1, Spring 1992, pp. 58-64.

Greenwood, Thomas G. and Reeve, James M. ”Activity-Based Cost
Management for Continuous Improvement: A Process Design
Framework,” Journal of Cost Management for the Manufacturing
Industry, Vol. 5, No. 4, Winter 1992, pp. 22-40.

Gupta, Mahendra. “Heterogeneity Issues in Aggregated Costing Systems,”
Journal of Management Accounting Research, Vol. 5, Fall 1993, pp.
180-2 12.

Haka, Susan F. and Marshall, Ronald. “ABC vs Traditional Costing in Short
and Long-run Planning,” working paper, Michigan State University,
1994.

Hill, Charles W. L. “Diﬂ‘erentiation Versus Low Cost of Diﬁ‘erentiation and
Low Cost: A Contingency Framework,” Journal of Management
Accounting Research, Vol. 3, Fall 1991, pp. 401-412.

134

Homgren, Charles T.; Foster, George; and Datar, Srikant M. Cost
Accounting: A Managerial Emphasis, Eighth Edition, Chapters 5, 12,
14, and 15, 1994, Prentice Hall, Englewood Cliffs, New Jersey.

Hwang, Yuhchang, Evans, John H. III, and Hegde, Vishwanath G. “Product
Cost Bias and Selection of an Allocation Base,” Journal of
Management Accounting Research, Vol. 5, Fall 1993, pp. 213-242.

Jensen, Michael C. and Meckling, William H. "Speciﬁc and General
Knowledge, and Organizational Structure,“ from Contract Economics
by Werin, Lars and Wijkander, Hans, 1992, Blackwell Publishers,
Oxford. 00. 251-274.

Johnson, H. Thomas. “Activity-Based Information: A Blueprint for World-
Class Management Accounting,” Management Accounting, June 1988,
pp. 23-30.

Johnson, H. Thomas and Kaplan, Robert S. Relevance Last: The Rise and
Fall of Management Accounting, 1987, Harvard Business School,
Boston.

Johnson, Richard A. and Wichern, Dean W. Applied Multivariate Statistical
Analysis, Third Edition, 1992, Prentice-Hall, Englewood Cliffs.

Kanter, RM. The Change Masters, 1983, Simon and Schuster, New York.
Kee, Robert. “Integrating Activity-Based Costing With the Theory of

Constraints To Enhance Production-Related Decision-Making,”
Accounting Horizons, December 1995, pp. 48-61.

Kotler, Philip. Marketing Management: Analysis, Planning,
Implementation, & Control, 1991, Prentice-Hall, Englewood Cliﬂ's,
New Jersey.

Lanford, Audri. “Avoiding Pilot System Pitfalls,” Inform, Vol. 7, No. 5, May
1993, pp. 32-33 and 38.

Lawrence, John J. and Yeh, Ryh-song. “The Use of Factor Analysis in
Management Research Sample Size Consideration,” working paper,
University of Idaho, 1996.

Levitt, Theodore. ”Exploit the Product Life Cycle,” Harvard Business Review,
Vol. 43, No. 6, November-December 1965, pp. 81-94.

135

Martinsons, Maris G. ”Cultivating the Champions for Strategic Information
Systems,” Journal of Systems Management, Vol. 44, No. 8, August
1993, pp. 31-34.

Moody's Investors Service, Moody ’3 Industrial Manual, 1995 edition, New
York.

Murray, Alan I. “A Contingency View of Porter’s ‘Generic Strategies’,”
Academy of Management Review, Vol. 13, No. 3, 1988, pp] 390-400.

Noreen, Eric. “Conditions Under Which Activity-Based Cost Systems Provide
Relevant Costs,” Journal of Management Accounting Research, Vol. 3,
Fall 1991, pp. 159-168.

Nunnally, Jum C. Psychometric Theory, Second Edition, McGraw-Hill, New
York, 1978.

Ostrenga, Michael R. "Activities: The Focal Point of Total Cost
Management," Management Accounting, Vol. 71, No. 8, February 1990,
pp. 42-49.

Pattell, James M., "Adapting a Cost Accounting System to Just-in-Time
Manufacturing: The Hewlett-Packard Personal Oﬁce Computer
Division,” from Accounting & Management: Field Study Perspectives,
ed. by Bruns, William J. Jr. and Kaplan, Robert 8., Harvard Business
School Press, Boston, 1987.

Porter, Michael E. Competitive Strategy: Techniques for Analyzing
Industries and Competitors, The Free Press, New York, NY, 1980.

Previts, Gary John, and Merino, Barbara Dubis. A History of Accounting in
America: An Historical Interpretation of the Cultural Signiﬁcance of
Accounting, John Wiley 8; Sons, New York, 1979, pp. 187 and 278.

Reich, B.H. and Benbasat, I. "An Empirical Investigation of Factors
Inﬂuencing the Success of Customer-Oriented Strategic Systems,”
Information Systems Research, Vol. 1, No. 3, September 1990, pp. 325-
347.

Richardson, Peter R. and Gordon, John R. M. ”Measuring Total
Manufacturing Performance," Sloan Management Review, Vol. 2 1,
Winter 1980, pp. 47-58.

136

Scherer, Frederic M. and Ross, David. "The Welfare Economics of
Competition and Monopoly" and "Industry Structure," Industrial
Market Structure and Economic Performance, Third Edition, 1990.

Shank, John K. and Govindarajan, Vijay. ”The Perils of Cost Allocation
Based on Production Volumes,” Accounting Horizons, December 1988,
pp. 71-79.

Shields, Michael D. and Young, S. Mark. "Effective Long-Term Cost
Reduction: A Strategic Perspective," Journal of Cost Management,
Vol. 6, No. 1, Spring 1992, pp. 16-30.

Smith, Keith V. and Leksan, Mark P. “A Manufacturing Case Study on
Activity-Based Costing,” Journal of Cost Management for the
Manufacturing Industry, Vol. 5, No. 2, Summer 1991, pp. 45-54.

US. Department of Commerce, Bureau of the Census, 1987 Census of
Manufacturing: Concentration Ratios in Manufacturing, 1992.

Varian, Hal R. Intermediate Microeconomics: A Modern Approach, Second
Edition, W.W. Norton & Company, New York, 1990.

Varian, Hal R. Microeconomic Analysis, Third Edition, W.W. Norton &
Company, New York, 1992.

Wernerfelt, Birger. “The Dynamics of Prices and Market Shares Over the
Product Life Cycle,” Management Science, Vol. 31, No. 8, August 1985,
pp. 928-939.

White, Halbert. “A Heteroskedasticity-Consistent Covariance Matrix
Estimator and a Direct Test for Heteroskedasticity,” Econometrica,
Vol. 48, No. 4, May 1980, pp. 817-838.

Yellen, Richard E. ”Introducing Group Decision Support Software (GDSS) in
an Organization," Journal of Systems Managerment, Vol. 44, No. 10,
October 1993, pp. 6-8.

Yourdon, Edward. Modern Structured Analysis, 1989, Prentice-Hall,
Englewood Cliﬁ’s, New Jersey.

Zimmerman, Jerold L. "The Costs and Beneﬁts of Cost Allocation," The
Accounting Review, Vol. 54, No. 3, July 1979, pp. 504-521.