Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Changing fundamental analysis in the new economy: the case of DuPont analysis and STEM firms
(USC Thesis Other)
Changing fundamental analysis in the new economy: the case of DuPont analysis and STEM firms
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
CHANGING FUNDAMENTAL ANALYSIS IN THE NEW ECONOMY:
THE CASE OF DUPONT ANALYSIS AND STEM FIRMS
by
Suteera Pongtepupathum
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirement for the Degree
DOCTOR OF PHILOSOPHY
BUSINESS ADMINISTRATION
August 2021
Copyright 2021 Suteera Pongtepupathum
ii
Acknowledgments
The path toward this dissertation has been up and down. Its completion is thanks to the special people who
support me in every way. First, I would like to express the deepest appreciation to my committee chair,
Professor Mark Soliman, who gives me not only his valuable insights on my dissertation but also his
emotional support throughout my Ph.D. journey. Without his kindness, guidance, and supports, completing
my dissertation would be just a dream. I also would like to thank my committee members, Associate
Professor Maria Ogneva and Associate Professor Kenneth Ahern, for their guidance and encouragement on
this project.
I am extremely grateful to my beloved twin sister, Wanida Pongtepupathum, for her unconditional love and
caring. I would definitely not be able to complete this 5-year long study without her listening to me and
understanding me in good times and bad times. I would like to express my gratitude toward my parents for
their wholehearted love and supports. I am very much thankful to my husband and brother for their
continuing support of my Ph.D. study. I would like to thank my amazing Ph.D. friends at Marshall School
of Business, especially Seun Yeon (Sunny) Yoo, Vivek Panday, Jung Koo Kan, and Ryan Erhard, for their
friendship and the great moments we had here. Lastly, I would like to thank the University of Southern
California for the financial, academic, and technical supports of my Ph.D. study.
iii
Table of Contents
Acknowledgments ......................................................................................................................................... ii
List of Tables ............................................................................................................................................... iv
List of Figures ............................................................................................................................................... v
Abstract ........................................................................................................................................................ vi
Chapter 1: Introduction ................................................................................................................................. 1
Chapter 2: Related literature and hypothesis development ........................................................................... 7
2.1 A changing structural nature of the economy and intangible capital .................................................. 7
2.2 DuPont analysis .................................................................................................................................. 7
2.3 Theory and hypotheses ....................................................................................................................... 9
Chapter 3: Sample and descriptive statistics ............................................................................................... 13
3.1 Variable measurement ...................................................................................................................... 13
3.2 Sample data ....................................................................................................................................... 14
3.3 Descriptive statistics ......................................................................................................................... 15
Chapter 4: Empirical analysis ..................................................................................................................... 18
4.1 Replication of prior research with an extended period ..................................................................... 18
4.2 STEM and non-STEM firms ............................................................................................................. 19
4.3 Intangible intensity ............................................................................................................................ 21
4.3 Before and after 2000 ........................................................................................................................ 22
4.4 DuPont analysis and industry concentration ..................................................................................... 23
4.5 DuPont analysis and product similarity ............................................................................................ 25
Chapter 5: Robustness check ...................................................................................................................... 27
5.1 Different measures of intangible intensity ........................................................................................ 27
5.2 Results using the Fama-MacBeth approach ...................................................................................... 28
Chapter 6: Conclusion ................................................................................................................................. 29
References ................................................................................................................................................... 31
iv
List of Tables
Table 1: Descriptive statistics ..................................................................................................................... 34
Table 2: Pairwise correlation table.............................................................. Error! Bookmark not defined.
Table 3: Time-series means and t-statistics for coefficients from annual cross-sectional regressions of
future one-year-ahead change in RNOA (RNOA_CHF) on the DuPont components ................................ 38
Table 4: STEM firms, the DuPont components, and future profitability .................................................... 40
Table 5: Intangible intensity, the DuPont components, and future profitability ......................................... 44
Table 6: The DuPont components and intangible intensity before and after 2000 ..................................... 46
Table 7: Industry concentration and the DuPont components .................................................................... 48
Table 8: Product Similarity and the DuPont components ........................................................................... 48
Table 9: Intangible intensity and the relation between the DuPont components and future profitability ... 49
Table 10: The DuPont components and intangible intensity before and after 2000 ................................... 50
Table 11: The DuPont component and intangible intensity using Fama-MacBeth regressions ................. 51
v
List of Figures
Figure 1: The mean intangible intensity during 1984-2016 for all firms, STEM and, non-STEM firms with
3 measures of intangible intensity ............................................................................................................... 33
Figure 2: The mean intangible intensity during 1984-2016 with 3 measures of intangible intensity, within
STEM industries including technology, science and internet firms ........................................................... 33
vi
Abstract
Prior research on DuPont analysis found that changes in asset turnover predicted future profitability
whereas profit margin did not. This was attributed to the structural nature of the U.S. economy as focused
primarily on manufacturing and asset turnover’s ability to capture physical asset utilization efficiency. The
U.S. economy has shifted toward service-and-technology-based industries over the last few years with the
surge of STEM (science, technology, engineering, and math) firms. Accordingly, I re-examine the
usefulness of DuPont analysis to predict future changes in profitability and whether intangible intensity
attributes to the change in the property of DuPont ratio. I find that changes in profit margin are now
positively associated with future changes in profitability, consistent with the notion that intangible capital
creates barriers to entry and production differentiation. Further, this relation is stronger for STEM firms
than non-STEM firms and also more pronounced for high intangible intensity firms than low intangible
intensity firms but only in industries with low concentration and high product similarity environment. Taken
together, these results highlight how changes in the structural makeup of the economy can change how
financial statement analysis is applied and adds to the literature in fundamental analysis.
Keywords: intangible intensity; STEM; fundamental analysis; forecasting; profitability; DuPont analysis
Data availability: all data are available from public sources identified in the paper
1
Chapter 1: Introduction
The nature of the U.S. economy has shifted drastically over the last few decades. In the early late 1990,
industrial and natural resource firms dominated the capital market. In 1996, companies like General
Electric, the Coca-Cola Company, and Exxon Mobile were among the top five largest firms by market
capitalization
1
. However, since then, firms’ composition has shifted to service-and-technology-based
industries as we enter into the information age. In 2000, technology companies such as Microsoft and Cisco
joined the list of the top five largest firms
2
. This shift not only highlights the increasing importance of
intangible capital such as innovative products, human capital, branding, patents, copyrights, software
development, database, etc. in driving the economy in general, but that intangible capital become even more
important to knowledge-and-technology-based companies than traditional industrial companies as their
primary sources of competitive advantage. Within this shifting landscape, the usefulness of traditional
approaches to financial statement analysis should be explored. DuPont analysis lends itself to this question
because it is an approach that relies on notions of capital intensity, competitive advantage and pricing power
and stands to change in the information age. Accordingly, I re-examine DuPont analysis in predicting firms’
changes in future profitability and whether intangible investments drive the change in the property of
DuPont ratios as the economy has moved away from a physical-capital-based economy to an intangible-
capital-based economy.
DuPont analysis provides a framework for forecasting future profitability by decomposing return
on net operating assets (RNOA) into asset turnover (ATO), which measures (mostly physical) asset
utilization and efficiency, and profit margin (PM), which measures pricing power. Prior research on DuPont
analysis documents that changes in ATO have predictive power with respect to future changes in RNOA
(Fairfield and Yohn, 2001; Nissim and Penman, 2001; Soliman, 2008), and information about changes in
ATO provides incremental information about firm’s operating characteristics for market participants
1
https://en.wikipedia.org/wiki/List_of_public_corporations_by_market_capitalization#1996
2
As of July 1, 2020, the top five largest companies were Apple, Microsoft, Amazon, Alphabet, and Facebook, and all of them are
technology firms
2
(Soliman, 2008). However, prior studies do not document the same property for changes in PM and have
found it to be generally uninformative; which is strange given the importance of margins as a tool in the
analysis of firms. Research in the fundamental analysis points out that changes in ATO, capturing new
information about the improved asset usage in generating revenues, can predict the future change in RNOA
because the efficiency of asset usages is less threatened by competition as imitating another company’s
production process is costly. In contrast, Soliman (2008) conjectures the reason for this finding is that
knowledge is quickly diffused and prone to imitation from competitors, the returns based on knowledge-
based assets may be more transitory (Romer, 1986), and thus this may result in the uninformativeness of
changes in PM in predicting future changes in RNOA.
However, returns on intangible resources may not necessarily be diminishing or transitory. In fact,
intangible capital may help increase profit or delay diminishing returns because such capital create barriers
to entry and are crucial to extracting willingness-to-pay from customers. The resource-based view of a firm
in the strategy literature argues that intangible assets (e.g., brands, trade-marks, etc.) are a crucial driver of
product differentiation or value-creating strategy in creating sustained competitive advantage (e.g.,
Lippman and Rumelt, 1982; Rumelt, 1984; Rumelt, 1991; Wernerfelt, 1984; Dierickx and Cool, 1989;
Barney 1991) because many intangible resources are internally developed (Dierickx and Cool 1989) and
thus make it difficult for rival firms to imitate. For example, organization capital is viewed as a collection
of technologies-such as unique business processes, practices, and compensation systems- that enable firms
to achieve an abnormal value of products through the efficiency of use (Lev and Radhakrishnan, 2005).
Organization capital is a valuable resource in sustaining a firm’s competitive advantages because it is firm-
specific and cannot be easily transferred to other organizations. Prior research has found that firms using
more organization capital are more productive after accounting for physical capital and labor (Eisfeldt and
Papanikolaou, 2013) and have better future operating performance measured by sales growth and operating
3
income growth (Lev, Radhakrishnan, and Zhang, 2009). Thus, organization capital should affect profit
margin and profitability and it is somewhat surprising that prior research fails to document this
3
.
To study the relation between the DuPont components and their predictive power for future changes
in profitability as the economy has changed toward intangible capital, I begin with a replication of prior
studies (Fairfield and Yohn, 2001; Nissim and Penman, 2001; Soliman, 2008) over the larger sample period
of 1984- 2016. Consistent with previously documented findings, changes in ATO(PM) are(are not)
positively associated with future changes in RNOA over the same sample period of 1984-2002. However,
the property of changes in PM have changed as I extended the period to 2016
4
, I find that changes in PM
are predictive of one-year-ahead change in RNOA, contrary to the previously documented evidence. This
finding suggests that changes in PM brings new information about improved pricing power in predicting
future changes in profitability and offers prima facie evidence that the shift of the economy may contribute
to the change in the relation between the DuPont components and future changes in RNOA.
Next, to further explore my conjecture that the shifting face of the economy may be causing the
result documented above, I examine whether intangible capital influences the relation between the DuPont
components and future changes in RNOA. To do this, I investigate the association between changes in PM
and future changes in RNOA between both STEM (science, technology, engineering, and mathematics) and
non-STEM firms. I focus on STEM firms as a proxy for intangible intensity because they are essential and
growing drivers in the U.S. economy, and STEM firms’ characteristics differ from traditional
manufacturing and industrial firms’(Fedyk et al 2017). Unlike brick-and-mortar firms that heavily invest in
physical assets, STEM firms have spent a large portion of their annual budget on intangible investments
3
Knowledge capital is another essential intangible resource that can create sustained competitive advantages. Firms develop
knowledge capital by spending on R&D activities. Such activities are critical in implementing a value-creating strategy because
they generate innovative or high-quality products and services that can extract higher customers’ willingness to pay, helping
sustain a firm’s profit margins and future profitability. Knowledge capital can maintain competitive advantages because it takes
time to develop technology and create innovation. Besides, the outcomes of R&D activities such as patents, copyrights, and
trademarks also create barriers to entry and thus make it difficult for rivals to imitate or at least delay the imitation from rival
firms. Overall, intangible capital is key to creating sustained competitive advantages and allowing the firm to maintain profit
margin and future profitability.
4
The extended period ends in 2016 because Peters and Taylor’s total Q dataset used for calculating intangible intensity ends in
2017. After dropping missing values and all data requirements, most remaining observations end in 2016.
4
such as R&D and human capital because innovations and technology are their competitive advantages in
extracting economic rents. The distinction of large intangible investments likely affects firms’ business
operations and profitability. Because intangible capital helps maintain firms’ competitive advantages, firms
with high intangible intensity are likely to be able to maintain profit margins and profitability better than
those with low intangible intensity. Therefore, I anticipate that the association between changes in PM and
future changes in RNOA is stronger for STEM firms than non-STEM firms. Consistent with my prediction,
I find that the association between changes in PM and one-year-ahead changes in RNOA is more pronounced
for STEM firms than non-STEM firms. Further analysis shows that the association between changes in PM
and future changes in RNOA exists only for STEM firms but not for non-STEM firms. This finding suggests
that new information about intangible capital is much more useful in predicting future changes in
profitability for STEM firms but not non-STEM firms. Lastly, a sub-STEM group analysis shows that
changes in PM are positively associated with future changes in RNOA for science, technology, and internet
firms. In contrast, changes in ATO are predictive of changes in RNOA for science and technology firms but
not for internet firms, consistent with the view that new information about asset utilization is less important
for internet firms since the physical assets are not the key resource in generating revenue.
Further, I extend the STEM/Non-STEM classification into a more generic classification based on
the intensity of intangibles to investigate the relation between changes in PM and future changes in RNOA.
