Journal of Operations Management 30 (2012) 521532

Contents lists available at SciVerse ScienceDirect

Journal of Operations Management

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

The effects of Six Sigma on corporate performance: An empirical investigation

Scott M. Shafer a, , Sara B. Moeller b,1
a
b

Wake Forest University, Schools of Business, Winston-Salem, NC 27109, United States

University of Pittsburgh, Katz Graduate School of Business and College of Business Administration, Pittsburgh, PA 15260, United States

a r t i c l e

i n f o

Article history:
Received 22 July 2011
Received in revised form
13 September 2012
Accepted 18 October 2012
Available online 29 October 2012
Keywords:
Six sigma
Event study
Process improvement
Corporate performance

a b s t r a c t
The purpose of this study is to investigate the impact of adopting Six Sigma on corporate performance.
Although there is a fairly large and growing body of anecdotal evidence associated with the benets of
implementing Six Sigma, there is very little systematic and rigorous research investigating these benets.
This research extends previous research in several important ways including utilizing a sample of 84 Six
Sigma rms that represent a wide variety of industries and rm characteristics, utilizing rigorously constructed control groups to ensure the validity of our comparisons and conclusions, and investigating the
impact of adopting Six Sigma on corporate performance over a ten year period. To carry out this investigation, the event study methodology is employed. The ten year period consists of three years prior to Six
Sigma implementation, the event year corresponding to the year Six Sigma is adopted, and six years post
Six Sigma implementation. To assess the impact of adopting Six Sigma on corporate performance we utilize commonly used measures including Operating Income/Total Assets (OI/A), Operating Income/Sales
(OI/S), Operating Income/Number of Employees (OI/E), Sales/Assets (S/A), and Sales/Number of Employees (S/E). The sample Six Sigma rms are compared to different benchmarks including the overall industry
performance and to the performance of carefully selected portfolios of control rms. The results of the
study indicate that adopting Six Sigma positively impacts organizational performance primarily through
the efciency with which employees are deployed. More specically, enhanced employee productivity
results were observed in both static analyses that assessed the performance of the sample Six Sigma
rms relative to their control groups at discrete points in time and dynamic analyses of the Six Sigma
rms rate of improvement relative to the rate of improvement of their control groups. Benets in terms
of improved asset efciency were not observed. Finally, there was no evidence that Six Sigma negatively
impacts corporate performance.
2012 Elsevier B.V. All rights reserved.

1. Introduction
The Six Sigma methodology was created by Motorola in the mid
1980s. Over time it has evolved into a comprehensive approach for
improving business performance. Key elements of the Six Sigma
approach include a clear focus on the customers needs, the use
of performance metrics, a focus on improving business processes
often through the reduction of inherent variation in the processes,
clearly dened process improvement specialist roles, the use of
data-driven and highly structured problem solving methodologies,
and ultimately the generation of tangible business results (Hahn
et al., 1999; Linderman et al., 2003; Schroeder et al., 2008). Pande
et al. (2000, p. xi) provide a representative denition of Six Sigma
as:

Corresponding author. Tel.: +1 336 758 3687.

E-mail addresses: shafersm@wfu.edu (S.M. Shafer), sbmoeller@katz.pitt.edu
(S.B. Moeller).
1
Tel.: +1 412 648 0137.
0272-6963/$ see front matter 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jom.2012.10.002

A comprehensive and exible system for achieving, sustaining, and maximizing business success. Six Sigma is uniquely
driven by close understanding of customer needs, disciplined
use of facts, data, and statistical analysis, and diligent attention
to managing, improving, and reinventing business processes.
Six Sigma is a particularly timely topic and appears to be gaining
momentum in practice (Linderman et al., 2003; Schroeder et al.,
2008). Perhaps one factor driving the current popularity of Six
Sigma is the growing body of anecdotal evidence touting the benets high prole organizations have reported from their Six Sigma
initiatives. For example, in the three years ending in 2001, GE estimated that it saved $8 billion as a result of its Six Sigma initiatives
(Arndt, 2002). In the following year, GE budgeted $600 million for
Six Sigma projects and targeted an additional $2.5 billion in savings.
As another example, Bank of America claimed benets in excess
of $2 billion and increased customer delight by 25% in less than
three years through its Six Sigma initiatives (Jones, 2004). Importantly, Bank of Americas experience demonstrates the applicability
of Six Sigma beyond traditional manufacturing processes. Indeed,
Honeywell found that the average savings it achieved from service

522

S.M. Shafer, S.B. Moeller / Journal of Operations Management 30 (2012) 521532

projects were double that of manufacturing projects (Bossidy and

Bonsignore, 1999). Motorola, the inventor of the Six Sigma methodology, estimated that over the 20 plus years it has deployed Six
Sigma it has documented savings in excess of $20 billion (Motorola,
2011). Six Sigma has also been credited as an important contributor to its winning the Malcom Baldrige Award for Quality in 1988
(Hahn et al., 1999).
Although there is fairly large and growing body of anecdotal
evidence associated with the benets of implementing Six Sigma,
there is very little systematic and rigorous research investigating
these benets. Linderman et al. (2003) argue that although Six
Sigma has had a substantial impact on industry, the academic community lacks theory as a basis for research on Six Sigma. Antony
(2004) agrees and notes that the despite the huge impact Six Sigma
has had on industry, the academic community lags behind in its
understanding of it. Schroeder et al. (2008) further argue that
research is needed to determine the impact Six Sigma has on performance improvement.
The purpose of this study is to investigate the impact adopting
Six Sigma has on corporate performance. To accomplish this objective we study the performance of organizations that have publically
announced or have received other publicity about their adoption
of Six Sigma. Beyond providing a clear adoption date, such public
disclosures may also serve as a proxy regarding the organizations
commitment to Six Sigma in a similar fashion to the way Hendricks
and Singhal (1997) used quality award winners as a proxy for effective TQM implementation.
The results of the study indicate the adoption of Six Sigma positively impacts organizational performance primarily through the
efciency with which employees, but not assets, are deployed.
There is no evidence that Six Sigma negatively impacts corporate
performance. In addition, the results suggest that better performing
rms adopt Six Sigma and they continue their performance advantage after adoption. Furthermore, the performance advantage for
the Six Sigma rms in terms of employee productivity tended to be
larger after adopting Six Sigma and tended to increase as additional
experience was gained with Six Sigma. The benets of adopting
Six Sigma were observed in both the static analysis that assessed
the performance of the sample Six Sigma rms at discrete points
in time and the dynamic analysis of the Six Sigma rms rate of
improvement on many different benchmarks.
This research extends previous research in several important
ways. First, we evaluate a variety of different benchmarks to ensure
that the benchmark choice is not driving the results. At one end of
the spectrum of benchmarks, we take a nave viewpoint and use an
industry adjusted performance of our sample Six Sigma rms. At the
other end of the spectrum, we follow Barber and Lyon (1996) and
compare a sample Six Sigma rms performance to the performance
of the closest matched rm and a portfolio of control rms matched
to it on the basis of industry, year, and similar past performance.
On all of the benchmarks, we do many robustness tests including
when and how we match the sample rm to the benchmark and
across all of these variations, our results are consistent.
Second, we investigate the impact of Six Sigma on operating
performance over a ten year period. Investigating the long-term
effects of adopting Six Sigma addresses important gaps in the
literature. To carry out this investigation, the event study methodology is employed. The ten year period consists of three years
prior to Six Sigma implementation, the event year corresponding
to the year Six Sigma was adopted, and six years post Six Sigma
implementation. Pre-implementation performance data is used for
performance matching Six Sigma sample rms with control rms as
well as to investigate the role past rm performance plays in motivating rms to adopt Six Sigma. A six-year post-implementation
period is used given an expected lag between Six Sigma implementation and the realization of performance benets. Previous

research has indicated a two and a half year or longer lag between
implementing total quality management (TQM) and improved performance (GAO, 1991; Powell, 1995). Likewise, Hendricks and
Singhal (2001a, b) suggest a three to ve year period to implement
an effective TQM program. The ten year period was also chosen
so that short-term and longer-term patterns in the performance
of the sample Six Sigma rms could be investigated. For example,
one of the most interesting results observed was that the Six Sigma
rms outperformed their matched portfolios in year 3 in terms of
Operating Income/Total Assets (OI/A), Operating Income/Number
of Employees (OI/E), and Sales/Number of Employees (S/E), then
experienced a signicant decline in performance prior to adopting
Six Sigma on these three measures, and nally exhibited a quick
rebound in year +1 upon adopting Six Sigma. Likewise, as an example of longer term patterns, the performance advantage for the
Six Sigma rms in terms of employee productivity tended to be
larger after adopting Six Sigma and tended to increase as additional
experience was gained with Six Sigma.
Third, beyond extending the research investigating the impact
of Six Sigma on rm performance, an additional contribution of this
research is to provide performance benchmarks for organizations
that have adopted or are considering adopting Six Sigma. Also, the
inclusion of commonly used measures of corporate performance
including OI/A, Operating Income/Sales (OI/S), OI/E, Sales/Assets
(S/A), and S/E facilitate comparisons with previous research.
This paper is organized as follows. Section 2 reviews the existing
empirical research related to process improvement methodologies
and rm performance. Section 3 provides the theoretical development for Six Sigmas impact on corporate performance, our
research hypotheses, and the performance variables included in
the study. Following this, our research methodology is discussed in
Section 4. Our empirical results are presented and discussed in Section 5. Finally, the paper is concluded in Section 6 with a discussion
of limitations and avenues for future research.

