J. Eng. Technol. Manage.

25 (2008) 287304

Contents lists available at ScienceDirect

J. Eng. Technol. Manage.

journal homepage: www.elsevier.com/locate/jengtecman

Measuring the leanness of manufacturing systemsA case

study of Ford Motor Company and General Motors
M.E. Bayou a,*, A. de Korvin b,1
a
b

School of Management, University of Michigan-Dearborn, Dearborn, MI 48126-2638, United States

Department of Computer and Mathematical Sciences, University of Houston Downtown, Houston, TX 77002, United States

A R T I C L E I N F O

A B S T R A C T

Article history:

In spite of the vast research published on lean manufacturing systems

in several disciplines in the last decade, the concept remains
underdeveloped for two reasons. First, it lacks a generally accepted
denition. Different authors dene lean in terms of its objectives,
which vary, overlap and differ in different rms. Second, no study has
developed a systematic and relative measure of lean production
systems. With the lack of such a measure, two companies cannot be
rated objectively on their progress toward becoming lean. This paper
has two goals: rst, to dene manufacturing leanness as a unifying
concept, and, second, to develop a systematic, long-term measure of
leanness. Manufacturing leanness is a strategy to incur less input to
better achieve the organizations goals through producing better output.
The systematic measure of leanness has seven characteristics:
relative, dynamic, long-term fuzzy logical, objective, integrative
and comprehensive. The leanness measure utilizes the fuzzy-logic
methodology since lean is a matter of degree. Applying the measure
to compare the production leanness of Ford Motor Company and
General Motors, the paper selects Honda Motor Company as the
benchmarking rm. Selecting just-in-time (JIT), Kaizen, and quality
controls as lean attributes, the paper uses surrogates for these
attributes extracted from audited nancial statements over the years
20012003. The results show that Fords system is more than 17%
leaner than GMs system vis-a`-vis the benchmarked companys
system.
2008 Elsevier B.V. All rights reserved.

Available online 12 November 2008

JEL classication:
M41
L62
Keywords:
Lean
Leanness
Fuzzy-logic leanness
Systematic measures
Benchmarking

* Corresponding author. Tel.: +1 313 593 9962; fax: +1 313 271 9837.
E-mail addresses: mbayou@umd.umich.edu (M.E. Bayou), deKorvin@dt.uh.edu (A. de Korvin).
1
Tel.: +1 713 221 8403.
0923-4748/$ see front matter 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.jengtecman.2008.10.003

288

M.E. Bayou, A. de Korvin / J. Eng. Technol. Manage. 25 (2008) 287304

1. Introduction
For decades, manufacturers have sought to optimize operations, supply chains and capital
assets (Pagatheodrou, 2005, p. 13). Facilitated by advances in information technology, the pursuit
of optimization has intensied the demand for speed, exibility, waste elimination, process
control, people utilization and global reach to gain competitive advantages (James-Moore and
Gibbons, 1997; Allway and Corbett, 2002; Pagatheodrou, 2005). Recently, achieving this goal has
become increasingly complex due to the fast moving global market, budget cuts and capacity
down sizing (Pagatheodrou, 2005, p. 12). Hence, lean manufacturing2 has become the key
approach to managing this complexity (Liker, 1998). The Toyota Production System (TPS), the
pioneering approach to manufacturing leanness, has become the basis for much of the
optimization movement that has dominated manufacturing developments since the last decade
(Liker, 1998; Hall, 2004).
In spite of the vast research published on lean manufacturing, the concept of leanness is
underdeveloped for two reasons: (a) it is often loosely dened in terms of its objectives, and (b) it lacks
a holistic, unifying measure. The objectives vary, overlap and differ in their emphasis in different rms,
e.g., on lean production versus lean behavior (Emiliani, 2000). Several studies have dened a portfolio
of tools or techniques to implement lean manufacturing (Rother, 1998; Hines and Taylor, 2000;
Emiliani, 2000; Sullivan et al., 2002). These studies, however, fall short of developing a systematic
measure of manufacturing leanness. For example, these studies lack the mechanism by which they can
determine whether Ford Motor Companys production system is leaner than General Motorss
production system, and by how much.
Manufacturing leanness is a concept that unies the various practices of promoting lean. Since
these practices measure different objects, e.g., inventory size, quality defects, Kaizen and asset
reduction (Emiliani, 2000), management needs a unifying measure of the effects of these practices, a
measure that meaningfully integrates the changes in these practices into one indicator, a scalar.
Accordingly, using this leanness scalar, two competing companies could be compared to determine
the relative degrees of their manufacturing leanness. A systematic measure of this sort can help
management and other stakeholders
(a) simplify the task of lean assessment by collapsing the effect of several associative practices that
seek waste reduction and output improvement,
(b) identify and analyze the effect of cost cutting programs,
(c) add a strategic long-term evaluative tool to the short-term nancial indicators for the assessment
of corporate performance.
The proposed measure of manufacturing leanness in this paper can help achieve these benets
since it formulates leanness as (1) a fuzzy concept because leanness is a matter of degree, e.g., a rm
can become slightly lean, moderately lean or highly lean, (2) relative across time for the same entity,
(3) relative across rms in an industry, (4) integrative of the multitude of different practices used to
achieve leanness, (5) comprehensive by associating resources used in production to the organization
goals, (6) long-term for studying the lean effect over a number of years and (7) objective since it uses
audited data.
The research problem of this paper is succinctly stated as follows: developing and applying a
systematic measure of manufacturing leanness. The balance of the paper is organized as follows. The
second section denes and explains the basic concept of leanness. The third section describes
the algorithm of developing the leanness measure. The fourth section describes the variables, data
and the required computations for measuring manufacturing leanness of American automakers
versus their Japanese counterparts. The limitations of the study and suggestions for future
research are presented in the fth section. Finally, a summary and conclusions are presented in the
last section.
2
Lean manufacturing is also called lean production and lean thinking. Automotive and aerospace industries were the rst
adopters of lean thinking. Now, its application has spread into other industries (Womack and Jones, 1996).