Following Peters and Taylor (2016), I measure intangible intensity as the ratio of intangible capital to total
capital. To measure intangible capital, I include both on-Balance-Sheet and off-Balance-Sheet intangible
assets by capitalizing knowledge capital and organizational capital using the perpetual inventory method. I
interpret R&D expenditures as knowledge capital and a fraction of selling, general, and administration
(SG&A) expenditures as organization capital. If intangible resources allow firms to sustain their
competitive advantages and contribute to firms’ ability to maintain their profit margin and profitability,
then I predict that firms with higher intangible intensity have a stronger association between changes in PM
5
and future changes in RNOA than those with lower intangible intensity. I find the evidence consistent with
this prediction.
Finally, I explore the role of the intangible intensity in DuPont analysis under different industry
concentration and product similarity space. I find that in low concentrated industries (those that have many
firms operating in them and likely have high competition), the relation between changes in PM and one-
year-ahead changes in RNOA is more pronounced for firms with higher intangible intensity (and is
insignificant in high industry concentrated industries). It’s consistent that firms operating in the low industry
concentration environment likely face high competition and heavily invest in intangibles to sustain their
competitive advantages over their competitors. However, in high concentrated industries, firms likely face
lower competition and thus the role of intangible investments becomes less critical. Further, I also find that
the association between changes in PM and future changes in profitability is more pronounced for firms
with higher intangible intensity than those with lower intangible intensity under a high product similarity
environment. But do not find this relation in a low product similarity environment. Consistent with the
theory that intangible investments can create endogenous barriers to entry (Sutton, 1991), the finding
indicates that firms invest in intangibles to differentiate their products to maintain competitiveness and,
thus, help their profit margin and future profitability.
This study contributes to the literature on fundamental analysis and the role that structural changes
in the economy have on how useful different approaches. My first finding shows that due to these changes,
PM can now predict future changes in RNOA although prior studies only found that changes in ATO are
predictive of future changes in RNOA(Fairfield and Yohn, 2001; Nissim and Penman, 2001; Soliman,
2008). I also document that the relation between changes in PM and future changes in RNOA is stronger
for high intangible intensity than low intangible intensity. This finding suggests that changes in the property
of DuPont ratio is attributed to increased intangible capital and highlights the important role of information
about intangible intensity in forecasting the persistence of ratios used in financial statement analysis.
6
This study also extends our understanding of how our economy has evolved by exploring how
STEM firms generate earnings differently than traditional firms. Previous studies about STEM firms mainly
focus on their valuation. For example, prior research documents that investors use information about R&D
expenditure in valuing biotech firms (Darrough and Ye, 2007; Guo et al., 2005; Joos and Zhdanov, 2008)
and about sales growth in valuing high growth firms (Demers and Lev, 2001; Trueman et al., 2000; Hand,
2001). Fedyk et al. (2017) documents that investors tend to weigh R&D and sales growth more than bottom-
line earnings in valuing STEM firms during the IPO and beings to explore financial statements through
different approaches of earnings management. This study extends our understanding of STEM firms by
documenting that the future changes in STEM firms’ profitability are positively associated with both
changes in ATO (improved asset utilization) and changes in PM (increased pricing power). However, future
changes in non-STEM firms’ profitability are positively associated with ATO changes but not with PM
changes. The difference in empirical results between STEM and non-STEM firms highlights the importance
of extensive investment in STEM firms’ intangible capital on their future profitability.
7
Chapter 2: Related literature and hypothesis development
2.1 A changing structural nature of the economy and intangible capital
We have witnessed the shift in the structural nature of the U.S. economy over the last few years. Before
2000, the U.S. economy generally focused on manufacturing. The dominant companies at that time were
industrial and natural resource companies like Exxon Mobile, The Coco-Cola Company, General Electric.
These firms heavily rely on physical assets, so asset utilization and efficiency are a crucial strategy in their
business operation and value creation. However, since then, the U.S. economy has moved toward
knowledge-and-technology-based sectors, and STEM firms are increasingly important and become a
growing driver of the economy as we enter into the digital age. For example, in early 2000, software and
networking hardware companies like Cisco and Microsoft were among the world's top five largest
companies by market capitalization. The shift in the economic landscape shows the increasing importance
of intangible capital in the economy, as depicted in Panel A of Figure 1, that all firms, on average, increase
their intangible capital from about 25 percent of their total capital in 1984 to almost 40 percent of their total
capital in 2016. The increase in intangible intensity primarily derives from STEM firms. Their intangible
capital increased from about 30 percent to over 50 percent of their total capital, as portrayed in Panel A of
Figure 1. With the changing nature of the economy, intangible capital is vital to study because such capital
is more crucial for knowledge-and-technology companies than traditional brick-and-mortar firms as
knowledge-based firms as their primary source of sustained competitive advantage.
2.2 DuPont analysis
The shifting economic landscape likely affects business operation and profitability, so the usefulness of
traditional tools to financial statement analysis should be investigated. DuPont analysis suits this question
because it is an approach that focuses on capital intensity, competitive advantage, and pricing power.
DuPont analysis provides a framework for firm valuation by assessing factors that affect financial
performance. Standard DuPont analysis decomposes return on equity (ROE) into three components,
including profit margin (PM), asset turnover (ATO), and equity multiplier (EM) as follows:
8
ROE= [NI/Sales X Sales/Assets X Asset/BVEquity]
PM represents the firm’s profitability relating to revenue. ATO captures the efficiency of assets in
generating sales. EM presents the firm’s capital structure. Because ROE can be affected by the firm’s
financial leverage, Nissim and Penman (2001) algebraically rearrange ROE to remove the effect of the
firm’s financial leverage and arrive as follows.
ROE= RNOA + [FLEV X SPREAD]
FLEV is financial leverage. SPREAD is the difference between borrowing costs and the return of
operations. RNOA is the return on the net operating assets. The new arrangement of ROE helps value firms
because it allows information users to focus on operating activities without the firm’s financial leverage.
Therefore, the above arrangement has become a standard for analyzing the firm’s profitability in the
valuation literature, and RNOA becomes a standard measure of the firm’s profitability from operating
activity.
RNOA can be decomposed into PM and ATO. These two measures capture different properties of a firm’s
operation. PM captures the firm’s ability to control costs in generating sales, convert sales into profit, and
give insight into product pricing. PM reflects pricing power, such as product innovation, brand name
recognition, first-mover advantage, and product differentiation, and thus are associated with intangible
capital. The change in PM (PM_CH) measures the growth rate in operating income relative to the growth
in sales. ATO captures the firm’s ability to utilize operating assets, such as working capital and property,
plant, and equipment. Therefore, ATO is often associated with physical capital rather than intangible capital.
Changes in ATO (ATO_CH) measures the growth rate in sales relative to growth in operating assets. As
these two measures capture different aspects of a firm’s business operation, prior studies in the fundamental
analysis literature investigated the usefulness of DuPont ratios in improving future profitability forecast.
Prior research documents that the decomposition of DuPont analysis is useful in forecasting future
profitability. Fairfield and Yohn (2001) find that changes in ATO are positively associated with the future
9
change in RNOA, whereas changes in PM are not. Soliman (2008) examines the association between the
DuPont components and future RNOA after controlling for fundamental signals and reaches a similar
conclusion (changes in ATO is positively associated with one-year-ahead future changes in RNOA).
Soliman (2008) also investigates the use of DuPont analysis by analysts and investors and documents that
changes in ATO are positively associated with the contemporaneous return, future abnormal returns, and
analyst revision. The findings suggest that information about changes in DuPont ratios is not fully processed
by market participants and provides incremental information about the firm’s operating characteristics.
Change et al. (2014) further investigate the usefulness of DuPont ratios in highly regulated the U.S. health-
care industry and find that, contrary to prior studies, changes in PM are positively associated with future
changes in RNOA, whereas changes in asset turnover are not. However, Change et al. (2014) do not control
for changes in RNOA in their regression models, which potentially cause an omitted variable problem, and
thus, the inference from their results should be cautiously interpreted. Overall, the literature finds changes
in ATO to be informative in predicting future changes in RNOA and useful to market participants, whereas
changes in PM are not. This is surprising given the significance of margins as a tool in analyzing firms.
Research in the fundamental analysis suggests that changes in ATO can predict one-year-ahead future
changes in in RNOA because the efficiency of asset utilization is less threatened by competition from rival
firms as it is costly to copy another company’s production process. Soliman (2008) proposed the reason of
uninformative changes in PM that knowledge is prone to imitation from rival firms. Therefore, the return
based on knowledge-based capital may be more transitory (Romer, 1986) and, as a result, it cannot predict
future changes in RNOA.
2.3 Theory and hypotheses
Contrary to the notion of the diminishing return based on intangible capital, the resource-based view of a
firm highlights the significance of intangible resources (e.g. advertising, R&D, organizational process,
brand names, etc.) as a key source of competitive advantage that generates profits because such capital
creates barriers to entry and product differentiation. In strategy literature, the resource-based view argues
10
that a firm’s endowment of unique resources drives significant and persistent heterogeneity, and such
resources are a primary determinant of its ability to create and sustain competitive advantage. Research in
the strategy literature posits and find that intangible capital is a critical driver of a value-creating or product
differentiation strategy (e.g., Lippman and Rumelt, 1982; Rumelt, 1984; Rumelt, 1991; Wernerfelt, 1984;
Dierickx and Cool, 1989; Barney, 1991) because intangible resources are internally developed (Dierickx
and Cool, 1989), rare, valuable, and imperfectly imitable (Barney, 1991). If acquired, intangible resources
are likely to be traded in imperfect factor markets (Barney, 1986). As a result, it is difficult or time-
consuming for rival firms to imitate.
In economics and finance literature, Sutton (1991) posits that firms spend on intangible capital such as
R&D and advertising to create endogenous barriers to entry and differentiate themselves from their
competitors. Sellling and Stickney (1989) suggest that firms that focus on product differentiation emphasize
on R&D, advertising, and capacity growth, and such spending should contribute to a higher profit margin.
Studies find supporting evidence on this perspective. For instance, Hoberg and Phillips (2016) find that
R&D and advertising are associated with future differentiation from competitors and increased profitability.
Brown and Kimbrough (2011) document that earnings noncommality, which is a portion of a firm’s
performance that is determined by firm-specific factors rather than industry-and-market-specific factors, is
positively associated with the intensity of intangible assets.
One example of intangible resources used as a strategic asset is organization capital, which Lev and
Radhakrishnan (2005) viewed as an accumulation of distinct business processes and practices that enable
firms to increase efficient use and yield an abnormal value of products. Some research views organization
capital as embodied in the firm’s key talent and has a firm-specific efficiency (Eisfeldt and Papanikolaou,
2013). Lev and Radhakrishnan (2005) argue that organization capital provides a firm’s competitive
advantage because it is firm-specific and, thus, cannot be simply transferred to other organizations. Prior
research has found that firms with larger organization capital are more productive, have higher average
returns (Eisfeldt and Papanikolaou, 2013), and have better future operating performance, such as the growth
11
of sales and operating income (Lev et al., 2009). Therefore, organization capital likely influences profit
margin and future profitability.
Under a DuPont analysis framework, a PM component provides information about pricing power such as
product innovation, better quality service, brand reputation, etc. Thus, this component is likely linked to
intangible capital. Firms that increase their investment in intangible capital should be able to improve their
pricing power and, thus, future profitability because intangible capital enables firms to create barriers to
entry and differentiate themselves and, thus, strengthen their competitive advantage against their rivals.
Therefore, new information about improved pricing power in generating revenue should be reflected in a
better the PM ratio. Therefore, I predict that the PM component should be informative in predicting future
profitability. This leads to my first hypothesis.
H1: Changes in profit margin are positively associated with one-year-ahead changes in return on net
operating assets
Next, I further explore my prediction that the changing nature of the economy may be attributing to the
positive association between changes in PM and future RNOA in recent few decades. I investigate whether
intangible capital influences the relation between the DuPont ratios and future changes in profitability. To
do this, I examine the association between changes in PM and one-year-ahead RNOA between both STEM
and non-STEM firms. I focus on STEM firms as a proxy for intangible intensity because they are
increasingly crucial in driving the U.S. economy and have different firms’ characteristics from
manufacturing and industrial firms’ (Fedyk et al., 2017).
STEM firms are a product of science and technology innovations. Unlike non-STEM firms such as
manufacturing and industrial firms whose primary profitability is driven from physical assets, STEM firms
focus and heavily invest in intangible capital such as R&D and human capital to create and sustain their
competitive advantage. Such difference in significant intangible investment likely affects firms’ business
operations and profitability. Because intangible capital enables firms to sustain their competitive advantage,
12
firms with high intangible intensity are more likely be able to maintain their profit margin and profitability
better than those with low intangible intensity. Therefore, I anticipate that the PM component is more
informative about future profitability for STEM firms than non-STEM firms. This leads to the following
hypothesis.
H2: the association between changes in profit margin and one-year-ahead changes in return on net
operating assets is stronger for STEM firms than non-STEM firms.
13
Chapter 3: Sample and descriptive statistics
3.1 Variable measurement
I calculate the DuPont ratios based on prior studies (Soliman, 2008). I use return on net operating assets
(RNOA) to measure firms’ profitability because such measure excludes the effects of financial leverage.