2. Review of empirical evidence of quality and process

improvement initiatives on corporate performance
While Six Sigma is the latest process improvement methodology, the inuence of earlier process improvement methodologies in
its development, particularly TQM and JIT/lean, are readily apparent. In this section we critically review the empirical research
investigating process improvement methodologies on corporate
performance in order to understand what has been studied and
then based on this understanding highlight the gaps in the literature addressed by the present study.
While there is a substantial body of empirical research
investigating quality and process improvement initiatives on corporate performance, rigorous research investigating the impact
of Six Sigma on corporate performance has been limited (Foster,
2007). This is supported by observing that only two of the 23
research contributions encountered in the literature review for
this study investigated the impact of Six Sigma on corporate
performance. Approximately half the studies investigating the
impact of various process improvement approaches on corporate performance utilized event studies and the other half utilized
surveys.
Fortunately, rigorous empirical research investigating the
impact of Six Sigma is beginning to emerge including the use of
event studies (Goh et al., 2003; Foster, 2007) and surveys (Lee
and Choi, 2006). While limited in quantity, this research tends
to contradict much of the anecdotal evidence because an overwhelmingly positive relationship between Six Sigma and corporate
performance has not been found. For example, Foster (2007) found
the impact of Six Sigma on operating and nancial performance

S.M. Shafer, S.B. Moeller / Journal of Operations Management 30 (2012) 521532

was mixed. Signicant main effects were found for cost per dollar
sales, EBITDA, sales, S/E, and number of employees while signicant main effects were not found for free cash ow per share, asset
turnover, return on assets (ROA), return on investment (ROI), and
total assets. A signicant limitation of the study is that the performance of the Six Sigma rms was compared to a random sample
chosen from the Fortune 500 as opposed to carefully selecting control groups. Barber and Lyon (1996) highlight the importance of
performance matching sample rms with control rms to ensure
that test statistics are well specied. Furthermore, the sample size
of 24 Six Sigma rms and the fact that only rms that announced
their adoption of Six Sigma from 1996 to 1998 limits the generalizability of the results. For example, if differences exist between
earlier adopters of Six Sigma and more recent adopters then the
rms included in the study would not provide a representative
sample of all organizations that have adopted Six Sigma.
Utilizing structural equation modeling, Lee and Choi (2006)
investigated how four Six Sigma management activities impact
process innovation, quality improvement, and corporate competitiveness improvement. The results of the study indicated that all
four management activities have a positive impact on process innovation. Furthermore, the results indicated that process innovation
signicantly affects quality improvement which in turn affects corporate competitiveness. Key limitations of this study are that only
a single organization was studied and as is the case with all survey
based research, there may be a self-report bias.
Of the process improvement approaches researched to date,
TQM has been researched the longest and accounts for almost half
of the empirical studies reviewed. In contrast to the studies investigating Six Sigma, the TQM event studies have generally found
a positive relationship between TQM and corporate performance
(Hendricks and Singhal, 1997; Easton and Jarrell, 1998; Hendricks
and Singhal, 2001b; Eriksson and Hansson, 2003). Hendricks and
Singhal (1997) found strong support for the hypothesis that quality
award winning rms outperformed a control sample on operating
income-based measures. In another study of quality award winning
rms, Eriksson and Hansson (2003) found that the quality award
recipients outperformed the competitor they were matched to in
terms of change in sales during the implementation period and
post implementation outperformed their matched competitor on
change in sales and return on assets. Easton and Jarrell (1998) found
strong evidence of overall improvement in operational and nancial performance over the long-term. More specically, across all
the variables investigated, more than half the TQM sample rms
outperformed their control group in terms of exceeding analysts
forecasts. Hendricks and Singhal (2001b) extended their previous research and investigated the role several rm characteristics
play in moderating the impact implementing effective TQM programs has on nancial performance over a four to ve year period.
The results indicated that smaller rms do signicantly better
than larger rms and rms that received awards from independent organizations signicantly outperformed rms that received
supplier-based awards. The results of the study provided weak
support that less capital intense rms outperform more capital
intense rms and that more focused rms outperform more diversied rms. Finally, there were no signicant differences observed
between early and late adopters of TQM.
A key strength of many of the TQM event studies are the use
of quality award winning rms which helps ensure only sample rms that effectively implemented TQM were included in the
study (Hendricks and Singhal, 1997, 2001b; Eriksson and Hansson,
2003). A key weakness of the TQM event studies is that TQM
sample rms were not performance matched with control rms
(Hendricks and Singhal, 1997, 2001b; Eriksson and Hansson, 2003)
or were matched in unconventional ways (Easton and Jarrell,
1998).

523

In addition to the use of event studies, the impact of TQM on

corporate performance has been investigated with other methodologies. For example, twenty of the highest scoring Malcolm
Baldrige National Quality Award applicants from 1988 and 1989
were studied by the United States General Accounting Ofce at the
request of Congressman Donald Ritter (GAO, 1991). The study found
that in almost all 20 cases, the companies had improved employee
relations, productivity, customer satisfaction, market share, and
protability. Key limitations of the study include the small sample
size and the fact that the results were not based on rigorous statistical analysis. Like the TQM event studies discussed earlier, a strength
of the study was the use of quality award winners which serves as
a proxy for having implemented an effective TQM program.
Surveys have also been used to investigate TQM. Adam (1994)
surveyed manufacturing rms and found a signicant relationship
between the quality improvement approach and operating and
nancial performance. Powell (1995) found that TQM rms outperformed non-TQM rms which in turn provided evidence that
TQM provides economic value to organizations. Powell also found
that long time adopters tended to be more satised with their TQM
programs. Handeld et al. (1998) found that customer satisfaction
was a primary driver of nancial performance. A key implication
associated with this nding is that the nancial returns associated
with investments in quality may be highly dependent on customer
satisfaction. Samson and Terziovksi (1999) investigated the issue of
whether elements of TQM could be used to predict organizational
performance and if so which elements are the best predictors of
organizational performance. Three of the elements of TQM: leadership, human resources management, and customer focus were
found to be signicant and positively related to organizational performance. Planning and process management were not found to be
signicantly related to organizational performance, while information and analysis was found to be signicant and negatively related
to performance. Other surveys have found a positive relationship
between the extent to which TQM practices were implemented and
rm performance (Douglas and Judge, 2001; Kaynak, 2003). While
these surveys each provide important insights it is also worth noting that survey results are limited by the respondents memory,
ability to respond, and honesty. Furthermore, there is always the
concern that those that responded to the survey are not representative of the general population of interest.
Nair (2006) performed a meta-analysis on 23 quality management empirical studies over the period of 19952004. The results
of the meta-analysis support the hypotheses that quality management practices are positively related to aggregate performance and
that this relationship is inuenced by moderating factors. Interestingly, while the results supported a positive relationship between
the use of quality information tools and aggregate performance,
support for a direct relationship between quality information tools
and nancial or operational performance was not found. Quality
information tools are widely used as part of Six Sigma programs.
Beyond TQM, a number of studies have also investigated the
relationship between ISO 9000 certication and rm performance.
On one side, Terziovski et al. (1997) and Singels et al. (2001) were
unable to nd evidence that ISO certication was related to organizational performance. In contrast, Terziovski et al. (2003) found
a positive relationship between organizational performance and
managements motives for adopting ISO 9000 and Corbett et al.
(2005) found that ISO 9000 manufacturing rms achieved significantly better nancial performance three years after obtaining
certication, however, the magnitude and timing of the improved
performance depended on the way the control group was specied.
Given ISO 9000s focus on documenting processes as opposed to
improving them per se, the mixed results across the studies investigating the adoption of ISO 9000 and organizational performance
are not surprising.