M.E. Bayou, A. de Korvin / J. Eng. Technol. Manage. 25 (2008) 287304

289

2. What is manufacturing leanness?

Many research studies in different disciplines dene lean in terms of the benets of a
lean manufacturing system (Liker, 1998; Allway and Corbett, 2002; Goldsby and Martichenko,
2004). The basic form of these denitions can be stated as follows: To be lean is to cut fat.
Emiliani (2000, p. 64) explains that the opposite of lean behavior is called fat behavior,
which is dened as any activity or action that creates or perpetuates behavioral waste. These
denitions are a tautology; they do not add to our knowledge.3 Creese (2000, p. CSC.05.1) denes
lean differently as a manufacturing philosophy to shorten lead times and reduce costs by
redirecting waste and improving employee performance, skills and satisfaction. This denition
emphasizes one side, the input dimension, of leanness. Womack et al. (1991) adopt the following
denition (p. 13), which emphasizes both of the input and output dimensions of manufacturing
lean:
Lean production . . . is lean because it uses less of everything compared with mass
productionhalf the human effort in factory, half the manufacturing space, half the investment
in tools, half the engineering hours to develop a new product in half time. Also, it requires
keeping far less than half the needed inventory on site, results in many fewer defects, and
produces a greater and ever growing variety of products.
This denition covers both the efciency and effectiveness attributes of manufacturing
performance, where efciency represents the relationship between input and output and
effectiveness represents the relationship between output and the organizations goals (Drucker,
1967; Baudin, 2004). Fig. 1 shows these relationships. A rm can achieve different levels of
leanness by improving its efciency, effectiveness or both. Moreover, this denition asserts that
efciency and effectiveness are not absolute objectives, i.e., one must continually improve these
objectives in order to improve from lean to leaner and leanness, as shown in Fig. 1. Thus, when
cost cutting improves efciency at the expense of effectiveness, the result is more of shortsighted
shrinkage than strategic leanness. Therefore, we improve on Womack et al.s (1991) denition as
follows: manufacturing leanness is a strategy to incur less input to better achieve the organizations
goals through producing better output, where input refers to the physical quantity of resources used
and their costs, and output refers to the quality and quantity of the products sold and the
corresponding customer services.
To demonstrate the importance of considering efciency and effectiveness individually and in
combination, let us consider these facts pertaining to lean companies as compared to traditional nonlean ones (Allway and Corbett, 2002, p. 46):
 The efciency dimension of leanness: Lean companies record 3550% improvement to labor
productivity, need approximately half the product-development time (concept to start production),
lower reject rates to one-tenth to one-hundredth of previous levels and streamline system
throughput time by 8090%.
 The effectiveness dimension of leanness: Lean companies have 1520 times the annual sales growth
and total return to stockholders.

Fig. 1 shows leanness as a dynamic variable improving from lean to leaner to leanest. The leanest
level is the state of perfection when the manufacturer incurs the least cost, e.g., zero inventory, zero
defects and endless product variety. As Womack et al. (1991, p. 14) explain: Of course, no lean
producer has ever reached this promised land and perhaps none ever will, but the endless quest for
perfection continues to generate surprising twist.
3
In the statement: lean is lack of fat, the predicate lack of fat is already contained in the subject lean. The denition is
analytic as in the denition: a rose is a rose. To add to our knowledge, the denition of lean should go beyond lack-of-fatness.
That is, the denition should be synthetic. This paper proposes a synthetic denition of manufacturing lean, where the predicate
adds knowledge beyond that conveyed by the subject alone.

290

M.E. Bayou, A. de Korvin / J. Eng. Technol. Manage. 25 (2008) 287304

Fig. 1. Denition of leanness: Continuous improvement of the combined efciencyeffectiveness attributes.

Companies follow different practices to become lean. Table 1 classies several of these practices
into three categories according to the frequency of coverage in industrial and management literatures
from 1977s to 1999: extensive, semi-extensive and light coverage. The practices shown in Table 1
emerge from a literature review reported by Shah and Ward (2003).
Before presenting the steps of developing the leanness measure, let us examine its major
characteristics.

Table 1
Lean practices and their extensive coverage in the industrial and management literature (19771999).
Extensive coverage






















JIT/continuous ow production
Pull system/Kanban
Quick-changeover techniques
Lot-size reductions
Continuous-improvement programs
Cross-functional work force
Preventative maintenance
Total quality management
Self-direct work teams
Cellular manufacturing
Focused-factory production
Cycle-time reduction
Process-capability measurements
New-process equipment
Safety-improvement programs
Bottleneck removal (production smoothing)
Quality-management programs
Re-engineered production process
Competitive benchmarking
Maintenance optimization
Planning and scheduling strategies

Source: Adapted from Shah and Ward (2003, p. 131).

Semi-extensive

Light coverage























M.E. Bayou, A. de Korvin / J. Eng. Technol. Manage. 25 (2008) 287304

291

3. Characteristics of the proposed leanness measure

The proposed leanness measure has the following characteristics:
(1) Relative and dynamic. Since leanness is a matter of degree, e.g., a manufacturing system can be
described as lean, leaner or leanest (Fig. 1), the measure is dynamic to encourage the search for
more and better means to improve leanness. Toyota considers leanness a process, a journey, not
an end state (Liker, 1998, p. 8).
(2) World-class-based. Developing this measure of leanness requires selecting a world-class producer
in the same industry as that of the manufacturer whose leanness is under evaluation. The practices
of this leader become the bases for the comparison.
(3) Long-term. The leanness measure covers 3 years of practices.
(4) Fuzzy-logical. To account for the relative nature of leanness, the leanness measure uses fuzzy-set
theory. Thus, the judgment large inventory can have several grades of large extending from 0.0
to 1.0:
Not large at all: (0.0); somewhat large: (0.5); . . .very large: (1.0)