RNOA is calculated as operating income before interest divided by the average net operating asset (NOA),
where NOA is operating assets – operating liabilities. Operating assets are total assets less cash and short-
term investment. Operating liabilities are total assets less the short-term and long-term debt, less book value
of total common and preferred equity, and less minority interest. RNOA is decomposed into PM and ATO’s
multiplicative components, where PM is operating income divided by total sales, and ATO is sales divided
by average NOA. Growth in the net operating asset is NOA_CH = ( 𝑁𝑂 𝐴 𝑡 − 𝑁𝑂 𝐴 𝑡 − 1
) /𝑁𝑂 𝐴 𝑡 − 1
. Changes
in profit margin are PM_CH = 𝑃 𝑀 𝑡 − 𝑃 𝑀 𝑡 − 1
, and changes in asset turnover are ATO_CH = 𝐴𝑇 𝑂 𝑡 −
𝐴𝑇 𝑂 𝑡 − 1
. Future changes in return on net operating assets are RNOA_CHF = 𝑅 𝑁 𝑂 𝐴 𝑡 + 1
− 𝑅 𝑁 𝑂 𝐴 𝑡 . Changes
in return on net operating assets from last year are RNOA_CHL = 𝑅 𝑁 𝑂 𝐴 𝑡 − 𝑅 𝑁𝑂 𝐴 𝑡 − 1
. All continuous
financial variables are winsorized at 1 percent and 99 percent to reduce the effect of outliers.
Following the literature on STEM firms, I identify STEM and non-STEM firms using a combination of SIC
code and a list of IPO firms from Jay Ritter’s website (Fedyk et al., 2017). STEM is an indicator variable
equal to one if observations are identified as STEM firms and zero otherwise.
My primary measure of intangible intensity (Intan_Intensity) is defined as total intangible capital divided
by total capital. Because the U.S. accounting standards allow firms to capitalize externally purchased
intangible assets but not the internally developed intangible assets such as R&D expenditures, the Balance
Sheet’s intangible assets do not fully reflect all intangible capital that firms have invested. Therefore, in the
study, total intangible capital consists of intangible assets on the Balance Sheet and off-Balance-Sheet
intangible capital, which includes knowledge capital (R&D capital) and organization capital. Total capital
is total assets plus off-Balance-Sheet intangible capital. I utilize Peters and Taylor’s total Q dataset for data
on knowledge capital and organization capital. In their study, Peters and Taylor estimate a firm’s knowledge
14
capital by accumulating past R&D expenses using the perpetual inventory method. Similar to the property,
plant, and equipment, knowledge capital is amortized over time. Peters and Taylor use the BEA’s industry-
specific R&D amortization rate to amortize knowledge capital. Organization capital is estimated by
accumulating 30 percent of SG&A (selling, general, and administrative expenses) and leaving the
remaining 70 percent of SG&A as current operating expenses. Once intangible intensity (Intan_Intensity)
is calculated, I then yearly rank and assign Intan_Intensity into quintile and scale it to range between 0 and
1 to derive Intan_QuintRank. I use Intan_QuintRank to perform all main empirical tests. I use the other two
measures of intangible intensity, which is intangible capital excluding goodwill (IntanNoGW_QuintRank)
and internally developed intangible capital (RDSGA_QuintRank) to perform the robustness check presented
in section 5. The definition of IntanNoGW_QuintRank and RDSGA_QuintRank is also in section 5.
3.2 Sample data
I investigate the informativeness of the DuPont components in predicting firms’ changes in future
profitability for a sample of the U.S. public firms over 1984 – 2017. Following Soliman (2008), I start the
sample period in 1984 to be able to compare my replicated empirical results to prior studies’. I obtain
financial information from Compustat. I obtain data on intangible capital from Peters and Taylor Total Q
dataset, which ends in 2017. Following the DuPont analysis literature, I exclude financial institutions (SIC
6000-6999), firms identified as public service, international affairs, or non-operating establishments (SIC
9000+), and firms with negative net operating assets from the sample observations. Following Peters and
Taylor (2017), I exclude firms with missing or non-positive book value of total assets or sales and firms
with less than $5 million in physical capital. I also drop observations with missing data on variables
necessary to perform empirical tests. Lastly, I drop my sample of 2017 because the number of observations
in 2017 becomes small once I drop all missing variables necessary to perform tests in my primary
15
regression
5
. My final sample consists of 111,584 firm-year observations
6
and 11,986 unique firms over the
period 1984-2016.
3.3 Descriptive statistics
In Figure 1, I illustrate the average intangible intensity for all firms, STEM firms, and non-STEM firms
during 1984-2016 in three different measures of intangible intensity: (1) Intan_Intensity (in Panel A), (2)
IntanNoGW_Intensity (in Panel B), and (3) RDSGA_Intansity (in Panel C). In Figure 1, Panel A and Panel
B show an increase in intangible capital relative to total capital over time for all firms. Panel A depicts an
average increase in intangible intensity for all firms from about 30 percent in 1984 to just over 50 percent
in 2016, while Panel B showed an increase from 33 percent to 40 percent in 2016. In Figure 1, Panel C
reports a relatively stable intangible intensity over the same period. In Figure 1, all panels show a significant
increase in intangible intensity around late 1990 to early 2000, which coincides with the internet bubble or
the dot-com crisis. The steep increase indicates that there might be a shift in firms’ behaviors in intangible
capital investment after the crisis.
Figure 2 plots the mean intangible intensity over time within STEM industries (Technology, Internet, and
Science) in three different measures of intangible intensity: (1) Intan_Intensity (in Panel A), (2)
IntanNoGW_Intensity (in Panel B), and (3) RDSGA_Intansity (in Panel C). In Figure 2, all panels show that
science firms' average intangible intensity has increased over time. This suggests that science firms have
increased intangible intensity by both investing intangible capital internally and acquiring it from the
market. All panels show a significant decline in internet firms’ intangible intensity around late 1990 to early
2000, which coincides with the internet bubble, and then a drastic increase in intangible intensity. Later,
the intangible intensity of internet firms stabilized at the same level as technology firms after the crisis.
5
About 900 observations are dropped from my sample.
6
Soliman (2008) excludes firm-year observations that are not tracked by IBES or do not have contemporaneous and future return
data on CRSP, whereas my sample includes those observations. Most of my results hold after excluding those observations.
16
Table 1 reports descriptive statistics. Panel A of Table 1 shows descriptive statistics for all firms.
Intan_Intensity measure shows the mean intangible capital is 32 percent relative to total capital. Excluding
goodwill, the average intangible intensity is 27 percent of total capital as reported by IntanNoGW_Intensity
measure. The mean intangible intensity that is internally generated (RDSGA_Intensity) is 24 percent
compared to total capital. PM and RNOA are lower than the value reported in Soliman (2008) because 1)
loss firms are not excluded in this analysis
7
since many loss firms are from their investment in intangible
capital such as R&D, and 2) the sample period is different. 31 percent of all firms are considered STEM
firms. The proportion of STEM firms in my sample is smaller than that reported by Fedyk et al. (2017)
because my sample includes firms with no investment in R&D, whereas Fedyk et al. (2017) exclude those
firms.
Panel B and C of Table 1 report summary statistics of non-STEM and STEM firms, respectively, whereas
Panel D of Table 1 presents the univariate test of the difference in the mean value. As expected, STEM
firms’ intangible capital (Intangible_Capital) is significantly larger than non-STEM firms’ since they rely
on intangible capital as their competitive advantages. The average intangible capital for STEM firms is
$2,129 million, whereas that for non-STEM firms is $1,180 million. Looking at intangible intensity
(Intan_Intensity), STEM firms’ intangible capital is 43 percent of their total capital, whereas Non-STEM
firms’ is 27 percent of their total capital. The other two measures, IntanNoGW_Intensity and
RDSGA_Intensity, show similar statistics.
Table 2 presents the pairwise correlation Table. Consistent with prior literature (Fairfield and Yohn 2001;
Soliman 2008), RNOA_CHF is positively associated with ATO_CH. Looking at the pairwise correlations,
there is a positive correlation between STEM and three measures of intangible intensity, consistent with the
notion that STEM firms highly invest in intangible capitals. There is also a positive correlation between
7
My decision of not excluding loss firms is similar to Dickenson (2011) who develops a proxy of firm life cycle using cahflow
patterns and use DuPont analysis and her proxy of firm life cycle to predict changes in future profitability. Because R&D
expenses are crucial in product differentiation strategy, excluding loss firms due to a significant investment in those expenses
may be inappropriate.
17
PM_CH and all three measures of intangible intensity, Intan_QuintRank, IntanNoGW_QuintRank, and
RDSGA_QuintRank. This is consistent with the argument that firms that invest more in intangible capital
are more likely to have a larger improved profit margin than those that invest less.
18
Chapter 4: Empirical analysis
4.1 Replication of prior research with an extended period
I begin the empirical analysis with the replication of prior studies (Fairfield and Yohn, 2001; Soliman,
2008) by estimating the following regression models using Fama-MacBeth methodology with Newey-West
adjustments for serial correlation:
RNOA_CHF = β 0+β 1PM_CH+ β 2ATO_CH+ β 3RNOA + β 4RNOA_CHL+ β 5NOA_CH+ AB controls (1)
RNOA_CHF = β 0+β 1PM_CH+ β 2ATO_CH+ β 3RNOA + β 4RNOA_CHL+ β 5RSST controls +AB controls
(2)
The above specification is similar to the one used by Soliman (2008). Model (1) shows the DuPont
components with changes in NOA (NOA_CH) and changes in RNOA (RNOA_CHL) as control variables.
Model (2) replaces changes in NOA (NOA_CH) with the three components of total accruals defined in
Richardson et al. (2005). Total accrual components include working capital (WC), noncurrent operating
(NCO), and financing (FIN). Therefore, NOA_CH is expanded into WC_CH, NCO_CH, and FIN_CH
(RSST controls). Following Soliman (2008), I also control for firms fundamental signals, proposed by Lev
and Thiagarajan (1993) and later used by Abarbanell and Bushee (1997) (AB controls) in both Model (1)
and Model (2). The measures of firm fundamentals include inventory, accounting receivables, capital
expenditures, gross margins, selling, and administrative expenses, effective tax rate, earnings quality, audit
quality, and labor forces. See Table 3 for their definitions.
Table 3 presents the results of Model (1) and (2). Column (1) and Column (2) show the results of the prior
research’s replication over the sample period between 1984 and 2002 in Soliman (2008). Consistent with
previous studies (Fairfield and Yohn, 2001; Soliman 2008), both Column (1) and Column (2) show that a
positive and significant association between ATO_CH and RNOA_CHF. The finding is consistent with the
notion that changes in ATO convey new information in predicting future changes in profitability. Column
(1) and Column (2) report that consistent with previous studies (Fairfield and Yohn, 2001; Soliman 2008),
19
coefficients on PM_CH are positive but not significant. This finding in the literature is somewhat surprising
given the significance of margins as a tool in valuing firms.
Next, I extend the sample period and re-examine the association between DuPont ratios and the future
changes in RNOA. Column (3) and Column (4) of Table 3 report the results with the extended period from
2003 to 2016 based on the estimates of Model (1) and (2), respectively. I find that over the extended period,
ATO_CH is positively associated with RNOA_CHF when estimated using Model (2) although the positive
association between ATO_CH and RNOA_CHF is not significant when estimated using Model (1). In
general, this finding is consistent with prior studies (Fairfield and Yohn, 2001; Soliman, 2008). Looking at
Column (3) and Column (4), I find a positive and significant association between PM_CH and RNOA_CHF,
consistent with my first hypothesis. This finding is new to the literature because prior research finds
PM_CH to be uninformative. The new finding suggests that changes in PM contain new information about
future changes in profitability and are consistent with the notion that increased intangible capital strengthens
firms’ barriers to entry and improves product differentiation.
Lastly, I investigate the property of the DuPont components over the whole sample period (1984-2016).
Column (5) and Column (6) present the results based on the estimates of Model (1) and (2), respectively.
Consistent with prior research, I find a positive and significant relation between ATO_CH and RNOA_CHF.
Moreover, I also find that PM_CH is positively associated with RNOA_CHF, consistent with my first
hypothesis. Overall, the evidence offers preliminary evidence that the shift of the economy toward
intangible capital may contribute to the change in the relation between DuPont components and one-year-
ahead RNOA.
4.2 STEM and non-STEM firms
To test my second hypothesis, I estimate the following regressions to test the significant difference in the
association between PM_CH and RNOA_CHF between STEM firms and non-STEM firms.
20
RNOA_CHF = β 0 + β 1PM_CH*STEM + β 2PM_CH + β 3ATO_CH+ β 4STEM + β 5RNOA +
β 6RNOA_CHL + β 7 NOA_CH+ AB controls+ FE (3)
RNOA_CHF = β 0 + β 1PM_CH*STEM + β 2PM_CH + β 3ATO_CH+ β 4STEM + β 5RNOA +
β 6RNOA_CHL + β 87RSST controls + AB controls+ FE (4)
Model (3) and (4) are modified from Model (1) and (2), respectively, by adding STEM variable and its
interaction term (PM_CH*STEM). STEM is an indicator variable equal to one if firms operates in STEM
industries (science, technology, and Internet), and zero otherwise. I change the estimation method from the
Fama-MacBeth approach to the OLS regressions with fixed effects because it allows me to control for
unobservable time-invariant firms’ characteristics (firm fixed effects) and time-varying factors common
(year fixed effects) to all firms in my sample. I cluster the standard errors at the firm level to control for
potential time-series dependence in the residual. In section 5 robustness check, I re-estimate the coefficients
using the Fama-MacBeth approach, and the results remain the same. My variable of interest is the
interaction term (PM_CH*STEM). The interaction term’s positive coefficients (β 1) suggest that the
association between PM_CH and RNOA_CHF is stronger for STEM firms than non-STEM firms. Similar
to Model (1) and (2), I include RSST controls to control for total accruals, and AB controls to control for
firms’ fundamental signals.