524

S.M. Shafer, S.B. Moeller / Journal of Operations Management 30 (2012) 521532

Finally, the research conducted to date has generally found a

positive relationship between lean/just-in-time (JIT) and corporate performance. Using the event study methodology, Huson and
Nanda (1995) found that adopting JIT resulted in increased earnings
per share driven primarily by a 24% increase in inventory turnover.
In another event study, Balakrishnan et al. (1996) reported similar results nding that the JIT sample rms substantially increased
their inventory turnover and decreased WIP inventory as a percentage of sales through the adoption of JIT. In a nal event study, Kinney
and Wempe (2002) found that both components of return on assets,
asset turnover and prot margin, improved for the JIT adopters
compared to the non-adopters. These researchers also found that
improved margins were the primary driver of the improvements in
ROA and that the improvements in ROA were concentrated among
the early adopters of JIT indicating a rst-mover advantage.
The relationship between JIT/lean and corporate performance
has also been investigated through survey based research. Shah
and Ward (2003) combined 22 lean practices into four lean bundles: JIT, continuous improvement, total preventive maintenance,
and human resource management. A positive association between
each lean bundle and operational performance was found and,
in total, the lean bundles explained 23% of the variation in operational performance. Fullerton et al. (2003) found statistically
signicant relationships between measures of protability and the
degree to which JIT practices were used. Cua et al. (2001) investigated the relationships between manufacturing performance and
multiple process improvement approaches related to continuous
improvement including TQM, JIT, and total productive maintenance. The results of the study supported the hypothesis that
higher levels of manufacturing performance can be attained when
multiple continuous improvement approaches are simultaneously
implemented.

2.1. Literature synthesis

The literature related to the impact of Six Sigma on corporate performance is largely anecdotal in nature and tends to
overwhelmingly cite the benets of Six Sigma on corporate performance (e.g. Benitez et al., 2007; Craven et al., 2006; Daniels, 2009;
Deshpande et al., 2004; Dudman, 2005; Johnson, 2005; Jones, 2004;
Mukherjee, 2008). However, not all the empirical evidence is positive. Chakravorty (2010) cites research suggesting that almost 60%
of Six Sigma initiatives at corporations do not generate the desired
results.
Furthermore, in reviewing the literature on quality and process
improvement methodologies, there is a lack of rigorous research
investigating Six Sigma in comparison to the volume of research
conducted in other areas despite the substantial interest in Six
Sigma from industry. In addition to this lack of research, the
research conducted to date suffers from signicant shortcomings.
For example, the scope of the Six Sigma studies to date has been
quite limited. Fosters (2007) event study included only 24 Six
Sigma organizations and Lee and Chois (2006) study surveyed only
employees of Samsung. Furthermore, in contrast to the empirical
TQM and JIT/lean research, the Six Sigma studies have not generally
found an overwhelmingly positive relationship between Six Sigma
and corporate performance. This is somewhat surprising given the
extent to which Six Sigma practices overlap with and perhaps complement TQM practices. For example, Zu et al. (2008) concluded
that Six Sigma includes practices distinct from TQM practices and
that these distinct Six Sigma practices complement traditional TQM
practices in terms of improving corporate performance. Thus, the
purpose of this study is to build on the limited empirical research
on Six Sigma and rigorously investigate the impact adopting Six
Sigma has on corporate performance.

3. Theory development, research hypotheses, and

performance variables
Six Sigma is theoretically different from other process improvement methodologies so investigating its effect on performance
is valuable. However, because of its similarities with other process improvement approaches, particularly TQM, there has been
an ongoing debate related to the extent to which it differs from
TQM (Schroeder et al., 2008). In terms of similarities between Six
Sigma and TQM, Schroeder et al. (2008) note the following:
Both TQM and Six Sigma emphasize the value of obtaining customer input and the use of quality function deployment in
product/service design.
Both Six Sigma and TQM emphasize process ownership and having clearly dened processes.
Both approaches recognize the importance of top management
leadership and support.
Involving employees is emphasized by both approaches. However, the approaches differ in the employees involved. In
particular, Six Sigma tends to rely on process improvement
specialists while TQM emphasizes involving all employees, especially shop oor employees.
Both approaches recognize the importance of collecting and
reporting quality data.
Considerable emphasis is given to understanding the needs of the
customer in both Six Sigma and TQM.
In terms of differences between Six Sigma and TQM, Zu et al.
(2008) identied three new practices associated with Six Sigma:
Six Sigma has well-dened process improvement specialist roles
(e.g. Green Belt, Black Belt, Master Black Belt) that are supported
with extensive training. Others have also highlighted the use
of full-time specialist roles as a key characteristic of Six Sigma
(Schroeder et al., 2008; Antony, 2004).
Six Sigma utilizes a structured process improvement methodology called DMAIC in combination with a well-dened set of
tools that are applied at various phases of the DMAIC methodology. The acronym DMAIC refers to the ve phase in a Six Sigma
process improvement project: dene, measure, analyze, improve,
and control.
An important focus of Six Sigma is the use of process
improvement metrics to monitor process performance and set
improvement goals. Six Sigma advocates the use of several new
process performance metrics such as defects per million opportunities (DPMO) and process sigma as well as the use of traditional
process performance metrics such as process capability and
rolled throughput yield.
Using exploratory and conrmatory factory analyses, Zu et al.
(2008) found that these three Six Sigma practices were implemented as distinct practices from seven traditional quality
management practices also considered in the study. Based on the
signicant relationships observed in the structural model developed by Zu et al. (2008), a model of the relationships between the
adoption of Six Sigma and organizational performance is presented
in Fig. 1. The three new Six Sigma practices which are highlighted
in the gure as dashed boxes illustrate not only how Six Sigma is
different from TQM, but also how Six Sigma impacts corporate performance. The seven TQM practices also investigated by Zu et al.
(2008) are not included in Fig. 1 for clarity of presentation and
because they are beyond the scope of the present study.
The model presented in Fig. 1 is consistent with other denitions
of Six Sigma in the literature. For example, Schroeder et al. (2008)
dene Six Sigma as an organized, parallel-meso structure to reduce

S.M. Shafer, S.B. Moeller / Journal of Operations Management 30 (2012) 521532

525

Product/
Service
Design
Top
Management
Support

Six Sigma
Role
Structure

Six Sigma
Focus on
Metrics

Quality
Performance
Process
Management
Business
Performance

Six Sigma
Improvement
Procedure

Fig. 1. Theoretical model of relationship between Six Sigma and organizational performance.

variation in organizational processes by using improvement specialists, a structured method, and performance metrics with the aim
of achieving strategic objectives. Importantly, all three Six Sigma
practices shown in Fig. 1 are key elements of this denition.
Beyond identifying the elements of Six Sigma, the denition
offered by Schroeder et al. (2008) describes the goal associated with
Six Sigma, namely, to reduce the variation inherent in organizational processes and as such claries how the Six Sigma practices
shown in Fig. 1 contribute to corporate performance. Accordingly,
Six Sigma seeks to improve business processes by studying and
reducing the inherent variation present in the process. Less process variation boosts quality performance by allowing the process
to obtain more consistent outcomes in terms of quality, lead times,
yield rates, and so on.
Of course, and is shown in Fig. 1, reducing process variation and
improving quality performance are simply the means to an end.
Fundamentally, the overarching goal is enhanced business performance including higher sales, increased market share, increased
operating income, improved return on assets, and so on. For example, improved process execution resulting in less process variation
leads to higher quality performance which in turn could provide a
competitive advantage that translates into increased market share
and higher revenues. Likewise, less process variation can help to
reduce waste and inefciency which in turn leads to lower costs
and higher protability.
Based on these insights and the model shown in Fig. 1, we offer
two hypotheses related to the impact of Six Sigma on business
performance. First, we hypothesize that implementing Six Sigma
will improve the organizations protability. To assess protability
and to be consistent with Zu et al. (2008), we rely on Operating
Income (OI) before depreciation, interest, and taxes as opposed to
net income (NI). OI is calculated as sales minus total cost (cost of
goods sold + selling and administrative expenses). OI provides a relatively clean measure of the cash that is generated from operations
since it is not impacted by decisions made on how to treat depreciation and amortization, by interest charges which are impacted by
the capital structure of the rm, or by taxes that can be inuenced
by a variety of decisions. However, two factors OI does not control
for are the affect that capital expenditures (particularly mergers
and acquisitions) have and the size of the rm. Therefore, in order
to retain this clean measure of cash ows while at the same time
controlling for capital expenditures and rm size, we normalize OI
by dividing it by the average of total assets (A), sales (S), and number of employees (E). Thus, we assess the impact of Six Sigma on
protability based on OI/A (or return on assets, ROA), OI/S (or return
on sales, ROS), and OI/E. According to Barber and Lyon (1996), ROA
is the most commonly used measure in studies aimed at detecting
abnormal operating performance.
Second, consistent with Zu et al. (2008), we hypothesize that
implementing Six Sigma will increase the organizations revenues.