(5) Objective. The sources of data used in the fuzzy-logic formulations are public record: audited
nancial statements.
(6) Integrative. The leanness measure combines all of the fuzzy values of selected practices to become
lean and collapses them into a single score.
(7) Policy management and implementation. The leanness measure applies to the manufacturing
production system of the entire corporation rather than to its parts (segments, subsidiaries or
departments). That is, the leanness measure is a micro-reductive concept in that senior
management sets the overall goal for leanness (top-down) and provides the incentives for bottomup implementation. This is consistent with Toyotas TPS centralized philosophy where a
corporation rst decides how much weight (capacity) it must have in order to be lean (Monden,
1993). Such efciency and effectiveness practices as just-in-time (JIT), Kaizen and TQM affect the
total entity in its struggle to become lean. To be effective, these practices must be performed
continuously, which TPS calls policy management, on a yearly basis, and PDCA (plan, do, check,
action) on a daily basis (Shook, 1998, p. 58). Emiliani (2000, p. 63) explains that the pressure
stockholders and customers exert on senior management to improve the rms nancial position
and operating results directs attention to the lean state of the rm.
3.1. The leanness measures algorithm
The algorithm consists of the following steps:
Step 1: Select the rm whose manufacturing system is to be evaluated for leanness.
Step 2: Using benchmarking and other techniques, select the best (world-class) producer in the
industry of the rm(s) in Step (1).
Step 3: Select a set of relevant practices to become lean, e.g., JIT, Kaizen and TQM (Table 1).
Step 4: Use proper surrogates for practices in Step (3). For example, to measure JIT practices, one may use
the change in the ratios of inventory to total current assets. This is necessary because JIT, Kaizen
and TQM are philosophies rather than concrete objects (Emiliani, 2000). Thus use of surrogates in
leanness analysis is common in practice. For example, Ford uses rst time through capability as
a surrogate for quality improvement and continuous improvement toward aligned objectives
as a surrogate for assessing effective work groups (Liker, 1998, p. 23).
Step 5: Extract data on the surrogates in Step (4) from the consolidated nancial statements of the
manufacturer under study and the world-class producer. From the extracted data, develop

292

M.E. Bayou, A. de Korvin / J. Eng. Technol. Manage. 25 (2008) 287304

Step 6:
Step 7:

Step 8:
Step 9:

ratios for measuring the surrogates. (Using ratios is necessary for performing valid
comparisons of companies of different sizes.)
From the world-class producers data, determine the continuum of the mean and the
continuum of the standard deviation (STDEV) for each practice in the practice set of Step (4).
For each practice, locate the manufacturers mean on the mean continuum determined in Step
(6). Repeat this step for the standard deviation. This process in Steps (6) and (7) is called
fuzzication of the mean and standard-deviation values of the lean practices.
For each practice, combine the fuzzy mean and fuzzy standard deviation using the common
fuzzy intersection (minimum) procedure.
Normalize the sum of values in Step (8). This normalized sum is the measure of leanness of the
manufacturer relative to that of the world-class producer.

This algorithm is applied in the following section to measure the leanness of the American
automakers manufacturing systems relative to that of the Japanese counterparts.
4. Applying the leanness measure to the automobile industry
Before applying the leanness measure using a fuzzy-logic methodology, we provide the following
brief review of the fuzzy-set concepts.
4.1. Fuzzy logic: basic concepts
Fuzzy-set theory accounts for the uncertainty inherited in natural language (Zadeh, 1965). Such
propositions as Company A is leaner than Company B, Company C is very lean and This company is
slightly lean are fuzzy because they are ambiguous and lack crisp boundaries that account for the grades
of the propositions object. A term like leanness we use in natural language is usually a summary of
complex, multifaceted concepts often incapable of precise characterization (Zadeh, 1978, p. 397). In this
paper, we dene leanness as a fuzzy subset. To formulate a fuzzy-logic model, one needs to dene the
universe (U), the elements (xi) of U where U = {x1 + x2 + . . .xn} and the fuzzy subset A included in U, where


x
A
mA xjx 2 U
The membership function for the fuzzy subset A, is usually expressed as follows:
mA: U ! [0,1], which associates with each element of x 2 U, the membership degree mx of x in A:
mA(x) = mx.
The most familiar fuzzy-logic operations are intersection, union and complement:
 Intersection of two fuzzy subsets A and B: mA \ mB = minimum {mA(x), mB(x)}
 Union of two fuzzy subsets A and B: mA [ mB = maximum {mA(x), mB(x)}
 Complement of A: mA0 x 1  mA x:
For example, assume that the just-in-time inventory system consists of the following policy:
 Have raw material (X1) delivered at the time only when the factory oor needs it.
 Produce nished goods (X2) only when customers place orders.
The fuzzy subset J (that the company follows a JIT policy where JIT is one component of leanness) is as follows:


0:8 0:5
JIT : J

X1
X2
Similarly, assume that Kaizen, as an element of leanness, refers to improvement of carrying cost of
raw materials and nished goods inventories. Let Y1 and Y2 denote carrying cost improvement in these
inventories, respectively. Then,