Of Table 4, Panel A reports the results of the above regressions. Consistent with my second hypothesis, I
find that the interaction term’s coefficients (β 1) are significant and positive, as shown in Column (1) and
Column (2). I also find that the coefficients on ATO_CH (β 3) are positive and significant, consistent with
prior studies (Fairfield and Yohn, 2001; Soliman 2008). The evidence is consistent with the economic
intuition that changes in ATO bring new information about improved asset utilization in generating future
profits.
Next, I further estimate the Model (1) and (2) conditional on being STEM firms and non-STEM firms. Of
Table 4, Panel B shows the results of the conditional regressions. I find that coefficients on PM_CH are
21
positively associated with RNOA_CHF for STEM firms, whereas they are not for non-STEM firms. I
interpret this finding that new information about intangible capital is much more informative in predicting
one-year-ahead changes in profitability for STEM firms but not non-STEM firms.
In Panel C, I explore the sub-STEM groups: science, technology, and the internet. Panel C shows the results
of the usefulness of DuPont ratios for each subgroup. Consistent with my second hypothesis, I find that the
coefficients on PM_CH are significant and positive in every subgroup. As firms in the STEM industries
increase their investment in intangibles to strengthen their competitive advantage, new information about
intangible capital is informative in predicting future changes in profitability. I also document that ATO_CH
is positively associated with RNOA_CHF for science and technology firms. In contrast, ATO_CH is not
positively associated with RNOA_CHF for internet firms. The finding is consistent with the view that new
information about improved asset efficiency is less critical for internet firms because the physical assets are
not the central resource in generating revenue.
4.3 Intangible intensity
In this section, I expand the STEM/Non-STEM classification into a more generic one based on the intensity
of intangible capital. If intangible capital allows firms to maintain their competitive advantage and
contribute to firms’ ability to sustain their profit margin and profitability, I conjecture that firms with higher
intangible intensity have a stronger relation between changes in PM and future changes in RNOA. To test
my prediction, I estimate the following regression models:
RNOA_CHF = β 0 + β 1PM_CH* Intan_QuintRank + β 2PM_CH + β 3ATO_CH+ β 4 Intan_QuintRank +
β 5RNOA + β 6RNOA_CHL + β 7 NOA_CH+ AB controls+ FE (5)
RNOA_CHF = β 0 + β 1PM_CH* Intan_QuintRank + β 2PM_CH + β 3ATO_CH+ β 4 Intan_QuintRank +
β 5RNOA + β 6RNOA_CHL + β 87RSST controls + AB controls+ FE (6)
Model (5) and (6) are modified Eq. (3) and (4), respectively, by substituting the STEM variable with the
Intan_QuintRank variable to capture the intensity of intangible capital. My variable of interest is the
22
interaction term between Intan_QuintRank and PM_CH. The positive coefficient (β 1) indicates that changes
in PM are more useful in predicting future changes in RNOA for firms with higher intangible intensity than
those with lower intangible intensity as intangible resources strengthen or maintain firms’ competitive
advantage. Similar to Model (3) and (4), I include RSST controls to control for total accruals and AB
controls to control for fundamental signals and fixed effects.
Table 5 reports the results of Model (5) and (6) in Column (1) and Column (2), respectively. Consistent
with my prediction, I find positive and significant coefficients on the interaction term
(PM_CH*Intan_QuintRank). Specifically, the coefficients on PM_CH for firms in the lowest quintile rank
of intangible intensity (Intan_QuintRank=0) are -0.0178 and -0.0172, as presented in Column (1) and
Column (2), respectively, whereas they are 0.0811 and 0.0814
8
for firms in the highest quintile rank of
intangible intensity (Intan_QuintRank=1), as shown in Column (1) and (2), respectively. The finding
suggests that new information about intangible capital is more informative in predicting future changes in
profitability for firms with higher intangible intensity than those with lower intangible intensity.
4.3 Before and after 2000
Figure 1 depicts an increase in intangible intensity around 2000, which coincides with the Dot-Com crisis
or the internet bubble. Although this event may induce changes in firms’ behavior regarding intangible
investment, I do not make a claim that a change in the association between DuPont ratios and one-year-
ahead changes in profitability is due to this event. I recognize that other events may have caused the change
in firms’ behaviors over the same time. In fact, the purpose of the analysis is to explore the time difference
between the industrial-firm-dominated period, which I consider the period before 2000, and the technology-
and-service-firm-dominated period, which I consider the period after 2000. I investigate whether the
influence of intangible capital on the association between changes in PM and future changes in RNOA
happened in the period after 2000, where knowledge-and-technology-based firms dominated the economy.
8
The coefficient in column (1), which is 0.0811, is calculated as 0.0841-0.0178 +0.0148. The coefficient in column (2), which is
0.0814, is calculated as 0.0844-0.0172 +0.0142
23
I predict that the influence of intangible capital on the association between changes in PM and one-year-
ahead changes in RNOA occurs in the technology-and-service-firm-dominated period (after 2000). To test
my prediction, I estimate the following regression models.
RNOA_CHF = β 0 + β 1PM_CH + β 2*Intan_QuintRank + β 3PM_CH* Intan_QuintRank + β 4Post+
β 5Post*PM_CH+ β 6Post*Intan_QuintRank+ β 7Post*PM_CH*Intan_QuintRank+β 8ATO_CH+ β 9RNOA
+ β 10RNOA_CHL + β 11 NOA_CH+ AB controls+ FE (7)
RNOA_CHF = β 0 + β 1PM_CH + β 2*Intan_QuintRank + β 3PM_CH* Intan_QuintRank + β 4Post+
β 5Post*PM_CH+ β 6Post*Intan_QuintRank+ β 7Post*PM_CH*Intan_QuintRank+β 8ATO_CH+ β 9RNOA
+ β 10RNOA_CHL + β 11 RSST controls+ AB controls+ FE (8)
The above regression models are modified from Model (5) and (6), respectively, by adding the Post
variable. Post is an indicator variable equal to 1 if year is after or equal to 2000, and 0 otherwise. My
variable of interest is the triple interaction term (Post*PM_CH*Intan_QuintRank). A positive coefficient
(β 7) of the triple interaction term indicates that the influence of intangible intensity on the association
between DuPont ratios and future changes in profitability happens in the period after 2000.
Table 6 reports the result. Consistent with my prediction, Column (1) and Column (2) show positive and
significant coefficients on the triple interaction term (Post*PM_CH*Intan_QuintRank). The influence of
intangible capital that changes in PM are more useful for firms with higher intangible intensity than those
with lower intangible intensity occurs in the technology-and-service-firm-dominated period (after 2000).
Moreover, I find that the interaction term’s coefficients (PM_CH*Intan_QuintRank) are not significant.
The finding suggests that the intensity of intangible capital does not influence the relation between changes
in PM and future changes in RNOA in the before 2000 period (when Post = 0).
4.4 DuPont analysis and industry concentration
In this section, I investigate the role of intangible intensity in DuPont analysis when firms operate under
different industry concentration environments. Low concentrated industries are those that have many firms
24
operate in them. Firms operating in such industries likely face high competition. To maintain their profit
and survival, firms heavily invest in intangibles to sustain their competitive advantages over their rivals.
Therefore, I predict that under low concentrated industries, the relation between changes in PM and one-
year-ahead changes in RNOA is more pronounced for firms with higher intangible intensity. However,
under high concentrated industries (those with fewer firms operating in them), firms likely experience lower
competition, and, thus, the role of intangible capital becomes less critical. Therefore, I conjecture that there
is no difference in the association between changes in PM and future changes in RNOA between firms with
high and low intangible intensity.
To test my predictions, I run the regression Model (5) and (6) conditional on under low and high
concentrated industries. My measure of industry concentration is the text-based measure of concentration
from Hoberg and Phillips (2016). I classify observations into high and low industry concentrations based
on the median HHI in a particular year. Observations are yearly classified as being under high industry
concentration if the HHI is higher than the median HHI, and vice versa.
Table 7 reports the results of the above analysis. Consistent with my prediction, I find that coefficients on
the interaction term (PM_CH*Intan_QuintRank) are positive and significant, as shown in Column (1) and
(2). The evidence is consistent with the notion that intangible capital is essential when firms operate in high
competition environment as such capital is used to create barriers to entry and differentiate themselves
(Sutton, 1991). Column (3) and (4) present the results under high industry concentration environment.
Consistent with my prediction, I find that the interaction term's coefficients (PM_CH*Intan_QuintRank)
are not statistically significant. Overall, the evidence
9
is consistent with Sutton’s theory of endogenous
barriers to entry and product differentiation.
9
In untabulated tables, the results remain the same if I use different intangible intensity measures (IntanNoGW_QuintRank and
RDSGA_QuintRank).
25
4.5 DuPont analysis and product similarity
I next explore the role of intangible intensity in DuPont analysis when firms operate under different product
similarity environment. Firms operating in the product market that offers similar products to theirs’ likely
face high competition. To stay competitive under high product similarity space, firms need to invest in
intangible resources to create barriers to entry and differentiate their products. Therefore, under a high
product similarity environment, I posit that the association between changes in PM and future changes in
RNOA is more pronounced for firms with higher intangible intensity than those with lower intangible
intensity. However, under low product similarity environment, firms likely experience low competition
since customers already perceive the firm’s product to be different from their competitors’. Therefore,
higher investment in intangible is not necessary. As a result, I conjecture that there is no difference in the
relation between changes in PM for future changes in profitability for firms with high intangible intensity
and those with low intangible intensity under low product similarity space.
To test my predictions, I run the regression Model (5) and (6) conditional on low and high product
similarity. My product similarity measure is the text-based measure of total similarity from Hoberg and
Phillips (2016). This measure is constructed by using the business description section of the 10-K, where
firms describe detail on the products or services they offer. Hoberg and Phillips (2016) use product words
in 10-K to assign each firm a spatial location and measure the firm pairwise distance. Total similarity score
is the sum of the pairwise similarity between a given firm and all other firms in the sample in a given year.
This measure captures competitive threats or potential to entry threats. Firms with higher total product
similarity are more likely to have higher competitive pressure as perceived by the management. To obtain
a high or low product similarity group, I yearly split total similarity scores into two groups based on the
median similarity scores in a particular year. I classify observation as under high product similarity space
if the total similarity score is higher than the median similarity scores and vice versa.
Table 8 presents the results. Consistent with my prediction, I find that the coefficients on the interaction
term (PM_CH*Intan_QuintRank) are not statistically significant as shown in Column (1) and Column (2),
26
whereas I find that the coefficients on interest (PM_CH*Intan_QuintRank) are positive and statistically
significant as reported in Column (3) and (4). Consistent with Sutton’s hypothesis that intangible
investments can create endogenous barriers to entry and product differentiation, the finding indicates that
firms invest in intangible capital to sustain their competitive advantage and help their profit margin and
future profitability when operating under high product similarity space.
27
Chapter 5: Robustness check
This section describes my results’ robustness check across different measures of intangible intensity and
regression methodology.
5.1 Different measures of intangible intensity
In this section, I show that the other two different measures of intangible intensity do not change my main
empirical results. The first alternative measure of intangible intensity is intangible intensity excluding
goodwill (IntanNoGW_Intensity). Goodwill is the premium the acquirer pays more than the value of
identifiable assets when purchases a business. Arguably, goodwill may or may not generate future benefits
for the acquirer because the acquirer may overpay when purchasing a business. As a result, such acquirer
later takes losses on the impairment of goodwill. Given the concern of future benefits related to goodwill,
I calculate a measure of intangible intensity excluding goodwill (IntanNoGW_Intensity) which is the ratio
of intangible capital excluding goodwill to total capital. Then, I yearly rank IntanNoGW_Intensity into
quintile and scale to range between 0 and 1 to derive IntanNoGW_QuintRank. I use
IntanNoGW_QuinRankk to re-run empirical analyses and find similar results.
Another alternative measure of intangible intensity is R&D and organization intensity (RDSGA_Intensity).
This measure focuses on internally generated intangible capital, whereas the other two measures include
externally acquired intangible assets. I calculate RDSGA_Intensity as the sum of R&D capital and
organization capital divided by total capital. RDSGA_Intensity is then yearly ranked into quintile and later
scaled to range between 0 and 1 to derive RDSGA_QuintRank. I use RDSGA_QuinRankk to re-run empirical
analyses and find similar results.
Table 9 presents the results of Model (5) and (6) using the two alternative measures of intangible intensity.
Panel A shows the results estimated by IntanNoGW_QuintRank, while Panel B reports the results estimated
by RDSGA_QuintRank. Consistent with my main results reported in Table 5, I find positive and significant
coefficients on the interaction term (PM_CH*IntanNoGW_QuintRank and PM_CH*RDSGA_QuintRank)
28
as reported in both Panel A and Panel B. I also find that the coefficients on other variables, including
PM_CH and ATO_CH, are consistent with the findings presented in Table 5.