Like OI, to control for capital expenditures and rm size, we normalize S by dividing it by the average of total assets and the number
of employees. Thus, we assess the impact of Six Sigma on revenues
based on S/A and S/E.
Zu et al. did not investigate Total Costs as a separate factor and
investigating it in the present study would not provide additional
insight since it can be derived from OI and S which are already
included (i.e., total cost = S OI). Essentially then, this study focuses
on the impact Six Sigma has on an organizations the top line (S)
and bottom line (OI).
4. Research methodology
4.1. Sample rm selection
In searching the web for organizations that adopted Six Sigma,
a list of approximately 400 organizations that were reported
adopters of Six Sigma was discovered. This list became the starting
point for identifying the sample Six Sigma organizations for this
study. Systematically, follow-up Google queries, searches of each
organizations website and annual reports, and queries of publication databases were executed for each public organization on
the list in an effort to identify the date when the organization
adopted Six Sigma. One advantage to studying Six Sigma compared
to other process improvement methodologies such as TQM or lean
is that with Six Sigma there is typically a clear start date as Six
Sigma is often initiated with some type of formal training program.
Ultimately, we were able to nd public announcements or articles discussing the year that Six Sigma was adopted for 88 of the
public organizations on the list. Of these 88 organizations, Compustat nancial data was available for 84 of the rms and these 84
rms comprised our sample Six Sigma rms. Because the disclosures used to identify the adoption date tended to coincide with
the actual adoption of Six Sigma, the inclusion of the sample Six
Sigma rms in this study is unrelated to their ultimate success or
lack of success with Six Sigma. Thus, while public disclosures may
be related to an organizations commitment to its Six Sigma initiative, there is no bias toward including only rms that have had
success with Six Sigma.
Table 1 provides distribution data on when the 84 sample rms
adopted Six Sigma. Approximately 45% of the sample rms adopted
Six Sigma in 2000 or 2001 and 81% adopted it between 1998 and
2002. It is interesting to observe the decline in identifying rms
announcing or receiving other publicity regarding their adoption of
Six Sigma post 2001. This may reect a perception that Six Sigma
has become more mainstream and therefore less worthy of a formal
announcement.
The 84 sample Six Sigma rms included in this study represents
a diverse set of rms. In particular, the sample rms represent 57
distinct 4-digit SIC codes and 27 unique 2-digit SIC Codes. Table 2

526

S.M. Shafer, S.B. Moeller / Journal of Operations Management 30 (2012) 521532

Table 1
Distribution of the year when sample rms adopted Six Sigma.
Year

Number of rms

Percentage of rms

1986
1988
1990
1994
1995
1997
1998
1999
2000
2001
2002
2003
2004
19862004

1
1
1
2
1
5
13
8
16
22
9
3
2
84

1.2
1.2
1.2
2.4
1.2
6.0
15.5
9.5
19.0
26.2
10.7
3.6
2.4
100

Table 2
Summary nancial data for Six Sigma sample rms in year Six Sigma adopted.a
OI
(000,000)
Mean
Median
Std Dev
Max
Min

3851
1380
7005
37,895
1391

S
(000,000)
18,001
10,072
26,987
183,691
506

A
(000,000)
51,036
9569
120,304
665,287
454

E
(000)
59.6
35.5
74.3
373.8
1.9

a
OI, S, and A in each sample rms event year adjusted to equivalent 2004 dollars
using CPI, The Federal Reserve Bank of Minneapolis, http://www.minneapolisfed.
org, May 23, 2011.

provides summary nancial data on the sample rms based on the

year the rms adopted Six Sigma. All nancial data were obtained
from the Compustat Annual Industrial File.
The research methodology employed in the present study
overcomes the shortcomings of other methodologies. To begin,
corporate performance is assessed on the basis of publically available and audited data thereby eliminating biases that may exist
in self-reported data. Also, with surveys it is not clear if the way
respondents interpret survey questions inuences their responses.
Another limitation associated with survey research is that it is
assumed that respondents have the knowledge to answer items
when in fact they may not. With interviews, the quality of the
data and its interpretation are highly dependent on the skills of
the interviewer.
4.2. Time period of analysis
To investigate the impact of Six Sigma on corporate performance, a study period of ten years was employed. This ten year
period consisted of the event year or year the company adopted Six
Sigma, a three year pre-implementation period prior to the event
year, and a six year post-implementation period following the
event year. Including the three-year pre-implementation period
permits investigating whether there are performance differences
between the sample Six Sigma rms that may have inuenced
their implementing Six Sigma in the rst place and also investigating the relationship between pre-implementation performance
and post-implementation performance. Another important reason
for assessing performance pre-implementation is so performance
matched control groups can be constructed to ensure that our test
statistics are well-specied (Barber and Lyon, 1996).
The six year post-implementation period is chosen to ensure
that the Six Sigma rms were given adequate time to realize
the benets of adopting Six Sigma. While there is little research
addressing the lag between adopting Six Sigma and the realization of benets from doing so, the research on TQM suggests a two

to three year lag period. For example, the study by the US GAO
found an average lag of approximately 2.5 years from the time
when the companies initiated their focus on quality until the performance improvements became evident (GAO, 1991). Easton and
Jarrell (1998) found no statistically signicant effects in terms of
unexpected performance for NI/S, NI/A, and OI/S one and two years
after the event year in their study of TQM and corporate performance. However, the unexpected average performance over years
three to ve on these three variables was statistically signicant.
In the Easton and Jarrell (1998) study, the lag in improved performance is even more understated since the event year was dened
as six months after the rst major TQM initiative.
For the purpose of this study, scal years are converted into
event years in order to pool observations over time. By convention,
event year 0 corresponds to the scal year a given sample rm
adopted Six Sigma, event year 1 corresponds to the year preceding
the year Six Sigma was adopted, and event year +1 corresponds to
the year following the year Six Sigma was adopted.
4.3. Assessing corporate performance
In an ideal world the impact of implementing Six Sigma on
a rms performance would be assessed by comparing how the
company performed both with and without Six Sigma. Unfortunately making this type of assessment is not possible and therefore
assessing the impact of Six Sigma on a rms performance requires
the identication of relevant performance benchmarks. We evaluate a variety of different benchmarks to ensure that the benchmark
choice is not driving the results.
At one end of the spectrum of benchmarks, we take a nave viewpoint and use an industry adjusted performance of our sample Six
Sigma rms. We calculate this by subtracting the industry median
performance from the performance of our sample rms. Industry median performance is used because of the non-normality of
the nancial data. Industry medians are created by rst computing
the 4-digit, 3-digit, 2-digit, and 1-digit SIC industry classications
medians excluding sample rms. Then the sample rm SIC code is
matched to the most detailed industry level median which has at
least ve other companies in the industry classication.2
There are several benets associated with using industry
adjusted performance. First, because organizations in different
industries face unique challenges and market conditions, adjusting a rms performance relative to its industrys performance
permits comparisons across industries. Second, assessing a rms
performance relative to its industry provides an evaluation of the
company relative to a large sample of its competitors. Third, industry adjusted performance is a very nave measure which relies
upon minimal assumptions. Across all performance measures in
year 1, the average number of rms included in a given Six
Sigma sample rms industry was 61.4 with a range of six to
590.
At the other end of the continuum, we also compare a sample Six Sigma rms performance to the performance of the closest
matched rm and a portfolio of control rms matched to it on the
basis of industry, year, and similar past performance. Because the
results were similar with either the best match or the portfolio of
matching rms, we report the results of the portfolio of matching
rms.
Matching control rms on the basis of similar pre-event performance helps on several dimensions (Barber and Lyon, 1996).
First, it helps to control for the endogeneity problem because the
level of performance may be due to managerial ability, rm-specic

We also match on 4-digit only and obtain similar results.