0:1 0:8

Kaizen : V
X1
X2

M.E. Bayou, A. de Korvin / J. Eng. Technol. Manage. 25 (2008) 287304

293

If A denotes an aggregation operation, we can dene

JIT A Vxi ; yi ai A bi
where ai denotes the membership of xi in JIT and bi denotes the membership of yi in V. For example, if
for A we select the min operator, we can state the following judgment about raw materials:
JIT \ Vx1 ; y1 minf0:8; 0:1g 0:1
This means that the truth of the statement: raw materials are delivered at the time only when the
factory oor needs it and the company has improved the carrying cost of its raw materials inventory
has a truth-value of 0.1. That is, on a scale 0.01.0, the 0.1 assessment indicates that the companys
commitment to leanness regarding raw materials management is low. A similar assessment can be
made about nished goods management. Briey, the aggregation operation produces the following
result for nished goods:
JIT \ Vx2 ; y2 minf0:5; 0:8g 0:5
The membership degrees discussed above are not probabilities because the former deals with
ambiguity and the latter concerns randomness. Zebda (1989, p. 20) explains that ambiguity (or
vagueness) differs from randomness. Randomness pertains to uncertainty (in terms of probability)
concerning the occurrence or non-occurrence of an event while ambiguity or fuzziness (a term rst
used by Zadeh (1965)) points to the lack of precision in natural language. Schrader et al. (1993)
lament that most empirical work on technical problem solving do not distinguish explicitly
between uncertainty and ambiguity; therefore, the two concepts are often used as if they are
interchangeable.
4.2. The key fuzzy propositions
To assess the relative level of leanness of Ford Motor Company and General Motors vis-a`-vis
Hondas level of leanness, we use an explanatory database (ED) (Zadeh, 1978, 1988, 1989). An ED
represents a collection of fuzzy relations used to test the constraints imposed upon the meaning of a
proposition expressed in natural language.4 We begin the study of leanness of automotive operations
of Ford and GM by stating the following proposition:
P: Ford Motor Company is much leaner than General Motors over the period 20012003.
The term in italics in this proposition is the focal variable, X, representing a fuzzy subset. The values
of this focal variable are constrained by the proposition. These values are the grades of leanness. The
focal variable has two components, X = (X1, X2) where
X 1 leanness Ford
X 2 leanness GM
which can be described by the fuzzy relation in the ED as follows:
MUCH:LEANER Lean1; Lean2 : m
in which m is the degree to which Lean1 is much leaner than Lean2, and Lean1 and Lean2 are Fords and
GMs relative leanness levels, respectively. Therefore, we have
P ! Lean Ford; LeanGM is MUCH:LEANER
The symbol ! stands for translates to.
In turn, each of the X1 and X2 are constrained by values of eight components of leanness (measured
by ratios developed from nancial indicators, as explained in the following section):
X 1i RATIOS R11 ; R12 ; . . . R18 
X 2i RATIOS R21 ; R22 ; . . . R28 
4
Zadeh (1989, p. 91) argues that a proposition imposes a constraint on its elements by which the meaning is established
through what he calls test-score semantics. He explains: A basic idea underlying test-score semantics is that a proposition in
a natural language may be viewed as a collection of elastic or, equivalently, fuzzy constraints. For example, the proposition Mary
is tall represents an elastic constraint on the height of Mary.

294

M.E. Bayou, A. de Korvin / J. Eng. Technol. Manage. 25 (2008) 287304

where Rii is ratio I of company is leanness. The constraints, Rii, induced by proposition P are explained
as follows.
4.3. Fuzzy leanness model and data
The constraints induced by proposition P are described by the explanatory database as shown in
Fig. 2. This database consists of relations in two-dimensional format where rows are called tuples and
the columns are called domains (Zadeh, 1989). We use the following nancial indicators of leanness,
extracted from the Securities and Exchange Commission (SEC) 10-K of Ford and GM and SEC Form 6-K
of Honda (www.sec.gov/EDGAR) for the years 20012003:
Inventory: Raw materials and work in process (RM&W)
Finished goods (FG)
Combined inventory of RM&G and FG (COMB)
 Current assets (CA)
 Sales
 Cost of goods sold (CGS)

Fig. 2. Panel A: Explanatory database for Ford Motor Company. Panel B: Explanatory database for General Motors Company.
Panel C: Explanatory database for Honda Motor Company.

M.E. Bayou, A. de Korvin / J. Eng. Technol. Manage. 25 (2008) 287304

295

Market share (MS): cars

Trucks
Combined MS (cars and trucks)
To study the extent of leanness of these companies automotive operations, we focus on three
elements over a period of 3 years, 20012003: JIT, Kaizen and quality management. These elements
have received substantial attention as the review of several literatures shows (Table 1). As surrogates
for JIT, we use relative inventory levels; for Kaizen, we use the relative change in inventories over the
3-year period; we also use the long-run change in cost of goods sold percentage of sales (CGS%) in year
2003 relative to 2001; for quality management, we use the relative change in market share of cars,
trucks and the combined products of cars and trucks. We develop the following ratios, which
constitute part of the database (Fig. 2):
JIT:
R1: Ending raw-materials and work-in-process inventory as a ratio of current assets for year
t = RM&WIPt/CAt
R2: Ending nished goods inventory as a ratio of current assets for year t = FGt/CAt

Fig. 2. (Continued )

296

M.E. Bayou, A. de Korvin / J. Eng. Technol. Manage. 25 (2008) 287304

Fig. 2. (Continued ).

R3: Combined inventoryt/CAt for year t

Kaizen:
R4: (EIt  EIt1)/[(CAt + CAt1)/2], where EIt and EIt1 = combined ending inventory balances of
years t and t  1
R5: CGSt/Salest, where CGSt = cost of goods sold for year t
Quality management:
R6: Cars: MSFord/(MSFord + MSGM + MSHonda)
R7: Trucks: MSFord/(MSFord + MSGM + MSHonda)
R8 Combined: MSFord/(MSFord + MSGM + MSHonda)
where MSFord, MSGM and MSHonda = market shares of Ford, GM and Honda, respectively in year t.
Ratios R1R3 are used to assess the extent of JIT application in each of Ford and GM, respectively. We
assume that the lower these ratios, the closer the company is to reaching the ultimate goal of JIT of
inventory minimization. A major goal of JIT is to minimize inventory (Monden, 1993; Imai, 1997; Ahmad
et al., 2003). Since Ratios R4 and R5 are indicators of the presence of inventory improvement cost
efciency in 1 year, we compute these measures over a 3-year period to examine inventory and cost
improvement, i.e., Kaizen in the long run. Ratios R6  R8 provide clues about product quality controls.