Table 10 shows the results of Model (7) and (8) estimated by IntanNoGW_QuintRank and
RDSGA_QuintRank as measures of intangible intensity. Both Panel A and Panel show that coefficients of
the triple interaction term (Post *PM_CH* IntanNoGW_QuintRank and Post *PM_CH*
RDSGA_QuintRank) are positive and significant, consistent with the findings reported in Table 6. In sum,
my main findings are robust to different measures of intangible intensity.
5.2 Results using the Fama-MacBeth approach
In this section, I re-estimate coefficients using the Fama-MacBeth method to be consistent with prior
research (Soliman (2008) and ensure that my findings are robust to the different estimations. Table 11
presents the results estimated by the Fama-MacBeth approach. Panel A reports the results of Model (3) and
(4) in Column (1) and (2), respectively, whereas Panel B shows the results of Model (5) and (6) in Column
(1) and (2), respectively. These regression models estimated by the Fama-MacBeth method do not include
firm fixed effects and year fixed effects, similar to Model (1) and (2).
Consistent with my findings in Panel A of Table 4, coefficients on the interaction term (PM_CH*STEM)
are positive and significant, as shown in both Column (1) and (2) when estimated by the Fama-MacBeth
method. Also, I also find that coefficients on the interaction term (PM_CH*Intan_QuintRank) are positive
and significant in both Column (1) and Column (2), consistent with the finding in Table 5. Overall, my
main findings are robust to the different estimation methods.
29
Chapter 6: Conclusion
Motivated by the shift in the U.S. economic landscape toward an intangible-based economy, this study re-
examines the usefulness of DuPont analysis in predicting future changes in profitability. Prior literature
documents that changes in ATO are informative in predicting future changes in profitability, whereas
changes in PM are not. This is surprising given that both components are crucial in analyzing firms’
profitability. I start my analysis by replicating prior studies in the fundamental analysis literature and
document consistent evidence that changes in ATO(PM) are(are not) positively associated with one-year-
ahead changes of RNOA. Extending the sample period to 2016, I find that changes in ATO are positively
associated with future changes in RNOA, consistent with prior studies. I also document new evidence to the
fundamental analysis literature that changes in PM are positively associated with one-year-ahead changes
of RNOA. This finding indicates that changes in PM convey additional information about improved pricing
power in predicting future changes in profitability.
As intangible capital becomes crucial in the economy, I further investigate the role of intangible intensity
in the relation between changes in DuPont ratios and future changes in RNOA. Consistent with Sutton’s
hypothesis of endogenous barriers to entry and product differentiation, I document that the positive
association between changes in PM and future changes in RNOA is stronger for STEM firms than non-
STEM firms and more pronounced for higher intangible intensity firms than lower intangible intensity
firms. Sub-STEM group analysis shows the positive association between changes in PM and future changes
in RNOA for all sub-STEM firms. However, evidence shows that changes in ATO are positively associated
with future changes in RNOA for science and technology firms but not internet firms, consistent with the
notion that internet firms do not rely on physical assets in generating value. Therefore, new information
about asset efficiency is less critical for internet firms.
Lastly, I investigate the role of the intangible intensity in DuPont analysis under different operating
environment. I find that the relation that changes in PM are more useful in predicting future changes in
RNOA for firms with higher intangible intensity only exists when firms operate under low concentrated
30
industries or high product similarity space. The findings are consistent with the notion that when facing
high competition, firms invest in intangible resources to create barriers to entry and differentiate themselves
from their competitors and, thus, sustain profit margin and future profitability. Considering the above
evidence, this study contributes to the fundamental analysis literature and extends our understanding of
how the U.S. economy has evolved by investigating how STEM firms generate profits differently from
traditional industrial firms.
31
References
Abarbanell, Jeffrey S., and Brian J. Bushee. "Fundamental analysis, future earnings, and stock prices." Journal of
accounting research 35, no. 1 (1997): 1-24.
Barney, Jay B. "Strategic factor markets: Expectations, luck, and business strategy." Management science 32, no. 10
(1986): 1231-1241.
Barney, Jay. "Firm resources and sustained competitive advantage." Journal of management 17, no. 1 (1991): 99-120.
Brown, Nerissa C., and Michael D. Kimbrough. “Intangible investment and the importance of firm-specific factors in
the determination of earnings.” Review of Accounting Studies 16, no. 3 (2011): 539-573.
Chang, Kathryn J., Doina C. Chichernea, and Hassan R. HassabElnaby. "On the DuPont analysis in the health care
industry." Journal of Accounting and Public Policy 33, no. 1 (2014): 83-103.
Darrough, Masako, and Jianming Ye. “Valuation of loss firms in a knowledge-based economy.” Review of Accounting
Studies 12, no. 1 (2007): 61-93.
Demers, Elizabeth, and Baruch Lev. “A rude awakening: Internet shakeout in 2000.” Review of Accounting Studies 6,
no. 2-3 (2001): 331-359.
Dickinson, Victoria. "Cash flow patterns as a proxy for firm life cycle." The Accounting Review 86, no. 6 (2011):
1969-1994.
Dierickx, Ingemar, and Karel Cool. “Asset stock accumulation and sustainability of competitive
advantage.” Management science 35, no. 12 (1989): 1504-1511.
Eisfeldt, Andrea L., and Dimitris Papanikolaou. “Organization capital and the cross‐section of expected returns.” The
Journal of Finance 68, no. 4 (2013): 1365-1406.
Fairfield, Patricia M., and Teri Lombardi Yohn. “Using asset turnover and profit margin to forecast changes in
profitability.” Review of Accounting Studies 6, no. 4 (2001): 371-385.
Fedyk, Tatiana, Zvi Singer, and Mark Soliman. “The sharpest tool in the shed: IPO financial statement management
of STEM vs. non-STEM firms.” Review of Accounting Studies 22, no. 4 (2017): 1541-1581.
Guo, Re-Jin, Baruch Lev, and Nan Zhou. "The valuation of biotech IPOs." Journal of accounting, auditing &
finance 20, no. 4 (2005): 423-459.
Hand, John RM. "The role of economic fundamentals, web traffic, and supply and demand in the pricing of US Internet
stocks." European Finance Review, no.5 (2001): 295-317.
Hoberg, Gerard, and Gordon Phillips. "Text-based network industries and endogenous product
differentiation." Journal of Political Economy 124, no. 5 (2016): 1423-1465.
Itami, Hiroyuki, and Thomas W. Roehl. Mobilizing invisible assets. Harvard University Press, 1991.
Joos, Philip, and Alexei Zhdanov. “Earnings and equity valuation in the biotech industry: Theory and
evidence.” Financial Management 37, no. 3 (2008): 431-460.
Lev, Baruch, and Suresh Radhakrishnan. “The valuation of organization capital.” In Measuring capital in the new
economy, pp. 73-110. University of Chicago Press, 2005.
Lev, Baruch, and S. Ramu Thiagarajan. "Fundamental information analysis." Journal of Accounting research 31, no.
2 (1993): 190-215.
Lev, Baruch, Suresh Radhakrishnan, and Weining Zhang. “Organization capital.” Abacus 45, no. 3 (2009): 275-298.
32
Lippman, Steven A., and Richard P. Rumelt. “Uncertain imitability: An analysis of interfirm differences in efficiency
under competition.” The bell journal of Economics (1982): 418-438.
Nelson, Richard R. “Why do firms differ, and how does it matter?.” Strategic management journal 12, no. S2 (1991):
61-74.
Nissim, Doron, and Stephen H. Penman. “Ratio analysis and equity valuation: From research to practice.” Review of
accounting studies 6, no. 1 (2001): 109-154.
Peters, Ryan H., and Lucian A. Taylor. "Intangible capital and the investment-q relation." Journal of Financial
Economics 123, no. 2 (2017): 251-272.
Richardson, Scott A., Richard G. Sloan, Mark T. Soliman, and Irem Tuna. "Accrual reliability, earnings persistence
and stock prices." Journal of accounting and economics 39, no. 3 (2005): 437-485.
Romer, Paul M. “Increasing returns and long-run growth.” Journal of political economy 94, no. 5 (1986): 1002-1037.
Rumelt, R. (1984). Towards a strategic theory of the firm. In B. Lamb (Ed.), Competitive strategic management.
Englewood Cliffs: Prentice Hall.
Rumelt, Richard P. "How much does industry matter?." Strategic management journal 12, no. 3 (1991): 167-185.
Selling, Thomas I., and Clyde P. Stickney. "The effects of business environment and strategy on a firm’s rate of return
on assets." Financial Analysts Journal 45, no. 1 (1989): 43-52.
Soliman, Mark T. “The use of DuPont analysis by market participants.” The Accounting Review 83, no. 3 (2008): 823-
853.
Sutton, John. "Market structure: theory and evidence." Handbook of industrial organization 3 (2007): 2301-2368.
Trueman, Brett, MH Franco Wong, and Xiao-Jun Zhang. "The eyeballs have it: Searching for the value in Internet
stocks." Journal of Accounting Research (2000): 137-162.
Wernerfelt, Birger. "A resource‐based view of the firm." Strategic management journal 5, no. 2 (1984): 171-180.
33
Figure 1: The mean intangible intensity during 1984-2016 for all firms, STEM and, non-STEM firms with 3 measures of intangible intensity. Panel A:
Intan_Intensity equals the ratio of intangible capital (Intangible assets + R&D capital + Organization capital) to total capital (total assets + R&D capital
+Organization capital). Panel B: IntanNoGW_Intensity equals to the ratio of intangible capital without goodwill (Intangible assets –Goodwill + R&D capital
+Organization capital) to total capital (total assets + R&D capital +Organization capital). Panel C: RDSGA_Intensity is the ratio of R&D and Organization capital
(R&D capital + Organization capital) to total capital (total asset + R&D capital + Organization capital).
Figure 2: The mean intangible intensity during 1984-2016 with 3 measures of intangible intensity, within STEM industries including technology, science and internet
firms. Panel A: Intan_Intensity equals the ratio of intangible capital (Intangible assets + R&D capital + Organization capital) to total capital (total assets + R&D
capital +Organization capital). Panel B: IntanNoGW_Intensity equals to the ratio of intangible capital without goodwill (Intangible assets –Goodwill + R&D capital
+Organization capital) to total capital (total assets + R&D capital +Organization capital). Panel C: RDSGA_Intensity is the ratio of R&D and Organization capital
(R&D capital + Organization capital) to total capital (total asset + R&D capital + Organization capital).