S.M. Shafer, S.B. Moeller / Journal of Operations Management 30 (2012) 521532

527

Table 3
Six Sigma rms median industry adjusted performance by event year.
Event year

OI/A

N
3
2
1
0
+1
+2
+3
+4
+5
+6
*
**
***

76
77
79
82
83
83
83
83
82
81

OI/S

Median
***

0.0290
0.0258***
0.0258***
0.0264***
0.0353***
0.0391***
0.0380***
0.0313***
0.0278***
0.0269***

N
76
77
79
82
83
83
83
83
82
81

OI/E
($000/employee)
Median
***

0.0363
0.0471***
0.0487***
0.0442***
0.0449***
0.0404***
0.0418***
0.0329***
0.0346***
0.0327***

N
76
76
79
79
83
83
82
82
81
79

S/A

Median
***

11.3375
11.5158***
11.3365***
13.2159***
17.3888***
18.7008***
13.4381***
12.1594***
11.7895***
15.0333***

S/E
($000/employee)

Median

Median

77
78
80
83
84
84
84
84
82
81

0.0538
0.0429
0.0046
0.0240
0.0086
0.0121
0.0420*
0.0195
0.0303
0.0516

77
77
80
80
84
84
83
83
81
79

37.5602***
26.0943***
23.7185***
35.2630***
33.5807***
40.8392***
45.7888***
49.6187***
55.6561***
59.7682***

Signicant at 10% level, sign test for median two-sided tail.

Signicant at 5% level, sign test for median two-sided tail.
Signicant at 1% level, sign test for median two-sided tail.

choices or the set of investment opportunities. By matching on performance, a researcher can control for various factors, unrelated to
an event, that affect the operating performance of assets (Barber
and Lyon, 1996, p. 366).
Second, matching on pre-event performance eliminates the
reversion to the mean effects that are common in accounting
data. Finally, matching sample rms with control rms based on
performance is also important to help ensure that our test statistics are well specied. Barber and Lyon concluded that selecting
control rms on the basis of pre-event performance is the only
way to ensure that the test statistics are well specied and that
matching on pre-event performance is considerably more important than selecting control rms on the basis of industry and/or
size.
The portfolio of same 4-digit SIC industry and year control rms
were selected based on similar performance on the variable being
assessed in the year prior (event year 1) and four years prior
(event year 4) to the sample rm adopting Six Sigma.3 Because
the results were similar with either event year match, we focus
on the year 1 match. Similar performance was operationalized
as the control rms performance being within 10% of the sample
rms performance in event year 1.4 Barber and Lyon (1996) found
that using a 90110% performance lter yields well-specied test
statistics. Note that because performance was matched separately
for each performance measure, the composition of the portfolio of
performance matched control rms for a given sample rm could
vary across the ve performance measures. Across all performance
measures in year 1, the average number of rms included in a
given Six Sigma sample rms matched portfolio was 4.4 with a
range of 134.
Finally, for the purposes of this study, our benchmarks are a conservative measure of performance because it is possible that other
rms included in the benchmark may have also implemented Six
Sigma or other process improvement programs. Along these lines,
if adopting Six Sigma does indeed enhance organizational performance, then including other Six Sigma rms in the benchmark

3
We also matched on 4-digit, 3-digit, 2-digit and 1-digit industry and found
similar results.
4
Because some of our measures are a very small percentage, we expanded the
match to include anything within an absolute one percent if the ratio is between 10
and +10 percent. In other words, if the sample rms operating income to sales is
three percent, the matching rm ratio is between two and four percent. Not including these expanded matches reduces our sample size but it does not change the
results. In addition, we required control rms to have ve years of data, event year
1 through +3.

would enhance the overall benchmark performance and reduce the

sample Six Sigma rms adjusted performance.5
5. Empirical results
5.1. Six Sigma rms industry adjusted performance
Tables 3 and 4 summarize the industry adjusted performance
for the Six Sigma sample rms.
In Table 3 the industry adjusted performance is calculated by
subtracting the sample rms median industry performance from
its individual performance. Positive industry adjusted performance
indicates that the sample rm outperformed its industry. For example, the median OI/A of a sample Six Sigma rm in the year Six Sigma
was adopted (event year 0) was 0.0264 higher than its median
industry OI/A. Also note that the difference between the sample Six
Sigma rm and its industry was used as opposed to using the percentage change in order to avoid problems with having a negative
denominator which can occur when OI is used. The percent change
has no meaning when the denominator is negative and removing
sample rms with a negative OI could bias our results.
Because the sample rms median industry performance is
subtracted from its individual performance, industry adjusted performance is effectively a paired difference. Histograms of the
industry adjusted performance measures were non-normal and
not symmetric as is often the case with nancial data, and therefore the non-parametric Wilcoxon sign rank test was used to
test the hypotheses that the industry adjusted performance was
equal to zero. Statistically signicant results are reported on the
basis of two-tailed tests. The use of non-parametric tests in this
study is supported by Barber and Lyon (1996) who found that
non-parametric tests are more powerful than their parametric
counterparts.
As Table 3 illustrates, the Six Sigma sample rms outperformed
their respective industries on all of the performance variables
except S/A across all event years including both the years prior to
implementing Six Sigma and the years post Six Sigma implementation. Thus, rms that adopted Six Sigma, on average, performed
better than the industry prior to their announcement and they
maintained their signicantly better performance after adoption.
Specically, while they do not generate more sales per assets, they

5
Though the results are not reported, we also winsorized and trimmed the data
and found similar results. We winsorized at the 2.5 and 97.5 percentiles which sets
the observations below the 2.5 percentile and above the 97.5 percentile equal to the
values at the 2.5 and 97.5 percentiles, respectively. We similarly trimmed the data.

528

S.M. Shafer, S.B. Moeller / Journal of Operations Management 30 (2012) 521532

Table 4
Industry adjusted rate of improvement.
Time period

3 to 1
1 to +1
1 to +3
+1 to +3
+1 to +6
3 to +6
*
**
***

OI/A

OI/S

OI/E
($000/employee)

S/A

S/E
($000/employee)

Median

Median

Median

Median

Median

76
79
79
83
79
72

0.0097
0.0055
0.0099
0.0077
0.0075
0.0044

76
79
79
83
79
72

0.0041
0.0035
0.0026
0.0023
0.0045
0.0012

76
79
78
82
77
70

1.3027
0.5119
2.3733**
2.0001
2.5371
6.7310***

77
80
80
84
80
73

0.0099
0.0502*
0.0017
0.0074
0.0072
0.0511

77
80
79
83
78
71

8.8337
7.7962
13.6938**
14.4206**
20.0235***
22.9155***

Signicant at 10% level, sign test for median two-sided tail.

Signicant at 5% level, sign test for median two-sided tail.
Signicant at 1% level, sign test for median two-sided tail.

did generate more sales per employee and are more efcient at generating operating income relative to assets, sales and employees.
Furthermore, the Six Sigma rms performance advantage over
the industry generally increased in the short-term period immediately after implementing Six Sigma compared to the period prior
to Six Sigma implementation. For example, referring to Table 3, the
average industry adjusted S/E over the period 3 to 1 was 29.12.
In contrast, the average industry adjusted S/E over the periods +1
to +3 was 40.07 indicating a larger performance gap after implementing Six Sigma. Similar improvements in the performance gap
were also observed for OI/A and OI/E. Over the longer-term period
of +4 to +6, the Six Sigma rms continued to increase their performance advantage in terms of S/E. In particular, the average industry
adjusted S/E over the periods +4 to +6 increased to 55.01 (compared
to an average of 29.12 over 3 to 1 and 40.07 over +1 to +3).
In terms of OI/A and OI/E, the Six Sigma rms outperformed their
industry over the period +4 to +6 by a larger margin than during
the pre-implementation period of 3 to 1, but by a smaller margin than the short-term period of +1 to +3 immediately following
Six Sigma adoption.
In contrast to the static analysis summarized in Table 3, Table 4
provides a dynamic comparison between the rate of improvement
of the sample Six Sigma rms and the median industry rate of
improvement. This provides a direct test whether Six Sigma rms
improve performance after adoption. More specically, Table 4
summarizes the industry adjusted rate of improvement across various time periods by calculating the difference between the change
in a given sample Six Sigma rms performance over the time period
and the change in median industry performance over the same time
period. A positive industry adjusted rate of improvement indicates
that the sample Six Sigma rms rate of improvement was greater
than the overall industrys rate of improvement over the specied
time period.
As illustrated in Table 4, in the pre-implementation period
comprised of event years 3 to 1, the sample rms rate of
improvement was not statistically different with the industry on all
ve performance measures. So prior to the adoption of Six Sigma,
the sample rms were not on a different performance trajectory
than the industry. Thus, the adoption of Six Sigma was not motivated by a change in performance in the two years prior.
Once Six Sigma was adopted, we nd evidence that the sample
rms rate of improvement in utilizing employees was signicantly
better than the industry. Over event year 3 to +6 and 1 to +3
the sample Six Sigma rms outperformed their industries on both
employee productivity measures OI/E and S/E (see Table 4). Correspondingly, the sample Six Sigma rms rate of improvement in
terms of OI/A, OI/S and S/A was not statistically different from their
respective industries.
More specically, the sample Six Sigma rms rate of improvement was greatest relative to their rms respective industry
rate of improvement on S/E. The sample Six Sigma rms rate of