M.E. Bayou, A. de Korvin / J. Eng. Technol. Manage. 25 (2008) 287304

297

Generally, the higher the product quality of a company as compared with the product quality of its
competitors, the higher its market share for the product. In his review of the history of research on
quality, Kolesar (1993) states that since higher quality leads to less scrap and rework, it become the basis
of international industrial competition. To sum, from data extracted from audited nancial statements
over a 3-year period, we compute inventory, CGS and MS ratios and use them as surrogates to study the
long-term extent of leanness of Ford and GM as compared with the leanness of the benchmarked rm.
We selected Honda Motor Company as the benchmarked rm for the following reasons.
4.4. The benchmarked rm
The fuzzy-logic application compares Fords and GMs leanness against that of Honda Motor
Company. Therefore, we have two propositions stated as follows:
P1: Honda Motor Company is an appropriate choice for benchmarking.
P2: Ford Motor Companys automotive operations are much leaner than those of General Motors visa`-vis Honda Motor Companys operations.
These propositions are consistent with Zadehs (1978, p. 395) argument that, [t]he truth-value of a
proposition is dened as its compatibility with a reference proposition, so that given two propositions
p and r, one can compute the truth of p relative to r. We use (Hondas system as a benchmark (i.e., a
reference) for comparing the extent of leanness of Fords and GMs operations over the long run, where
the long run is dened as a 3-year period.5 The truth of the rst proposition is logically developed as
follows. Honda is the only major Japanese automobile company whose consolidated nancial
statements are published according to the US generally accepted accounting principles (GAAP). Hence,
comparing Fords and GMs audited nancial information with that of Honda to assess their leanness is
valid. In addition, Hondas return on investment (ROI) of 8.83, 10.90 and 11.16 for the years 2001
2003, respectively exceed those of Ford (0.23, 0.21 and 0.70) and GM (0.45, 1.21 and 1.32) (S&P
Compustat, 2005) and also exceed the ROI industry average (4.70, 4.90 and 3.70) for the same periods
(RMA, 2004, p. 671). Furthermore, during this period, Honda has brought to market several innovative,
fuel efcient vehicles. For example, in 2004, Honda introduced the Honda FCX reportedly considered
as the worlds most advanced fuel cell vehicle and the rst and only fuel cell car to be certied by the
Environmental Protection Agency (EPA) and California Air Resources Board (CARB) for regular daily
uses . . .. Honda also leads the automobile industry in the development of cleaner and more efcient
gasoline-powered vehicles including gaselectric hybrid technology (Lexis Nexis, 2004). In brief,
Hondas concern with lean production systems has started in the early 1960s when it began building
cars and building its BP (best practice, best process and best performance) as the core supplier
development activity (MacDufe and Helper, 1997). The truth of the second proposition is developed
in the following sections.
4.5. Fuzzy components of leanness
The ED of Ford Motor Company is given by Zadeh (1978, 1989):
ED POPULATION COMPANY; RATIOS; YEARS RATIOS VALUE; STDEV
Components uv1  uv4 values are the means of ratios R1R4s values over 3 years, respectively (Fig. 2,
Relation 3):
X Rt
; where t 1 is 2001; t 2 is 2002 and t 3 is 2003
uvi meanRi
3
t
Thus, for Fords rst component, uv1 R2001
R2002
R2003
=3 0:080
1
1
1
5
We use the consolidated nancial statements of Honda Motor Company. This is justied on two grounds. First, Honda does
not separate its automotive operations from its nancial operations as Ford and GM do. Second, Hondas nancial operations
contribute only 3% of revenues to its total revenue according to its recent announcements (Compustat, 2005).

M.E. Bayou, A. de Korvin / J. Eng. Technol. Manage. 25 (2008) 287304

298

Table 2
Fuzzication of the lean practices the components means.
Fuzzy
components

Hondas
uVi

mVi

u1
u2
u3
u4
u5
u6
u7
u8

0.076
0.152
0.228
0.126
0.005
0.083
0.613
0.173

1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000

Beginning of the
continuum, 0.0 (a)

End of the
continuum, 1.0 (b)

Fords
uVi

mVi

GMs
uVi

mVi

0.152 or greater
0.304 or greater
0.456 or greater
0.126 or greater
0.005 or greater
0.083 or less
0.613 or less
0.173 or less

0.076 or less
0.152 or less
0.228 or less
0.063 or less
0.005 or less
0.083 or greater
0.613 or greater
0.173 or greater

0.080
0.134
0.214
0.011
(0.045)
(0.058)
(0.068)
(0.056)

0.947
1.00
1.000
1.000
1.000
0.151
0.444
0.338

0.094
0.135
0.228
(0.009)
0.013
0.038
0.064
0.046

0.763
1.000
1.000
1.000
0.000
0.729
0.552
0.633

Components u5u8s values are the percentage change between years 2001 and 2003:
uv5

R2003
 R2001

5
5
0:045
2001
R5

Similar computations for uv6 , uv7 and uv8 yield 0.058, 0.068 and 0.056, respectively.
We use the standard deviation of the 3-year values of the eight components R1R8 to measure the
uctuations in these ratios. We assume that the lower the STDEV, the stronger the companys
commitment to leanness. Lower standard deviation may also indicate that the company is closer to its
target level of leanness because at that level, uctuations will be minimal. This resembles the nearly at
part of the learning curve when learning reaches its ultimate level. We denote the components STDEV by
uSi , i = 1, 2, . . . 8. Relation 3 of Fig. 2 shows the uS1 uS8 STDEVs. The next step is to fuzzify these components.
4.5.1. Fuzzy attributes of leanness
To fuzzify the uVi and uSi and develop the membership grades of the leanness elements, we use Hondas
fuzzy components since Honda is the benchmarked entity as explained above. Therefore, we plot the
values of each of uVi and uSi using a continuum dened by Hondas opposite points (Table 2). We also use
open interval trapezoidal membership functions to locate a fuzzy attribute on the continuum. Using two
points, a and b corresponding to lower and higher values of Hondas attributes, we dene the
membership function as follows for fuzzifying uV1 (the mean of the RM&W ratio over 3 years for Ford):
 Set the interval for the mean value of RM&W ratio (uV1 ). Since Hondas uV1 0:076 (Fig. 2, Relation 3),
then a = 0.076. We set the value for the other extreme point, b, as 2a = b = 0.152. That is, if a companys
long-term mean of RM&W inventory ratio is twice that of Honda, then the companys relative
inventory level is so high that it either does not have a JIT system in place or this system is not working
toward making the company lean. This multiplier of 2 is an arbitrary selection that can be changed to a
different value by the analyst. Accordingly, the membership function is shown as follows:
8
1;
if uV
>
1  a
>
 V

<
u1  a
V
V
mF u1 m1 1 
; if a < uV
1 <b
>
b  a
>
:
V
0;
if u1  b
Plot these values (Fig. 3). On Fig. 3, locate points a and b to build the trapezoidal membership
function with an open interval.
 Using the plotted membership function in Fig. 3, determine Fords fuzzy attribute, mV1 that
corresponds to its uV1 of 0.080 (Fig. 2, Relation 3). The fuzzy component uV1 of 0.080 corresponds to
0.947 membership degree computed as follows (interpolating and assuming linearity (Table 2):
1

uV
0:080  0:076
1 a
0:947
1
0:152  0:076
ba

Similar procedures are applied to the remaining fuzzy components uV2 and uV3 and uS1 uS8 to yield
the eight elements of each of mVi and mSi as shown in Table 3.