.2 .3 .4 .5
Intangible Intensity
1982 1987 1992 1997 2002 2007 2012 2017
year
Intangible Intensity-All Intangible Intensity-STEM
Intangible Intensity-NonSTEM
PanelA: Intan_Intensity
Intangible Intensity
.2
.25
.3
.35
.4
Intangible Intensity
1982 1987 1992 1997 2002 2007 2012 2017
year
Intangible Intensity-All Intangible Intensity-STEM
Intangible Intensity-NonSTEM
Panel B: IntanNoGW_Intensity
Intangible Intensity
.15
.2
.25
.3
.35
Intangible Intensity
1982 1987 1992 1997 2002 2007 2012 2017
year
Intangible Intensity-All Intangible Intensity-STEM
Intangible Intensity-NonSTEM
Panel C: RDSGA_Intensity
Intangible Intensity
.2 .3 .4 .5 .6
Intangible Intensity
1985 1990 1995 2000 2005 2010 2015 2020
year
Intangible Intensity-Technology Intangible Intensity-Science
Intangible Intensity-Internet
PanelA: Intan_Intensity
STEM Industries
.2 .3 .4 .5 .6
Intangible Intensity
1982 1987 1992 1997 2002 2007 2012 2017
year
Intangible Intensity-Technology Intangible Intensity-Science
Intangible Intensity-Internet
Panel A: Intangible capital
STEM Industries
.2 .3 .4 .5
Intangible Intensity
1982 1987 1992 1997 2002 2007 2012 2017
year
Intangible Intensity-Technology Intangible Intensity-Science
Intangible Intensity-Internet
Panel C: RDSGA_Intensity
STEM Industries
34
Table 1: Descriptive statistics
Panel A: For all firms
Variables N Mean SD Min P25 P50 P75 Max
RNOA_CHF 111,584 -0.013 0.34 -1.7 -0.06 -0.0029 0.044 1.7
PM 111,584 -0.015 0.53 -4.1 0.017 0.07 0.13 0.43
ATO 111,584 2.5 2.6 0.14 1 1.8 3 17
PM_CH 111,584 0.0024 0.25 -1.3 -0.024 0.00059 0.022 1.5
ATO_CH 111,584 0.025 1.4 -5.9 -0.21 -0.0026 0.18 9.1
Intan_Intensity 111,584 0.32 0.21 0 0.15 0.32 0.47 0.98
IntanNoGW_Intensity 111,584 0.27 0.19 -0.6 0.11 0.26 0.4 0.98
RDSGA_Intensity 111,584 0.24 0.18 0 0.085 0.22 0.36 0.97
Intangible_Capital ($million) 111,584 1,474 7,252 0 23 96 484 278,770
RD_Capital ($million) 111,584 317 2,268 0 0 0.084 35 80,145
SGA_Capital ($million) 111,584 509 2,370 0 9.9 42 192 130,290
STEM 111,584 0.31 0.46 0 0 0 1 1
RNOA 111,584 0.086 0.48 -2.8 0.035 0.12 0.22 1.5
RNOA_CHL 111,584 -0.014 0.36 -2 -0.061 -0.0026 0.046 1.8
Panel B: For non-STEM firms
Variables N Mean SD P25 P50 P75
RNOA_CHF 76,989 -0.012 0.24 -0.049 -0.0031 0.036
PM 76,989 0.03 0.4 0.027 0.073 0.13
ATO 76,989 2.4 2.5 0.95 1.8 3
PM_CH 76,989 -0.00096 0.21 -0.021 0.0003 0.017
ATO_CH 76,989 -0.002 1.1 -0.19 -0.0036 0.16
Intan_Intensity 76,989 0.27 0.2 0.097 0.26 0.42
IntanNoGW_Intensity 76,989 0.22 0.17 0.072 0.2 0.34
RDSGA_Intensity 76,989 0.19 0.16 0.054 0.16 0.3
Intangible_Capital ($million) 76,989 1,180 5,316 17 88 468
RD_Capital ($million) 76,989 151 1,194 0 0 2.9
SGA_Capital ($million) 76,989 488 2,295 8.7 45 215
Panel C: For STEM firms
Variables N Mean SD P25 P50 P75
RNOA_CHF 34,595 -0.015 0.49 -0.1 -0.0022 0.078
PM 34,595 -0.12 0.73 -0.037 0.062 0.14
ATO 34,595 2.6 2.8 1.2 1.9 2.9
PM_CH 34,595 0.01 0.32 -0.037 0.0018 0.037
ATO_CH 34,595 0.084 1.9 -0.29 0.00079 0.25
Intan_Intensity 34,595 0.43 0.18 0.31 0.43 0.56
IntanNoGW_Intensity 34,595 0.37 0.17 0.26 0.37 0.48
RDSGA_Intensity 34,595 0.34 0.17 0.22 0.33 0.44
Intangible_Capital ($million) 34,595 2,129 10,302 36 113 513
RD_Capital ($million) 34,595 685 3,637 6.4 34 144
SGA_Capital ($million) 34,595 555 2,530 12 36 148
35
Panel D: The difference between STEM vs Non-STEM firms
STEM Non-STEM
Variables N Mean SD N Mean SD Diff t-stat
RNOA_CHF 34,595 -0.015 0.49 76,989 -0.012 0.24 0.003 1.28
PM 34,595 -0.12 0.73 76,989 0.03 0.4 -0.15 43.06
ATO 34,595 2.6 2.8 76,989 2.4 2.5 0.2 -11.39
PM_CH 34,595 0.01 0.32 76,989 -0.00096 0.21 -0.01096 -6.69
ATO_CH 34,595 0.084 1.9 76,989 -0.002 1.1 -0.086 -9.51
Intan_Intensity 34,595 0.43 0.18 76,989 0.27 0.2 -0.16 -1.3e+02
IntanNoGW_Intensity 34,595 0.37 0.17 76,989 0.22 0.17 -0.15 -1.3e+02
RDSGA_Intensity 34,595 0.34 0.17 76,989 0.19 0.16 -0.15 -1.4e+02
Intangible_Capital ($million) 34,595 2,129 10,302 76,989 1,180 5,316 -949 -20.26
RD_Capital ($million) 34,595 685 3,637 76,989 151 1,194 -534 -36.61
SGA_Capital ($million) 34,595 555 2,530 76,989 488 2,295 -67 -4.37
This table shows descriptive statistics of all main variables. All continuous variables are winsorized at 1%.
Variables are defined as follows:
Intangible_Capital = Total intangible capital (Peters and Taylor Total Q: k_int) which include (1) internally
developed intangible capital, R&D capital (Peters and Taylor Total Q: k_int_know) plus Organization capital (Peters and Taylor
Total Q item: k_int_org), and (2) externally purchased intangible capital (Compustat: Intan);
RD_Capital = R&D capital (Peters and Taylor Total Q: k_int_know);
SGA_Capital = Organization capital (Peters and Taylor Total Q item: k_int_org);
Intan_intensity = Intangible_Capital divided by total capital. Total capital is calculated as total assets
(Compustat: at) plus R&D capital (Peters and Taylor Total Q: k_int_know) plus organization capital (Peters and Taylor Total Q:
k_int_org);
IntanNoGW_intensity = Total intangible capital excluding Goodwill (Compustat: gdwl) divided by total capital.
RDSGA_intensity = R&D capital (Peters and Taylor Total Q: k_int_know) plus Organization capital (Peters
and Taylor Total Q item: k_int_org) divided by total capital.
STEM = An indicator variable equal to 1 if a firm operates in STEM industries, including
Science, Technology, and Internet. Following Fedyk et al. (2017), I identify STEM firms using a combination of SIC code and a
list of companies from Jay Ritter’s website.
NOA (Net operating assets) = Operating assets minus operating liabilities; Operating assets is calculated as total
assets (Compustat item: at) less less cash and short-term investment (Compustat item: che); Operating liabilities is calculated as
total assets (Compustat item: at) less total debt (Compustat item: dlc + dltt) less book value of total common (Compustat item:
ceq) and preferred equity (Compustat item: upstk) less minority interest (Compustat item: mib);
NOA_CH = (NOAt-NOAt-1)/NOAt-1;
PM (Profit margin) = Operating income (Compustat item: oiadp) divided by Sales (Compustat item: sale);
PM_CH = PMt-PMt-1;
ATO (Asset turnover) = Sales (Compustat item: sale) divided by Average NOAt (NOAt+NOAt-1)/2;
ATO_CH = ATOt-ATOt-1;
RNOA = PM * ATO;
RNOA_CHF = RNOAt+1 –RNOAt;
RNOA_CHL = RNOAt -RNOAt-1;
WC = current operating assets (COA) - current operating liabilities (COL); COA = current
assets (Compustat item: act) – cash and short- term investmen (Compustat item: che) t; COL = current liabilities (Compustat
item:lct) – debt in current liabilities (Compustat item: dlc);
WC_CH = WCt-WCt-1;
NCO = non-current operating assets (NCOA) – non-current operating liabilities (NCOL);
NCOA = total assets (Compustat item: at) – current asset (Compustat item: act) – investments and advances (Compustat item:
ivao); NCOL = total liabilities (Compustat item: lt) – current liabilities (Compustat item: lct)– long-term debt (Compustat item:
dltt);
NCO_CH = NCOt-NCOt-1;
FIN = financial assets (FINA) – financial liabilities (FINL); FINA = short-term investment
(Compustat item: ivst) + long-term investment (Compustat item: ivao); FINL = long-term debt (Compustat item: dltt)+ debt in
current liabilities (Compustat item: dlc)+ preferred stock (Compustat item: upstk);
FIN_CH = FINt-FINt-1.
36
Table 2: Pairwise correlation table
Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
(1) RNOA_CHF 1.000
(2) PM -0.111* 1.000
(3) ATO -0.047* 0.073* 1.000
(4) PM_CH -0.001 0.153* 0.044* 1.000
(5) ATO_CH 0.051* -0.022* 0.277* 0.117* 1.000
(6) RNOA -0.279* 0.653* 0.140* 0.043* -0.064* 1.000
(7) RNOA_CHL -0.086* 0.132* 0.035* 0.319* 0.186* 0.264* 1.000
(8) Intan_Intensity 0.032* -0.123* 0.188* 0.029* 0.035* -0.160* -0.015* 1.000
(9)
IntanNoGW_Intensity
0.032* -0.185* 0.268* 0.030* 0.059* -0.209* -0.018* 0.877* 1.000
(10) RDSGA_Intensity 0.031* -0.210* 0.337* 0.031* 0.075* -0.225* -0.017* 0.755* 0.903* 1.000
(11) Intan_QuintRank 0.024* -0.093* 0.192* 0.023* 0.024* -0.123* -0.014* 0.941* 0.860* 0.767* 1.000
(12)
IntanNoGW_QuintRank
0.023* -0.132* 0.264* 0.024* 0.044* -0.139* -0.012* 0.838* 0.953* 0.877* 0.867* 1.000
(13) RDSGA_QuintRank 0.020* -0.148* 0.320* 0.023* 0.055* -0.135* -0.010* 0.732* 0.860* 0.948* 0.765* 0.886* 1.000
(14) STEM -0.004 -0.128* 0.034* 0.020* 0.028* -0.144* -0.020* 0.355* 0.377* 0.390* 0.337* 0.368* 0.380* 1.000
* shows significance at the .05 level
This table reports pairwise correlation table. All continuous variables are winsorized at 1%.
Variables are defined as follows:
Intangible_Capital = Total intangible capital (Peters and Taylor Total Q: k_int) which include (1) internally developed intangible capital, R&D
capital (Peters and Taylor Total Q: k_int_know) plus Organization capital (Peters and Taylor Total Q item: k_int_org), and (2) externally purchased intangible capital
(Compustat: Intan);
RD_Capital = R&D capital (Peters and Taylor Total Q: k_int_know);
SGA_Capital = Organization capital (Peters and Taylor Total Q item: k_int_org);
Intan_intensity = Intangible_Capital divided by total capital. Total capital is calculated as total assets (Compustat: at) plus R&D capital (Peters
and Taylor Total Q: k_int_know) plus organization capital (Peters and Taylor Total Q: k_int_org);
IntanNoGW_intensity = Total intangible capital excluding Goodwill (Compustat: gdwl) divided by total capital.
RDSGA_intensity = R&D capital (Peters and Taylor Total Q: k_int_know) plus Organization capital (Peters and Taylor Total Q item: k_int_org)
divided by total capital.
Intan_QuintRank = Intan_Intensity is yearly ranked into quintile and scaled to range between 0 and 1.
IntanNoGW_QuintRank = IntanNoGW_Intensity is yearly ranked into quintile and scaled to range between 0 and 1.
RDSGA_QuintRank = RDSGA_Intensity is yearly ranked into quintile and scaled to range between 0 and 1.
STEM = an indicator variable equal to 1 if a firm operates in STEM industries, including Science, Technology, and Internet. Following
Fedyk et al. (2017), I identify STEM firms using a combination of SIC code and a list of companies from Jay Ritter’s website.
NOA (Net operating assets) = Operating assets minus operating liabilities; Operating assets is calculated as total assets (Compustat item: at) less less cash and
short-term investment (Compustat item: che); Operating liabilities is calculated as total assets (Compustat item: at) less total debt (Compustat item: dlc + dltt) less book
value of total common (Compustat item: ceq) and preferred equity (Compustat item: upstk) less minority interest (Compustat item: mib);
37
NOA_CH = (NOAt-NOAt-1)/NOAt-1;
PM (Profit margin) = Operating income (Compustat item: oiadp) divided by Sales (Compustat item: sale);
PM_CH = PMt-PMt-1;
ATO (Asset turnover) = Sales (Compustat item: sale) divided by Average NOAt (NOAt+NOAt-1)/2;
ATO_CH = ATOt-ATOt-1;
RNOA = PM * ATO;
RNOA_CHF = RNOAt+1 –RNOAt;
RNOA_CHL = RNOAt -RNOAt-1
38
Table 3: Time-series means and t-statistics for coefficients from annual cross-sectional regressions of future one-year-ahead
change in RNOA (RNOA_CHF) on the DuPont components
1984-2002 2003-2016 1984-2016
RNOA_CHF RNOA_CHF RNOA_CHF RNOA_CHF RNOA_CHF RNOA_CHF
PM_CH 0.0328 0.0328 0.0290
**
0.0308
**
0.0293
**
0.0297
**
(1.42) (1.45) (2.56) (2.70) (2.06) (2.15)
ATO_CH 0.00699
**
0.0100
***
0.00681 0.00741
**
0.00576
***
0.00819
***
(2.55) (3.64) (1.71) (2.32) (2.86) (4.10)
RNOA -0.205
***
-0.200
***
-0.188
***
-0.185
***
-0.202
***
-0.198
***
(-14.22) (-13.83) (-10.14) (-10.16) (-18.74) (-18.50)
RNOA_CHL -0.0317
*
-0.0319
*
-0.0405
**
-0.0413
**
-0.0375
***
-0.0378
***
(-1.77) (-1.82) (-2.68) (-2.71) (-3.09) (-3.16)
NOA_CH -0.0403
***
-0.0187
*
-0.0339
***
(-6.92) (-2.09) (-7.41)
WC_CH -0.278
***
-0.197
***
-0.236
***
(-7.86) (-3.61) (-7.91)
NCO_CH -0.108
***
-0.0867
***
-0.101
***
(-3.83) (-4.31) (-5.50)
FIN_CH -0.0675
***
-0.0562
**
-0.0626
***
(-3.27) (-2.56) (-4.11)
Constant 0.0136
***
0.0112
**
0.0127
**
0.0107
*
0.0131
***
0.0110
***
(3.31) (2.56) (2.26) (1.84) (3.87) (3.05)
N 63719 63719 47865 47865 111584 111584
adj. R
2
0.1104 0.1095 0.1091 0.1079 0.1113 0.1103
AB controls Yes Yes Yes Yes Yes Yes
This table presents results of the association between the DuPont components and future changes in RNOA using Fama-MacBeth
regressions (Model (1) and (2)). Column 1 and 2 show replication of prior studies (Fairfield and Yohn 2001; Soliman 2008) over
the sample period between 1984 and 2002. Column 3 and 4 report result over the sample period between 2003 and 2016, which is
the period after Soliman (2008). Column 5 and 6 present results over the whole sample period between 1984 and 2016. All
continuous variables are winsorized at 1% statistics in parentheses
*
p < 0.1,
**
p < 0.05,
***
p < 0.01
Variables are defined as follows:
NOA (Net operating assets) = Operating assets minus operating liabilities; Operating assets is calculated as total assets
(Compustat item: at) less less cash and short-term investment (Compustat item: che); Operating liabilities is calculated as total
assets (Compustat item: at) less total debt (Compustat item: dlc + dltt) less book value of total common (Compustat item: ceq)
and preferred equity (Compustat item: upstk) less minority interest (Compustat item: mib);
NOA_CH = (NOAt-NOAt-1)/NOAt-1;
PM (Profit margin) = Operating income (Compustat item: oiadp) divided by Sales (Compustat item: sale);
PM_CH = PMt-PMt-1;
ATO (Asset turnover) = Sales (Compustat item: sale) divided by Average NOAt (NOAt+NOAt-1)/2;
ATO_CH = ATOt-ATOt-1;
RNOA = PM * ATO;
RNOA_CHF = RNOAt+1 –RNOAt;
RNOA_CHL = RNOAt -RNOAt-1;
WC = current operating assets (COA) - current operating liabilities (COL); COA = current assets
(Compustat item: act) – cash and short- term investmen (Compustat item: che) t; COL = current liabilities (Compustat item:lct) –
debt in current liabilities (Compustat item: dlc);
WC_CH = WCt-WCt-1;
NCO = non-current operating assets (NCOA) – non-current operating liabilities (NCOL); NCOA = total
assets (Compustat item: at) – current asset (Compustat item: act) – investments and advances (Compustat item: ivao); NCOL =
total liabilities (Compustat item: lt) – current liabilities (Compustat item: lct) – long-term debt (Compustat item: dltt);
NCO_CH = NCOt-NCOt-1;
FIN = financial assets (FINA) – financial liabilities (FINL); FINA = short-term investment (Compustat
item: ivst) + long-term investment (Compustat item: ivao); FINL = long-term debt (Compustat item: dltt) + debt in current
liabilities (Compustat item: dlc) + preferred stock (Compustat item: upstk);
FIN_CH = FINt-FINt-1.