improvement on S/E was signicantly greater than their respective

industries rate of improvement in the short term periods 1 to +3
and +1 to +3 at a 5% level of signicance and longer term over the
periods +1 to +6 and 3 to +6 at a 1% level of signicance. The sample Six Sigma rms also had a greater rate of improvement on the
other employee productivity measure, OI/E, over the short term
period 1 to +3 at the 5% level of signicance and over the long
term period 3 to +6 at the 1% level of signicance. Differences in
the rate of improvement between the sample Six Sigma rms and
their respective industries were not observed for OI/A and OI/S.
Statistically signicant results in terms of the rate of improvement
of S/A were also generally not observed except for the period 1
to +1 where the sample Six Sigma rms rate of improvement was
greater at the 10% level of signicance.
Aside from S/A, we did not nd other signicant improvements
in the short run as measured by the 1 to +1 event year period.
This is consistent with the relatively long time it takes to roll out
Six Sigma programs.
So relative to their industry, we nd the Six Sigma rms are
better performers both prior to and after the adoption of Six Sigma.
In terms of the rate of improvement, the results suggest that the
adoption of Six Sigma signicantly improves the efciency with
which an organization deploys its employees but it does not affect
its efciency in deploying assets or operating income relative to
sales.

5.2. Six Sigma rms performance relative to matched portfolios

of control rms
Tables 5 and 6 summarize the performance for the Six Sigma
sample rms relative to the performance of matched portfolios of
control rms. In Table 5 the sample of Six Sigma rms adjusted performance is calculated by subtracting the median performance of
its matched portfolio of control rms from its performance. Positive
portfolio matched performance indicates that the sample Six Sigma
rm outperformed its matched portfolio of control rms. For example, the sample Six Sigma rms OI/A was a median 0.0014 higher
than their respective matched portfolio of control rms in the year
Six Sigma was adopted (event year 0). Like the industry adjusted
analysis, the difference in performance between the sample rm
and its matched portfolio of control rms was used as opposed to
calculating the percent change to avoid the negative denominator
problem.
Paired differences were created by subtracting the sample rms
matched portfolios performance from its performance. For the
same reasons as discussed for the industry adjusted data, the nonparametric sign test was used to test the hypotheses that the
differences between the sample rms and their portfolios of control
rms were equal to zero. Statistically signicant results are again
reported on the basis of two-tailed tests.

S.M. Shafer, S.B. Moeller / Journal of Operations Management 30 (2012) 521532

529

Table 5
Six Sigma rms median adjusted performance based on portfolio of matched control rms by event year.
Event year

OI/A

N
3
2
1
0
+1
+2
+3
+4
+5
+6
*
**
***

OI/S

Median

39
43
48
48
48
48
48
48
45
42

N
*

0.0131
0.0079
0.0003
0.0014
0.0125**
0.0114**
0.0036
0.0018
0.0144
0.0032

OI/E
($000/employee)
Median

39
41
46
46
46
46
46
46
42
39

0.0006
0.0012
0.0012
0.0023
0.0063
0.0101*
0.0097**
0.0121***
0.0130**
0.0279***

S/A

Median

31
33
37
37
37
37
37
36
34
32

2.1602
0.9961
0.3529
2.7288**
7.7801**
6.7384***
9.6990***
10.3645***
14.5282***
14.6827*

S/E
($000/employee)

Median

Median

40
43
46
46
46
46
46
44
40
39

0.0409
0.0317
0.0029
0.0023
0.0202
0.0675
0.0004
0.0090
0.0080
0.0254

44
46
48
48
48
48
48
46
43
41

15.9693*
13.1148***
0.5978
7.6650
19.0194***
26.7210***
25.6801***
35.8037***
29.8120**
35.6440**

Signicant at 10% level, sign test for median two-sided tail.

Signicant at 5% level, sign test for median two-sided tail.
Signicant at 1% level, sign test for median two-sided tail.

Table 6
Adjusted rate of improvement based on portfolio of matched control rms.
Time period

OI/A

Median

3 to 1
1 to +1
1 to +3
+1 to +3
+1 to +6
3 to +6

39
48
48
48
42
33

0.0144*
0.0140**
0.0037
0.0038
0.0145
0.0048

39
46
46
46
39
33

*
**
***

OI/S

OI/E
($000/employee)

S/A

S/E
($000/employee)

Median

Median

Median

Median

0.0031
0.0041
0.0117**
0.0140
0.0048
0.0342***

31
37
37
37
32
27

2.3375**
7.1443**
11.4627***
3.6265*
8.5194
11.2845

40
46
46
46
39
35

0.0272
0.0212
0.0057
0.0071
0.0030
0.0290

44
48
48
48
41
37

12.9774*
11.7228**
26.0909***
20.6390***
17.2024**
12.3567

Signicant at 10% level, sign test for median two-sided tail.

Signicant at 5% level, sign test for median two-sided tail.
Signicant at 1% level, sign test for median two-sided tail.

While statistically signicant differences were observed

between the performance of the sample Six Sigma rms and their
performance-matched portfolios of control rms in various event
years, as expected fewer statistically signicant results were found
in comparison to the industry adjusted results discussed earlier. For
example, the lack of signicant results in year 1 is expected due
to fact that control rms were selected based on their performance
relative to the sample rms performance in event year 1. Referring to the results in Table 5, prior to implementing Six Sigma, the
sample rms were generally at parity with their control rm portfolios while they outperformed their respective industries in the
industry adjusted analysis. Notable exceptions to this (see Table 5)
include slightly signicant results at the 10% level for OI/A, OI/E,
and S/E in year 3 and S/E at the 1% level in year 2. Likewise, in
the year of implementation (event year 0), the sample rms were at
parity on all performance measures except for OI/E at the 5% level.6
While generally at parity prior to adopting Six Sigma, over the
short-term the sample Six Sigma rms made gains on several of the
performance variables. For example, the sample Six Sigma rms
OI/A was signicantly higher at the 5% level in years +1 and +2.
Likewise, the sample rms had signicantly higher OI/S at the 10%
level in year +2 and in year +3 at the 5% level. In terms of OI/E,
the sample Six Sigma rms outperformed their matched portfolio
in year +1 at the 5% level and in years +2 and +3 at the 1% level.
Finally, the sample Six Sigma rms outperformed their matched
portfolios on S/E in years +1, +2, and +3 at the 1% level.

6
In unreported analysis, we matched on year 4, rather than year 1, performance and though there were some minor differences in the pre-adoption period,
the post-adoption results and rates of improvement results are unchanged.

Turning to long-term performance, the sample Six Sigma rms

did not outperform their matched portfolio on either asset productivity measure, namely, OI/A and S/A. Signicant results for S/A
were also not found in the industry adjusted analysis. The sample
Six Sigma rms outperformed their matched portfolio on OI/S at the
1% level in years +4 and +6 and in year +5 at the 5% level. Similar to
the industry adjusted analysis, the sample Six Sigma rms also outperformed their matched portfolios on both employee productivity
measures (OI/E and S/E) in all years +4 through +6.
Overall, though the results are generally a bit weaker, similar to
the industry adjusted analysis we nd rms that adopt Six Sigma
have higher performance in terms of employee productivity. Furthermore, in both the industry adjusted analysis and the matched
portfolio analysis, the rms that adopt Six Sigma did not outperform their respective benchmarks in terms of their sales to asset
efciency. Finally, while the Six Sigma rms outperformed their
industry in all years investigated in terms of OI/S, evidence of better
performance relative to their matched portfolios was not observed
until event year +2 with the results generally becoming more signicant as additional experience was gained with Six Sigma.
In contrast to the static analysis summarized in Table 5 comparing the performance of the sample rms to matched portfolios
of control rms at particular points in time, Table 6 provides a
dynamic comparison between the rates of improvement of the
sample rms and their respective matched portfolios of control
rms. In particular, Table 6 summarizes the portfolio adjusted
rate of improvement across alternative time periods by calculating the difference between the change in a sample Six Sigma rms
performance over the time period and the median change in its
portfolios performance over the same time period. A positive portfolio adjusted rate of improvement indicates that the sample Six
Sigma rms rate of improvement was greater than the rate of