M.E. Bayou, A. de Korvin / J. Eng. Technol. Manage. 25 (2008) 287304

299

Fig. 3. Plotting the membership function for the fuzzy components: Ford versus Honda.

A slightly different procedure is applied to the determination of mV4 mV8 . For determining mV4 , we
use Hondas uV4 of 0.126 as the highest point, b, on the membership continuum; the lowest point, a, on
this continuum is half this value, i.e., 0.063. Thus, the membership equation is:
8
1;
if uV4  a
>
>
 V

<
u

a
mF uV4 mV4 1  4
; if a < uV4 < b
>
b  a
>
:
0;
if uV4  b
Since Fords uV4 0:011 is less than 0.063, mV4 1. That is, the RM&W (uV4 ) component measures the
mean value of the annual change in inventory R4 (where the annual change is dened as the ending
inventory balance less the beginning inventory divided by the beginning inventory balance, as
explained above). A positive change in this ratio indicates a move away from the JIT goal of minimizing
inventory level, and vice versa. Since Hondas uV4 of +0.126 represents an increase rather than a
decrease in inventory, we assign a membership degree of zero if Fords or GMs uV4 is equal to or greater
Table 3
Fuzzication of the lean practices the components standard deviations.
Fuzzy
components

Hondas
uSi

mSi

u1
u2
u3
u4
u5
u6
u7
u8

0.010
0.024
0.033
0.254
0.014
0.008
0.018
0.010

1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000

Beginning of the
continuum, 0.0 (a)

End of the
continuum, 1.0 (b)

Fords
uSi

mSi

uSi

mSi

0.020
0.048
0.066
0.508
0.028
0.016
0.036
0.020

0.010
0.024
0.033
0.254
0.014
0.008
0.018
0.010

0.007
0.015
0.019
0.049
0.027
0.012
0.019
0.014

1.000
1.000
1.000
1.000
0.071
0.500
0.994
0.600

0.033
0.038
0.070
0.021
0.007
0.012
0.019
0.014

0.000
0.417
0.000
1.000
1.000
0.500
0.994
0.600

or
or
or
or
or
or
or
or

greater
greater
greater
greater
greater
greater
greater
greater

or
or
or
or
or
or
or
or

less
less
less
less
less
less
less
less

GMs

300

M.E. Bayou, A. de Korvin / J. Eng. Technol. Manage. 25 (2008) 287304

than 0.126, and a degree of 1.0 if this ratio is equal to or less than half that of Hondas uV4 (i.e.,
0.063). Again, the selection of this multiple of 0.5 of Hondas uV4 is arbitrary and can be changed to
t a particular application of this fuzzy leanness model. Indeed, a benchmarked rm may not be
the leader in every component of leanness. For example, Toyotas preoccupation with
standardization and procedures runs counter to the spirit of decentralization and managements
autonomy (Berggren, 1992; Monden, 1993); yet its TPS is the world-class approach to production
leanness. This is another argument for using a unifying concept and measure of leanness because
weaknesses in one component could be more than compensated for by strengths in other
components of lean.
To determine the mV5 mV8 fuzzy components, the midpoint between points a and b is set at zero;
thus, for mV5 , since Hondas uV5 0:005, Fords uV5 of 0.045 leads to mF uV5 mV5 1 using the
following equation (where a = 0.005, and b = +0.005):
8
1;
if uV5  a
>
>
 V

<
u

a
mF uV5 mV5 1  5
; if a < uV5 < b
>
b  a
>
:
0;
if uV5  b
The fuzzy component uV5 measures the improvement in CGS% from 2001 to 2003, as dened above.
A negative change in this ratio indicates a cost improvement (Kaizen). A positive change indicates a
lack of improvement.
Similarly, for each of the quality control surrogates, uV6 uV8 , the midpoint between points a and b is
zero, i.e., uVi  0  uVi , for i = 6, 7, and 8:
8
0;
if uV6  a
>
>
 V

<
u

a
mF uV6 mV6 1  6
; if a < uV6 < b
>
b  a
>
:
1;
if uV6  b
For example, since Hondas uV6 0:083, the interval for determining the membership function is
0.083  0  +0.083.
For Ford, uV6 0:058 leads to mF uV6 mV6 0:151,
For GM, uV6 0:038, yields mF uV6 mV6 0:729, computed as follows:
1

uV6  a
0:038  0:083
0:729
1
0:083  0:083
ba

The determination of mF(uV7 ) and mF(uV8 ) follows the same procedures. In addition, the
determination of membership degrees for the standard deviation in the fuzzy components uS1 uS8
to yield mS1 mS8 follows procedures similar to the ones used to determine the membership degrees for
uV6 uV8 except that a = 0 for each of these procedures since the standard deviation cannot be negative.
The nal step is the aggregation of these fuzzy attributes.
4.5.2. Aggregation of the fuzzy attributes
To assess the relative attributes of a companys leanness, we focus on both the mean value and
standard deviation of selected nancial indicators over the long run. We now aggregate the grades of
belief of these two measures using the fuzzy intersection operation, as explained above:

mF mi mF uVi \ mF uSi mVi \ mSi

Furthermore, to account for signicant values for assessing a companys leanness, we use an a-cut
of 0.33 as the threshold of the aggregation (Klir and Folger, 1988, p. 16).