AB controls consist of the following variables:
AB_INV = change in inventory (Compustat: invt) – change in sales (Compustat: sale);
AB_AR = change in accounts receivable (Compustat: rect) – change in sale (Compustat: sale);
AB_CAPEX = change in industry Capex – change firm Capex (Compustat: capx);
39
AB_GM = change in sales (Compustat: sale) – change in cost of good sale (Compustat: cogs);
AB_ETR = Effective tax rate [( 1 / 3 ∑ 𝐸𝑇 𝑅 𝑡 − 𝜏 3
𝜏 = 1
)-ETRt]*EARN_CH, where ETR = tax expense (Compustat: txt)/EBT
(Compustat: pi+am) and EARN_CH = change in EPS (Compustat: epspi) divided by the market value of
equity per share at the end of fiscal year t-1 (Compustat: prcc_f)
AB_EQ = 0 for LIFO, 1 for FIFO or other;
AB_AQ = 0 for unqualified, 1 for qualified or other;
AB_SandA = change in selling and admin expense (Compustat: xsga) – change in sale; and
AB_LF = (past sales/past employee (Compustat: emp))-sale/employee)/(past sale/past employee).
40
Table 4: STEM firms, the DuPont components, and future profitability
Panel A: Comparison between STEM vs Non-STEM firms
(1) (2)
RNOA_CHF RNOA_CHF
PM_CH*STEM 0.0621
***
0.0612
***
(2.83) (2.79)
PM_CH -0.00567 -0.00463
(-0.58) (-0.47)
ATO_CH 0.00639
**
0.00660
***
(2.52) (2.66)
STEM -0.00434 -0.00450
(-0.42) (-0.44)
RNOA -0.451
***
-0.450
***
(-40.19) (-39.43)
RNOA_CHL -0.0128 -0.0134
(-1.30) (-1.36)
NOA_CH -0.0164
***
(-3.96)
WC_CH -0.202
***
(-6.97)
NCO_CH -0.0576
***
(-4.34)
FIN_CH -0.00906
(-0.60)
Constant 0.0270
***
0.0274
***
(4.56) (4.64)
N 111584 111584
adj. R
2
0.180 0.181
AB controls Yes Yes
Firm FE &Year FE Yes Yes
This table reports results of the association between the DuPont components and future changes in RNOA between STEM firms
and Non-STEM firms (Model (3) and (4)). All continuous variables are winsorized at 1%. Standard errors are clustered by firm. t
statistics in parentheses
*
p < 0.1,
**
p < 0.05,
***
p < 0.01
Variables are defined as follows:
STEM = an indicator variable equal to 1 if a firm operates in STEM industries, including Science,
Technology, and Internet. Following Fedyk et al. (2017), I identify STEM firms using a combination of SIC code and a list of
companies from Jay Ritter’s website.
NOA (Net operating assets) = Operating assets minus operating liabilities; Operating assets is calculated as total assets
(Compustat item: at) less less cash and short-term investment (Compustat item: che); Operating liabilities is calculated as total
assets (Compustat item: at) less total debt (Compustat item: dlc + dltt) less book value of total common (Compustat item: ceq)
and preferred equity (Compustat item: upstk) less minority interest (Compustat item: mib);
NOA_CH = (NOAt-NOAt-1)/NOAt-1;
PM (Profit margin) = Operating income (Compustat item: oiadp) divided by Sales (Compustat item: sale);
PM_CH = PMt-PMt-1;
ATO (Asset turnover) = Sales (Compustat item: sale) divided by Average NOAt (NOAt+NOAt-1)/2;
ATO_CH = ATOt-ATOt-1;
RNOA = PM * ATO;
RNOA_CHF = RNOAt+1 –RNOAt;
RNOA_CHL = RNOAt -RNOAt-1;
WC = current operating assets (COA) - current operating liabilities (COL); COA = current assets
(Compustat item: act) – cash and short- term investmen (Compustat item: che) t; COL = current liabilities
(Compustat item:lct) – debt in current liabilities (Compustat item: dlc);
WC_CH = WCt-WCt-1;
NCO = non-current operating assets (NCOA) – non-current operating liabilities (NCOL); NCOA = total
assets (Compustat item: at) – current asset (Compustat item: act) – investments and advances (Compustat item: ivao); NCOL =
total liabilities (Compustat item: lt) – current liabilities (Compustat item: lct)– long-term debt (Compustat item: dltt);
NCO_CH = NCOt-NCOt-1;
FIN = financial assets (FINA) – financial liabilities (FINL); FINA = short-term investment (Compustat
41
item: ivst) + long-term investment (Compustat item: ivao); FINL = long-term debt (Compustat item: dltt)+ debt in current
liabilities (Compustat item: dlc)+ preferred stock (Compustat item: upstk);
FIN_CH = FINt-FINt-1.
Panel B: Conditional regressions
Non-STEM STEM
RNOA_CHF RNOA_CHF RNOA_CHF RNOA_CHF
PM_CH 0.00142 0.00226 0.0560
***
0.0587
***
(0.14) (0.22) (2.61) (2.73)
ATO_CH 0.00373 0.00517
**
0.00816
**
0.00797
**
(1.43) (2.03) (2.42) (2.47)
RNOA -0.488
***
-0.488
***
-0.438
***
-0.434
***
(-31.92) (-31.14) (-29.93) (-29.35)
RNOA_CHL -0.0168 -0.0173 -0.0104 -0.00971
(-1.20) (-1.23) (-0.82) (-0.76)
NOA_CH -0.0358
***
0.00487
(-8.00) (0.70)
WC_CH -0.227
***
-0.154
**
(-7.65) (-2.49)
NCO_CH -0.0992
***
0.0547
**
(-7.21) (2.09)
FIN_CH 0.0117 -0.0319
(0.83) (-1.12)
Constant 0.0514
***
0.0518
***
-0.0140 -0.0152
(9.79) (9.83) (-1.10) (-1.20)
N 76989 76989 34595 34595
adj. R
2
0.203 0.204 0.173 0.173
AB controls Yes Yes Yes Yes
Firm FE &Year FE Yes Yes Yes Yes
This table shows results of the association between the DuPont components and future changes in RNOA conditional on
STEM/Non-STEM firms. All continuous variables are winsorized at 1%. Standard errors are clustered by firm. t statistics in
parentheses
*
p < 0.1,
**
p < 0.05,
***
p < 0.01
Variables are defined as follows:
STEM = an indicator variable equal to 1 if a firm operates in STEM industries, including Science,
Technology, and Internet. Following Fedyk et al. (2017), I identify STEM firms using a combination of SIC code and a list of
companies from Jay Ritter’s website.
NOA (Net operating assets) = Operating assets minus operating liabilities; Operating assets is calculated as total assets
(Compustat item: at) less less cash and short-term investment (Compustat item: che); Operating liabilities is calculated as total
assets (Compustat item: at) less total debt (Compustat item: dlc + dltt) less book value of total common (Compustat item: ceq)
and preferred equity (Compustat item: upstk) less minority interest (Compustat item: mib);
NOA_CH = (NOAt-NOAt-1)/NOAt-1;
PM (Profit margin) = Operating income (Compustat item: oiadp) divided by Sales (Compustat item: sale);
PM_CH = PMt-PMt-1;
ATO (Asset turnover) = Sales (Compustat item: sale) divided by Average NOAt (NOAt+NOAt-1)/2;
ATO_CH = ATOt-ATOt-1;
RNOA = PM * ATO;
RNOA_CHF = RNOAt+1 –RNOAt;
RNOA_CHL = RNOAt -RNOAt-1;
WC = current operating assets (COA) - current operating liabilities (COL); COA = current assets
(Compustat item: act) – cash and short- term investmen (Compustat item: che) t; COL = current liabilities
(Compustat item:lct) – debt in current liabilities (Compustat item: dlc);
WC_CH = WCt-WCt-1;
NCO = non-current operating assets (NCOA) – non-current operating liabilities (NCOL); NCOA = total
assets (Compustat item: at) – current asset (Compustat item: act) – investments and advances (Compustat item: ivao); NCOL =
total liabilities (Compustat item: lt) – current liabilities (Compustat item: lct)– long-term debt (Compustat item: dltt);
NCO_CH = NCOt-NCOt-1;
FIN = financial assets (FINA) – financial liabilities (FINL); FINA = short-term investment (Compustat
item: ivst) + long-term investment (Compustat item: ivao); FINL = long-term debt (Compustat item: dltt)+ debt in current
liabilities (Compustat item: dlc)+ preferred stock (Compustat item: upstk);
42
FIN_CH = FINt-FINt-1.
43
Panel C: Sub-STEM groups: Science, technology, and internet
Science Technology Internet
RNOA_CHF RNOA_CHF RNOA_CHF RNOA_CHF RNOA_CHF RNOA_CHF
PM_CH 0.0400
**
0.0421
**
0.0617
***
0.0636
***
0.136
**
0.132
**
(2.20) (2.31) (5.00) (5.15) (2.32) (2.25)
ATO_CH 0.0253
***
0.0194
***
0.00223 0.00382
***
0.00372 0.00346
(4.39) (3.44) (1.47) (2.62) (0.68) (0.67)
RNOA -0.349
***
-0.349
***
-0.489
***
-0.489
***
-0.531
***
-0.525
***
(-19.78) (-19.61) (-71.55) (-69.98) (-15.98) (-15.75)
RNOA_CHL -0.0381
**
-0.0240 0.00629 0.00692 -0.00280 -0.000302
(-2.45) (-1.55) (1.03) (1.13) (-0.12) (-0.01)
NOA_CH 0.0862
***
-0.0196
***
0.00668
(6.76) (-4.93) (0.37)
WC_CH 0.504
***
-0.290
***
0.886
***
(2.79) (-6.70) (2.61)
NCO_CH 0.415
***
-0.00188 0.118
(5.17) (-0.09) (1.12)
FIN_CH -0.0700 0.0130 -0.0622
(-1.27) (0.68) (-0.56)
Constant -0.0992 -0.0981 0.0216 0.0207 -0.0872 -0.136
(-0.97) (-0.96) (1.15) (1.10) (-0.20) (-0.32)
N 4860 4860 29332 29332 1572 1572
adj. R
2
-0.024 -0.026 0.102 0.102 0.061 0.065
AB controls Yes Yes Yes Yes Yes Yes
Firm FE & Year FE Yes Yes Yes Yes Yes Yes
This table shows results of the sub-STEM group. All continuous variables are winsorized at 1%. Standard errors are clustered by
firm. t statistics in parentheses
*
p < 0.1,
**
p < 0.05,
***
p < 0.01
Variables are defined as follows:
STEM = an indicator variable equal to 1 if a firm operates in STEM industries, including Science,
Technology, and Internet. Following Fedyk et al. (2017), I identify STEM firms using a combination of SIC code and a list of
companies from Jay Ritter’s website.