530

S.M. Shafer, S.B. Moeller / Journal of Operations Management 30 (2012) 521532

improvement of their matched portfolios of control rms over the

specied time period.
As illustrated in Table 6, in the pre-implementation period consisting of event years 3 to 1, the sample Six Sigma rms rate of
improvement is signicantly lower than the rate of improvement
of their respective matched portfolios in terms of OI/A, OI/E, and
S/E. Since Table 5 shows that the sample rms outperformed the
portfolio in event year 3 on these measures, the previously well
performing sample rms were experiencing a decline in their relative performance in the two years prior to adopting Six Sigma.
In addition, for only these three performance measures there is a
quick signicant improvement in performance from the 1 to +1
period. So for sample rms that outperformed the portfolio in year
3, they experience a signicant decline in performance prior to
Six Sigma then a similar quick rebound upon adopting Six Sigma.
This result raises an interesting question regarding to the extent
to which contextual factors may inuence the lag between adopting Six Sigma and its impact on organizational performance. In the
present case, rms that outperformed their matched portfolio in
event year 3 entered a downward trajectory and then experienced a quick bump in performance immediately after adopting
Six Sigma. This evidence is consistent with better performing rms
having an advantage when it comes to reaping the benets of
adopting Six Sigma and/or better performing rms react faster at
the rst signs of deteriorating performance.
The signicantly higher rate of improvement for Six Sigma rms
relative to the matched portfolio for operating income and sales
per employee are generally consistent with the industry adjusted
results. Specically, the sample Six Sigma rms had a greater rate
of improvement relative to their matched portfolio in years 1 to
+3 on OI/E and S/E at the 1% level and the +1 to +6 period for S/E at
the 5% level.
In contrast to the industry adjusted results, we also nd signicant improvement in operating income to sales. Specically, in
the 1 to +3 period OI/S is signicantly higher at the 5% level. Also,
over the entire ten year period studied from event years 3 to +6 the
sample rms rate of improvement exceeded the matched portfolio
rate of improvement on OI/S at the 1% level.
5.3. Results summary
The purpose of this research was to investigate the impact the
adoption of Six Sigma has on corporate performance. The results of
the study indicate Six Sigma positively impacts organizational performance primarily through the efciency with which employees
are deployed. The benets of adopting Six Sigma were observed in
both the static analysis and the analysis of the Six Sigma rms rate
of improvement when the sample rms were compared to both
their industries and a portfolio of performance matched rms.
The results of the static analysis provide evidence that rms
that adopt Six Sigma are strong performers. The clearest evidence of this was observed in the industry adjusted analysis where
the sample Six Sigma rms outperformed their respective industries on all performance measures except S/A in all years prior to
adopting Six Sigma and in all years studied post adoption. Furthermore, while the results were not as strong, the sample Six Sigma
rms also outperformed their matched portfolios of control rms.
More specically, while the Six Sigma rms tended to be at parity with their matched portfolio prior to adopting Six Sigma, they
outperformed their matched portfolio in the years following implementation on OI/A (event years +1 and +2), OI/S (event years +2
through +6), OI/E (event years +1 through +6), and S/E (event years
+1 through +6). Again, there was no difference in S/A.
In contrast to the static analysis, in the dynamic analysis of
the Six Sigma rms rate of improvement, the strongest evidence
that Six Sigma rms achieve a greater rate of improvement was

observed in comparison to their matched portfolios of control rms.

One of the most interesting results observed was that the Six Sigma
rms outperformed their matched portfolios in year 3 in terms of
OI/A, OI/E, and S/E, then experienced a signicant decline in performance prior to Six Sigma on these three measures, and nally
exhibited a quick rebound upon adopting Six Sigma. This evidence
is consistent with previously better performing rms having an
advantage when it comes to reaping the benets of adopting Six
Sigma and/or better performing rms reacting faster at the rst
signs of deteriorating performance.
We also found evidence that the sample rms rate of improvement in terms of employee productivity was signicantly better
than their industries and matched portfolios. The sample Six Sigma
rms rate of improvement relative to the benchmarks performance was most persistent in terms of S/E where statistically
signicant results were found over the periods 1 to +3, +1 to +3,
and +1 to +6 in both the industry adjusted and matched portfolio
analysis. Statistically signicant results were also found for OI/E in
both the industry adjusted and matched portfolio analysis over the
period 1 to +3. Thus, the results of this study indicate that Six
Sigmas greatest impact was on the efciency with which employees are deployed. In both the industry adjusted analysis and the
performance-matched analysis, the Six Sigma rms outperformed
their respective benchmark groups on both OI/E and S/E in all
years after implementing Six Sigma. Furthermore, the performance
advantage for the Six Sigma rms on both employee productivity
measures tended to be larger after adopting Six Sigma and tended
to increase as additional experience was gained with Six Sigma.
Other results of the study further support increasing benets
from Six Sigma as experience is gained. In the performancematched portfolio analysis, there was no difference between the
sample rms and their portfolio of rms on OI/S until event year
+2 with the results generally becoming more statistically signicant (i.e., lower p-values) as additional experience was gained with
Six Sigma. As another example, the rate of improvement increased
as additional experience was gained with Six Sigma on S/E in the
industry adjusted analysis (see Table 4).
In total, the results suggest that better performing rms adopt
Six Sigma and that they continue their performance advantage after
adopting Six Sigma. Furthermore, the results suggest that Six Sigma
has its greatest impact on employee productivity and not on asset
productivity. Finally, there was no evidence that adopting Six Sigma
negatively impacts corporate performance.

6. Limitations and future research directions

6.1. Limitations
As an empirical research study, it is appropriate to highlight and
discuss the key limitations associated with the study. First, it should
be noted that the source of the data for the study was Compustat.
While Compustat data is a widely used and highly regarded source
of data, errors in the data and missing data items could have an
impact on the results obtained in the study.
Second, the focus of our study was on rms that publically
announced or have received other publicity about their adoption
of Six Sigma. While such publicity could serve as a proxy for the
sample rms commitment to Six Sigma, no attempt was made
to assess the effectiveness to which the sample Six Sigma rms
implemented Six Sigma. This is in contrast to several of the TQM
studies where the sample rms only consisted of quality award
winners as a proxy for effective TQM implementation. Thus, any
statistically signicant results regarding superior performance of
the sample Six Sigma rms are conservative to the extent that the
sample included rms that did not implement Six Sigma effectively.

S.M. Shafer, S.B. Moeller / Journal of Operations Management 30 (2012) 521532

In particular, observing statistically signicant differences between

the Six Sigma sample rms and their industries or their matched
portfolios becomes more difcult as the percentage of sample Six
Sigma rms that effectively implemented Six Sigma decreases.
Third, our sample of Six Sigma rms only included rms that
publically announced or received other publicity about their adoption of Six Sigma. Thus, it is possible that these rms are not
representative of all rms that have adopted Six Sigma.
Fourth, all empirical research on operational systems encounters the challenge of nding direct measures for performance
outcomes. Our choice of performance measures was driven by our
theoretical framework and the measures used in past research.
Fifth, though we did analyze prior performance trends, we did
not explicitly control for a rms motivation for adopting Six Sigma.
There are clearly a variety of endogenous factors, including motivation that could inuence why a rm chooses to adopt Six Sigma
and that also impact performance. For instance, the adoption of
Six Sigma may be motivated by recent poor nancial performance,
as a complement to a successful TQM implementation, or perhaps
because rms with better management tend to adopt Six Sigma.
The issue of endogeneity is a serious issue but any study of this type
suffers from the same problem. In this paper we use many methods
to attempt to manage endogeneity in order to isolate Six Sigma as
the important factor and we are careful to appropriately interpret
our results. However, as with all empirical research, it is a balancing act. While knowing a rms motivation for adopting Six Sigma
might be insightful, we are not concerned that not controlling for
this biased our results based on the performance matching research
of Barber and Lyon (1996).
Finally, as with all studies of this type, adopting Six Sigma may
serve as a proxy for unobservable rm characteristics. As such, this
type of research is not prescriptive and does not imply that merely
adopting Six Sigma guarantees improved rm performance.