V
S
ma mi 1; if mi \ mi   a
0;
Otherwise
Aggregating the fuzzy attributes and reducing them to a scalar (Klir and Folger, 1988, p. 16), we use
the concept of cardinality and sigma count. Cardinality of a non-fuzzy set is the number or proportion

M.E. Bayou, A. de Korvin / J. Eng. Technol. Manage. 25 (2008) 287304

301

of elements of the universal set U. A sigma count is the arithmetic sum of the membership grades in
the fuzzy subset F (Zadeh, 1978, 1989; Siegel et al., 1995). Thus, to determine the proportion of the
lean components that are signicant, we compute
P
X
CountLEAN \ MEAN \ STDEV \ THRESHOLD
P
CountLEANjJIT \ KAIZEN \ QUALITY
CountLEAN
P
P
V
1 \ mi \ mSi \ ma mi
mG mi

N
N
where LEAN = {R1, R2, . . .R8}, and Ri = Component i of leanness. Table 4 shows the results of these
aggregations.
Let us determine the truth of the following propositions;P3: Fords automotive operations are lean
over the period 20012003 vis-a`-vis Hondas attributes.
P Ford
mG mi 8=6  1  4:729
0:197

8
N
P4: GMs automotive operations are lean over the period 20012003 vis-a`-vis Hondas attributes.
P GM
mG mi 8=5  1  3:069
0:230

8
N

Table 4
Aggregation of belief grades regarding leanness key characteristics.
Ford Motor

R1
R2
R3
R4
R5
R6
R7
R8

RM&WIP
FG
Combined
D Inventory
D CGS%
MS cars
MS trucks
MS combined

General Motors

R1
R2
R3
R4
R5
R6
R7
R8

RM&WIP
FG
Combined
D Inventory
D CGS%
MS cars
MS trucks
MS combined

Honda Motor

R1
R2
R3
R4
R5
R6
R7
R8

RM&WIP
FG
Combined
D Inventory
D CGS%
MS cars
MS trucks
MS combined

uVi

uSi

mvi

mSi

mvi \ mSi

a = 0.33
ma

mG

0.080
0.134
0.214
0.011
0.045
0.058
0.068
0.056

0.007
0.015
0.019
0.049
0.027
0.012
0.019
0.014

0.947
1.000
1.000
1.000
1.000
0.151
0.444
0.338

1.000
1.000
1.000
1.000
0.071
0.500
0.994
0.600

0.947
1.000
1.000
1.000
0.071
0.151
0.444
0.338

1.000
1.000
1.000
1.000
0.000
0.000
1.000
1.000

0.947
1.000
1.000
1.000
0.000
0.000
0.444
0.338

uVi

uSi

mvi

mSi

mvi \ mSi

0.094
0.135
0.228
0.009
0.013
0.038
0.064
0.046

0.033
0.038
0.070
0.021
0.007
0.012
0.019
0.014

0.763
1.000
1.000
1.000
0.000
0.729
0.552
0.633

0.000
0.417
0.000
1.000
1.000
0.500
0.994
0.600

0.000
0.417
0.000
1.000
0.000
0.500
0.552
0.600

a = 0.33
ma

mG

0.000
1.000
0.000
1.000
0.000
0.000
0.000
1.000

0.000
0.417
0.000
1.000
0.000
0.500
0.552
0.600

mG

uSi

mvi

mSi

mvi \ mSi

a = 0.33
ma

mG

0.076
0.152
0.228
0.126
0.005
0.083
0.613
0.173

0.010
0.024
0.033
0.254
0.014
0.008
0.018
0.010

1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000

1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000

1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000

1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000

1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000

M.E. Bayou, A. de Korvin / J. Eng. Technol. Manage. 25 (2008) 287304

302

P5: Fords automotive operations are much leaner than GMs automotive operations over the period
20012003 vis-a`-vis Hondas attributes.

mFord
0:197
G
0:46

0:197 0:230
m mGM
G
Ford
G

Our degree of belief in the truth of proposition P5 is strongly supported by the ratio of 0.46 since our
idea that Ford is lean is supported by 0.197, which is far less than that of General Motors, P4, of 0.230.
Moreover, during the period 20012003, Ford is more than 17% ((0.230  0.197)/0.197) leaner than
GM.
5. Limitations
This study has several limitations:
(1) We focused on only a few components of leanness, mainly JIT, Kaizen and quality control.
However, these practices of leanness have received more attention than many other practices, as
the extensive literature review of lean shows in Table 1.
(2) We used surrogates to measure the components of leanness. We may be justied in using
surrogates for these components because JIT, Kaizen and quality control are more philosophical
concepts than concrete instruments. For example, the JIT philosophy includes much more than
simply reducing inventory. JIT pertains to factory layout, relationship with suppliers, and a whole
attitude toward eliminating Muda (waste in Japanese) (Imai, 1997).
(3) Consolidated nancial statements are the source of our data. We argue that the concept of leanness
is a holistic concept in that a proper examination of it requires an examination of the total entitys
characteristics.
(4) Honda Motor Company may not be the best company to select for benchmarking purposes;
Toyota is a better choice. However, Toyota does not publish its consolidated nancial
statements according to US GAAP. The second best Japanese automobile company that does is
Honda.
5.1. Suggestions for future research
An extension of the present study may use fuzzy clustering analysis. Several companies in an
industry are classied into different clusters according to leanness criteria. Consequently, we can
determine to what degree a company belongs to a given cluster. Another extension of this study may
incorporate uncertainties in the membership measurement, which the current paper does not
consider, e.g., in Fig. 3. For example, instead of assuming that the extreme interval point a in Fig. 3 is
not precisely known, then it may lie in an interval [a1, a2]; thus, the trapezoid functions of Fig. 3 would
be replaced by interval valued fuzzy sets, as Fig. 4 shows.

Fig. 4. Extension for future research: fuzzy interval for membership measurement.