6.2. Future research issues

As an early research study addressing the relationship between
Six Sigma and corporate performance, there are numerous avenues
for extending this research that would contribute to our understanding of the impact Six Sigma has on corporate performance.
To begin, research is needed that investigates rm characteristics
such as rm size, capital intensity, degree of diversication, industry, and the maturity of Six Sigma implementation. For example,
do larger rms have an advantage in deploying Six Sigma because
it is easier for them to absorb the training cost and dedicating
full-time resources to process improvement activities? Likewise,
given the evidence from the study that Six Sigma facilitates the
deployment of employees, are highly capital intensive rms able
to reap benets similar in magnitude to less capital intensive
rms? Lastly, are there differences in performance between various
industries?
Research is also needed to investigate the relationship between
how Six Sigma was implemented and corporate performance. For
example, in this study, the extent to which Six Sigma was implemented throughout the rm was not controlled for. Logically we
would not expect to observe the same magnitude of benets for
rms that adopted Six Sigma across the entire enterprise versus
others that implemented it more narrowly in a limited number of
divisions, geographies, or plants. Another implementation issue in
need of additional research relates to whether there are differences
in performance between early and late adopters of Six Sigma. In a
related vein, does the previous adoption of other process improvement initiatives such as TQM, ISO 9000, and/or lean moderate the
relationship between the adoption of Six Sigma and corporate performance? Is there a difference in performance between rms that

531

have integrated their Six Sigma initiatives with lean versus rms
that have kept them separate or only adopted Six Sigma?
Related to these research issues, another important area of
research is to investigate the relationship between an organizations motivation for adopting Six Sigma and corporate
performance. For example, can patterns be observed suggesting
that some rms adopted Six Sigma during periods when they were
outperforming the industry while other rms adopted Six Sigma
during periods when they were underperforming the industry? If
so, is there a difference in performance between the more proactive
rms versus the more reactive rms? And even more fundamentally, is there a difference between the types of rms that benet
from adopting Six Sigma versus those that do not? Addressing these
research issues would provide important insights to managers and
researchers on the relationship between Six Sigma and corporate
performance.

References
Adam Jr., E.E., 1994. Alternative quality improvement practices and organization
performance. Journal of Operations Management 12 (1), 2744.
Antony, J., 2004. Some pros and cons of Six Sigma: an academic perspective. TQM
Magazine 16 (4), 303306.
Arndt, M., 2002. Quality isnt just for widgets. Business Week, 72 (July 22).
Barber, B.M., Lyon, J.D., 1996. Detecting abnormal operating performance: the empirical power and specication of test statistics. Journal of Financial Economics 41,
359399.
Balakrishnan, R., Linsmeier, T.J., Venkatachalam, M., 1996. Financial benets from JIT
adoption: effects of customer concentration and cost structure. The Accounting
Review 71 (2), 183205.
Benitez, Y., Forrester, L., Hurst, C., Turpin, D., 2007. Hospital reduces medication
errors using DMAIC and QFD. Quality Progress, 3845, January.
Bossidy, L.A., Bonsignore, M.R., 1999. To our shareonwers. Honeywell 1999 Annual
Report.
Chakravorty, S.S., 2010. Where process-improvement projects go wrong. The Wall
Street Journal, 25, January.
Cua, K.O., McKone, K.E., Schroeder, R.G., 2001. Relationships between implementation of TQM, JIT, and TPM and manufacturing performance. Journal of Operations
Management 19 (6), 675694.
Corbett, C.J., Montes-Sancho, M.J., Kirsch, D.A., 2005. The nancial impact of ISO 9000
certication in the United States: an empirical analysis. Management Science 51
(7), 10461059.
Craven, E.D., Clark, J., Cramer, M., Corwin, S.J., Cooper, M.R., 2006. New
YorkPresbyterian Hospital uses six sigma to build culture of quality and innovation. Journal of Organizational Excellence, 1119, Autumn.
Daniels, S.E., 2009. Contacts that count: project improves member call rate, nets $3.3
million in savings. Quality Progress, 4247, January.
Deshpande, D.P., Halder, S., Choudhary, A., Raychaudhuri, A., Biswas, S., 2004. DMAIC
approach brings breakthrough results. Six Sigma Forum Magazine, 2429,
February.
Douglas, T.J., Judge Jr., W.Q., 2001. Total quality management implementation and
competitive advantage: the role of structural control and exploration. Journal of
Operations Management 44 (1), 158169.
Dudman, L., 2005. Big improvements for small parts. Quality Progress, 6772,
December.
Easton, G.S., Jarrell, S.L., 1998. The effects of total quality management on corporate
performance: an empirical investigation. Journal of Business 71 (2), 253307.
Eriksson, H., Hansson, J., 2003. The impact of TQM on nancial performance. Measuring Business Excellence 7 (1), 3650.
Foster Jr., S.T., 2007. Does Six Sigma improve performance? Quality Management
Journal 14 (4), 720.
Fullerton, R.R., McWatters, C.S., Fawson, C., 2003. An examination of the relationship
between JIT and nancial performance. Journal of Operations Management 21
(4), 383404.
GAO, 1991. Management practices. 1991. U.S. companies improve performance
through quality efforts. Report to the Honorable Donald Ritter, House of Representatives GAO/NSIAD-91-190.
Goh, T.N., Low, P.C., Tsui, K.L., Xie, M., 2003. The impact of Six Sigma implementation
on stock price performance. TQM and Business Excellence 14 (7), 753763.
Hahn, G.J., Hill, W.J., Hoerl, R.W., Zinkgraf, S.A., 1999. The impact of Six Sigma
Improvementa glimpse into the future of statistics. The American Statistician
53 (3), 208215.
Handeld, R., Ghosh, S., Fawcett, S., 1998. Quality-driven change and its effects on
nancial performance. Quality Management Journal 5 (3), 1330.
Hendricks, K.B., Singhal, V.R., 1997. Does implementing an effective TQM program
actually improve operating performance? Empirical evidence from rms that
have won quality awards. Management Science 43 (9), 12581274.
Hendricks, K.B., Singhal, V.R., 2001a. The long-run stock price performance of Firms
with effective TQM Programs. Management Science 47 (3), 359368.

532

S.M. Shafer, S.B. Moeller / Journal of Operations Management 30 (2012) 521532

Hendricks, K.B., Singhal, V.R., 2001b. Firm characteristics, total quality management,
and nancial performance. Journal of Operations Management 19 (3), 269285.
Huson, M., Nanda, D., 1995. The impact of just-in-time manufacturing on rm performance in the US. Journal of Operations Management 12 (34), 297310.
Johnson, K., 2005. Six Sigma delivers on-time service. Quality Progress, 5759,
December.
Jones Jr., M.H., 2004. Six Sigma. . .at a Bank? Six Sigma Forum Magazine, 1317,
February.
Kaynak, K., 2003. The relationship between total quality management practices
and their effects on rm performance. Journal of Operations Management 21,
405435.
Kinney, M.R., Wempe, W.F., 2002. Further evidence of the extent and origins of JITs
protability effects. The Accounting Review 77 (1), 203225.
Lee, K.-C., Choi, B., 2006. Six Sigma management activities and their inuence on
corporate competitiveness. Total Quality Management 17 (7), 893911.
Linderman, K., Schroeder, R.G., Zaheer, S., Choo, A.S., 2003. Six Sigma: a goal-theoretic
perspective. Journal of Operations Management 21 (2), 193203.
Motorola, 2011. Motorola Six Sigma business improvement programs.
<http://www.motorola.com/web/Business/ Moto University/ Documents/
Static Files/Business Improvement Campaigns.pdf> (accessed 15.07.11).
Mukherjee, S., 2008. A dose of DMAIC: hospitals six sigma and lean efforts benet
patients and protability. Quality Progress, 4451, August.
Nair, A., 2006. Meta-analysis of the relationship between quality management
practices and rm performanceimplications for quality management theory
development. Journal of Operations Management 24, 948975.

Pande, P.S., Neuman, R.R., Cavanagh, R.R., 2000. The Six Sigma Way. McGraw-Hill,
New York, NY.
Powell, T.C., 1995. Total quality management as competitive advantage: a review
and empirical study. Strategic Management Journal 16, 1537.
Samson, D., Terziovksi, M., 1999. The relationship between total quality management practices and operational performance. Journal of Operations
Management 17, 393409.
Schroeder, R.G., Linderman, K., Liedtke, C., Choo, A.S., 2008. Six Sigma: definition and underlying theory. Journal of Operations Management 26,
536554.
Shah, R., Ward, P.T., 2003. Lean manufacturing: context, practice bundles, and performance. Journal of Operations Management 21 (2), 129149.
Singels, J., Ruel, G., van de Water, H., 2001. ISO 9000 series certication and performance. International Journal of Quality and Reliability Management 18 (1),
6275.
Terziovski, M., Samson, D., Dow, D., 1997. The business value of quality management systems certication evidence from Australia and New Zealand. Journal of
Operations Management 15 (1), 118.
Terziovski, M., Power, D., Sohal, A.S., 2003. The longitudinal effects of the ISO 9000
certication process on business performance. European Journal of Operational
Research 146, 580595.
Zu, X., Fredendall, L.D., Douglas, T.J., 2008. The evolving theory of quality management: the role of six sigma. Journal of Operations Management 26,
630650.