M.E. Bayou, A. de Korvin / J. Eng. Technol. Manage. 25 (2008) 287304

303

6. Summary and conclusions

The burgeoning research studies on lean production system have emphasized that the Toyota
production system approach to lean is the key to growth and survival amid the increasing global
competition, fast changes in manufacturing and service logistics and economic recessions. However,
these studies fall short of delivering a systematic measure of leanness by which companies can be
compared across time and with each others regarding the degree of leanness of their production
systems. In addition, the concept of leanness is underdeveloped because many studies dene lean in
terms of its objectives, which vary, overlap and differ among rms and among researchers. This paper
denes leanness as a dynamic, relative and long-term concept. Accordingly, it develops a relative
measure of lean, which has the following characteristics: relative, dynamic, long-term, integrative,
holistic and objective.
Using a fuzzy-logic methodology, this case study lists the steps of an algorithm and applies it to the
automobile industry. The measure compares the production systems of Ford Motor Company and
General Motors using Honda Motor Companys system as a benchmark over the years 20012003,
using their audited consolidated nancial statements and other pertinent information reported in the
SEC Form 10-K for Ford and GM and Form 6-K for Honda. The results of analyzing several propositions
over the period 20012003 show that on a scale from 0.0 to 1.0, the production systems of Ford and
GM are 0.197 and 0.230, respectively, as compared to Hondas system, which is set at 0.000 as the
frame of reference. Therefore, our belief that Fords system is leaner than GMs system is supported by
0.46[0.197/(0.197 + 0.230)]. The results also indicate that Fords system is more than 17% leaner than
GMs system over the 3-year period.
References
Ahmad, S., Schrader, R.G., Sinha, K.K., 2003. The role of infrastructure practices in the effectiveness of JIT practices: implications
for plant competitiveness. Journal of Engineering and Technology Management 20, 161191.
Allway, M., Corbett, S., 2002. Shifting to lean service: stealing a page from manufacturers playbooks. Journal of Organizational
Excellence 21 (Spring (2)), 4554.
Baudin, M., 2004. Lean Logistic, The Nuts and Bolts of Delivering Materials and Goods. Productivity, New York, NY.
Berggren, C., 1992. Alternatives to Lean Production, Work Organization in the Swedish Auto Industry. ILR Press, Ithaca, NY.
Creese, R.C., 2000. Cost management in lean manufacturing enterprises. AACE International Transactions. Morgantown, C5A
11A.
Drucker, P., 1967. The Effective Executive. Harper Business Essentials, New York, NY.
Emiliani, M.L., 2000. Cracking the code of business. Management Decisions 38 (2), 6079.
Goldsby, T., Martichenko, R., 2004. Lean Six Sigma Logistics. J. Ross Publishing, Boca Raton, FL.
Hall, R.H., 2004. Lean and the Toyota Production System. Target.
Hines, P., Taylor, D., 2000. Going Lean. Lean Enterprise Research Centre Cardiff Business School, Cardiff, UK.
Imai, M., 1997. Gemba Kaizen. McGraw-Hill, New York.
James-Moore, S.M., Gibbons, A., 1997. Is lean manufacturing universally relevant? An integrative methodology. International
Journal of Operations Product Management 17 (9), 899911.
Klir, G.J., Folger, T.A., 1988. Fuzzy Sets, Uncertainty, and Information. Prentice-Hall, Englewood Cliffs, NJ.
Kolesar, P.J., 1993. The relevance of research on statistical process control to the total quality movement. Journal of Engineering
and Technology Management 10, 317338.
Liker, J.K., 1998. Introduction: bringing lean back to the U.S.A. In: Liker, J.K. (Ed.), Becoming Lean, Inside Storied of U.S.
Manufacturers. Productivity Press, Portland, OR, pp. 340.
MacDufe, J.P., Helper, S., 1997. Creating lean suppliers: diffusing lean production through the supply chain. California
Management Review 39 (Summer (4)), 118151.
Monden, Y., 1993. Toyota Management System: Linking the Seven Key Functional Areas. Productivity Press, Portland.
Pagatheodrou, Y., 2005. The price of leanness. Industrial Management 47 (Jan/Feb (1)), 814.
Rother, M., 1998. Which way will you turn on the road to lean? In: Likert, J.K. (Ed.), Becoming Lean. Productivity, Inc., Portland,
OR.
Schrader, S., Riggs, W.M., Smith, R.P., 1993. Choice over uncertainty and ambiguity in technical problem solving. Journal of
Engineering and Technology Management 10, 7399.
Shah, R., Ward, P., 2003. Lean manufacturing: context, practice bundles, and performance. Journal of Operations Management
21, 129149.
Shook, J.Y., 1998. Bringing Toyota production system to the United States, a personal perspective. In: Liker, J.K. (Ed.), Becoming
Lean, Inside Storied of U.S. Manufacturers. Productivity Press, Portland, OR, pp. 4170.
Siegel, P.H., Rigby, J.T., Bourgeois, B., 1995. An application of fuzzy set theory to the problem of evaluating net realizable value of
accounts receivable. In: Siegel, P., de Korvin, A., Omer, K. (Eds.), Application of Fuzzy Sets and the Theory of Evidence to
Accounting, vol. 3, JAI Press, Inc., Stamford, Conn., pp. 89102.
Sullivan, W.G., McDonald, T.N., Van Aken, E., 2002. Equipment replacement decisions and lean manufacturing. Robotics and
Computer Integrated Manufacturing 18, 255265.

304

M.E. Bayou, A. de Korvin / J. Eng. Technol. Manage. 25 (2008) 287304

Womack, J.P., Jones, D.T., Roos, D., 1991. The Machine That Changed the World. Harper Collins, New York, NY.
Womack, J.P., Jones, D.T., 1996. Lean Thinking. Simon and Schuster, New York, NY.
Zadeh, L.A., 1965. Fuzzy sets. Information and Control 8, 338353.
Zadeh, L.A., 1978. Pruf-meaning representation language for natural language. International Journal of Man-Machine Studies 10,
395460.
Zadeh, L.A., 1988. Fuzzy logic. Computer 1, 8393.
Zadeh, L.A., 1989. Knowledge representation in fuzzy logic. IEEE Transactions Knowledge and Data Engineering 1, 1.
Zebda, A., 1989. Fuzzy set theory and accounting. Journal of Accounting Literature 8, 76105.