NOA (Net operating assets) = Operating assets minus operating liabilities; Operating assets is calculated as total assets
(Compustat item: at) less less cash and short-term investment (Compustat item: che); Operating liabilities is calculated as total
assets (Compustat item: at) less total debt (Compustat item: dlc + dltt) less book value of total common (Compustat item: ceq)
and preferred equity (Compustat item: upstk) less minority interest (Compustat item: mib);
NOA_CH = (NOAt-NOAt-1)/NOAt-1;
PM (Profit margin) = Operating income (Compustat item: oiadp) divided by Sales (Compustat item: sale);
PM_CH = PMt-PMt-1;
ATO (Asset turnover) = Sales (Compustat item: sale) divided by Average NOAt (NOAt+NOAt-1)/2;
ATO_CH = ATOt-ATOt-1;
RNOA = PM * ATO;
RNOA_CHF = RNOAt+1 –RNOAt;
RNOA_CHL = RNOAt -RNOAt-1;
WC = current operating assets (COA) - current operating liabilities (COL); COA = current assets
(Compustat item: act) – cash and short- term investmen (Compustat item: che) t; COL = current liabilities
(Compustat item:lct) – debt in current liabilities (Compustat item: dlc);
WC_CH = WCt-WCt-1;
NCO = non-current operating assets (NCOA) – non-current operating liabilities (NCOL); NCOA = total
assets (Compustat item: at) – current asset (Compustat item: act) – investments and advances (Compustat item: ivao); NCOL =
total liabilities (Compustat item: lt) – current liabilities (Compustat item: lct)– long-term debt (Compustat item: dltt);
NCO_CH = NCOt-NCOt-1;
FIN = financial assets (FINA) – financial liabilities (FINL); FINA = short-term investment (Compustat
item: ivst) + long-term investment (Compustat item: ivao); FINL = long-term debt (Compustat item: dltt)+ debt in current
liabilities (Compustat item: dlc)+ preferred stock (Compustat item: upstk);
FIN_CH = FINt-FINt-1.
44
Table 5: Intangible intensity, the DuPont components, and future profitability
(1) (2)
RNOA_CHF RNOA_CHF
PM_CH*Intan_QuintRank 0.0841
***
0.0844
***
(3.25) (3.26)
PM_CH -0.0178
*
-0.0172
*
(-1.75) (-1.70)
ATO_CH 0.00587
***
0.00653
***
(2.67) (3.07)
Intan_QuintRank 0.0148
*
0.0142
*
(1.73) (1.66)
RNOA -0.450
***
-0.449
***
(-39.75) (-39.01)
RNOA_CHL -0.0129 -0.0137
(-1.31) (-1.39)
NOA_CH -0.0164
***
(-3.98)
WC_CH -0.203
***
(-7.00)
NCO_CH -0.0580
***
(-4.35)
FIN_CH -0.00906
(-0.60)
Constant 0.0179
***
0.0186
***
(2.76) (2.88)
N 111584 111584
adj. R
2
0.180 0.181
AB controls Yes Yes
Firm FE &Year FE Yes Yes
This table reports results of Model (5) and (6). All continuous variables are winsorized at 1%. Standard errors are clustered by
firm. t statistics in parentheses
*
p < 0.1,
**
p < 0.05,
***
p < 0.01
Variables are defined as follows:
Intan_QuintRank = Intan_Intensity is yearly ranked into quintile and scaled to range between 0 and 1. Intan_Intensity
is calculated as Total intangible capital divided by total capital. Total intangible capital (Peters and Taylor Total Q: k_int)
includes (1) internally developed intangible capital, R&D capital (Peters and Taylor Total Q: k_int_know) plus Organization
capital (Peters and Taylor Total Q item: k_int_org), and (2) externally purchased intangible capital (Compustat: Intan). Total
capital is calculated as total assets (Compustat: at) plus R&D capital (Peters and Taylor Total Q: k_int_know) plus organization
capital (Peters and Taylor Total Q: k_int_org);
NOA (Net operating assets) = Operating assets minus operating liabilities; Operating assets is calculated as total assets
(Compustat item: at) less less cash and short-term investment (Compustat item: che); Operating liabilities is calculated as total
assets (Compustat item: at) less total debt (Compustat item: dlc + dltt) less book value of total common (Compustat item: ceq)
and preferred equity (Compustat item: upstk) less minority interest (Compustat item: mib);
NOA_CH = (NOAt-NOAt-1)/NOAt-1;
PM (Profit margin) = Operating income (Compustat item: oiadp) divided by Sales (Compustat item: sale);
PM_CH = PMt-PMt-1;
ATO (Asset turnover) = Sales (Compustat item: sale) divided by Average NOAt (NOAt+NOAt-1)/2;
ATO_CH = ATOt-ATOt-1;
RNOA = PM * ATO;
RNOA_CHF = RNOAt+1 –RNOAt;
RNOA_CHL = RNOAt -RNOAt-1;
WC = current operating assets (COA) - current operating liabilities (COL); COA = current assets
(Compustat item: act) – cash and short- term investmen (Compustat item: che) t; COL = current liabilities (Compustat item:lct) –
debt in current liabilities (Compustat item: dlc);
WC_CH = WCt-WCt-1;
NCO = non-current operating assets (NCOA) – non-current operating liabilities (NCOL); NCOA = total
assets (Compustat item: at) – current asset (Compustat item: act) – investments and advances (Compustat item: ivao); NCOL =
total liabilities (Compustat item: lt) – current liabilities (Compustat item: lct)– long-term debt (Compustat item: dltt);
NCO_CH = NCOt-NCOt-1;
FIN = financial assets (FINA) – financial liabilities (FINL); FINA = short-term investment (Compustat
45
item: ivst) + long-term investment (Compustat item: ivao); FINL = long-term debt (Compustat item: dltt)+ debt in current
liabilities (Compustat item: dlc)+ preferred stock (Compustat item: upstk);
FIN_CH = FINt-FINt-1.
46
Table 6: The DuPont components and intangible intensity before and after 2000
(1) (2)
RNOA_CHF RNOA_CHF
PM_CH -0.0126 -0.0122
(-0.78) (-0.76)
Intan_QuintRank 0.0213
**
0.0206
**
(2.36) (2.29)
PM_CH*Intan_QuintRank -0.00312 -0.00243
(-0.07) (-0.05)
Post 0.00356 0.00115
(0.42) (0.14)
Post *PM_CH -0.00844 -0.00818
(-0.43) (-0.42)
Post *Intan_QuintRank -0.0113 -0.0111
(-1.23) (-1.21)
Post *PM_CH*Intan_QuintRank 0.135
**
0.134
**
(2.57) (2.56)
ATO_CH 0.00584
***
0.00649
***
(2.65) (3.05)
Constant 0.0150
**
0.0158
**
(2.37) (2.49)
N 111584 111584
adj. R
2
0.181 0.182
Controls Yes Yes
Firm FE & Year FE Yes Yes
This table shows results of the association between the interaction term (PM_CH*Post*Intan_QuintRank) and future changes in
RNOA before and after 2000 (Model (7) and (8)). All continuous variables are winsorized at 1%. Standard errors are clustered by
firm. t statistics in parentheses
*
p < 0.1,
**
p < 0.05,
***
p < 0.01
Variables are defined as follows:
Post = An indicator variable equal to 1 if year is after 2000, and 0 if year is before or equal to 2000.
Intan_QuintRank = Intan_Intensity is yearly ranked into quintile and scaled to range between 0 and 1. Intan_Intensity
is calculated as Total intangible capital divided by total capital. Total intangible capital (Peters and Taylor Total Q: k_int)
includes (1) internally developed intangible capital, R&D capital (Peters and Taylor Total Q: k_int_know) plus Organization
capital (Peters and Taylor Total Q item: k_int_org), and (2) externally purchased intangible capital (Compustat: Intan). Total
capital is calculated as total assets (Compustat: at) plus R&D capital (Peters and Taylor Total Q: k_int_know) plus organization
capital (Peters and Taylor Total Q: k_int_org);
Post = An indicator variable equal to 1 if year is after 1997 and 0 if year is before or equal to 1997.
High Industry Concentration = An indicator variable equal to 1 if a given TNIC industry concentration measure is higher than
median industry concentration measure in a given year and 0 otherwise.
Yearly rank TNIC industry concentration measure (Hoberg and Phillips, 2016) into high and low industry concentration. High
Industry Concentration is equal to 1 if TNIC industry concentration measure is higher than median and 0 otherwise.
NOA (Net operating assets) = Operating assets minus operating liabilities; Operating assets is calculated as total assets
(Compustat item: at) less less cash and short-term investment (Compustat item: che); Operating liabilities is calculated as total
assets (Compustat item: at) less total debt (Compustat item: dlc + dltt) less book value of total common (Compustat item: ceq)
and preferred equity (Compustat item: upstk) less minority interest (Compustat item: mib);
NOA_CH = (NOAt-NOAt-1)/NOAt-1;
PM (Profit margin) = Operating income (Compustat item: oiadp) divided by Sales (Compustat item: sale);
PM_CH = PMt-PMt-1;
ATO (Asset turnover) = Sales (Compustat item: sale) divided by Average NOAt (NOAt+NOAt-1)/2;
ATO_CH = ATOt-ATOt-1;
RNOA = PM * ATO;
RNOA_CHF = RNOAt+1 –RNOAt;
RNOA_CHL = RNOAt -RNOAt-1;
WC = current operating assets (COA) - current operating liabilities (COL); COA = current assets
(Compustat item: act) – cash and short- term investmen (Compustat item: che) t; COL = current liabilities
(Compustat item:lct) – debt in current liabilities (Compustat item: dlc);
WC_CH = WCt-WCt-1;
47
NCO = non-current operating assets (NCOA) – non-current operating liabilities (NCOL); NCOA = total
assets (Compustat item: at) – current asset (Compustat item: act) – investments and advances (Compustat item: ivao); NCOL =
total liabilities (Compustat item: lt) – current liabilities (Compustat item: lct) – long-term debt (Compustat item: dltt);
NCO_CH = NCOt-NCOt-1;
FIN = financial assets (FINA) – financial liabilities (FINL); FINA = short-term investment (Compustat
item: ivst) + long-term investment (Compustat item: ivao); FINL = long-term debt (Compustat item: dltt)+ debt in current
liabilities (Compustat item: dlc)+ preferred stock (Compustat item: upstk);
FIN_CH = FINt-FINt-1.
48
Table 7: Industry concentration and the DuPont components
Low Industry Concentration High Industry Concentration
RNOA_CHF RNOA_CHF RNOA_CHF RNOA_CHF
PM_CH -0.00901 -0.00861 -0.0420 -0.0398
(-0.45) (-0.43) (-1.15) (-1.09)
Intan_QuintRank 0.0163 0.0165 0.0197 0.0186
(0.78) (0.78) (1.18) (1.12)
PM_CH*Intan_QuintRank 0.0895
**
0.0896
**
0.109 0.106
(2.29) (2.29) (1.53) (1.49)
ATO_CH 0.00817
**
0.00808
**
0.00737
*
0.00883
**
(2.13) (2.22) (1.66) (2.04)
Constant 0.0190 0.0195 0.0514
***
0.0500
***
(1.17) (1.20) (3.89) (3.79)
N 29277 29277 29265 29265
adj. R
2
0.183 0.183 0.207 0.207
Firm FE & Year FE Yes Yes Yes Yes
Controls Yes Yes Yes Yes
This table presents the results on the association between the interaction term (PM_CH*Intan_QuintRank) about future changes
in RNOA under high and low industry concentration environment. Low industry concentration is equal to one when TNIC HHI is
lower than the median TNIC HHI in a particular year, and zero otherwise. All continuous variables are winsorized at 1%.
Standard errors are clustered by firm. t statistics in parentheses
*
p < 0.1,
**
p < 0.05,
***
p < 0.01
The definition of the variables, see previous tables.
Table 8: Product Similarity and the DuPont components
Low Product Similarity High Product Similarity
RNOA_CHF RNOA_CHF RNOA_CHF RNOA_CHF
PM_CH -0.0327 -0.0297 -0.0111 -0.00998
(-0.91) (-0.83) (-0.56) (-0.50)
Intan_QuintRank 0.0297
*
0.0273
*
0.0226 0.0224
(1.84) (1.69) (1.03) (1.02)
PM_CH*Intan_QuintRank 0.0777 0.0748 0.101
***
0.100
***
(1.01) (0.96) (2.68) (2.66)
ATO_CH 0.00638 0.00838
*
0.00836
**
0.00797
**
(1.36) (1.82) (2.29) (2.30)
Constant 0.0696
***
0.0693
***
-0.00976 -0.00860
(5.68) (5.64) (-0.54) (-0.48)
N 29280 29280 29262 29262
adj. R
2
0.207 0.206 0.185 0.185
Firm FE &Year FE Yes Yes Yes Yes
Controls Yes Yes Yes Yes
This table reports the results on the association between the interaction term (PM_CH*Intan_QuintRank) about future changes in
RNOA under high and low product similarity environment. Low product similarity is equal to one if total similarity score is lower
than the median total similarity score in a particular year and zero otherwise.
All continuous variables are winsorized at 1%. Standard errors are clustered by firm. t statistics in parentheses
*
p < 0.1,
**
p <
0.05,
***
p < 0.01
The definition of the variables, see previous tables.
49
Robustness check:
Table 9: Intangible intensity and the relation between the DuPont components and future profitability
(1) (2)
RNOA_CHF RNOA_CHF
Panel A: Intangible capital excluding goodwill
PM_CH*IntanNoGW_QuintRank 0.0846
***
0.0851
***
(3.29) (3.31)
PM_CH -0.0218
**
-0.0212
**
(-2.13) (-2.09)
ATO_CH 0.00573
***
0.00639
***
(2.60) (3.00)
IntanNoGW_QuintRank 0.0320
***
0.0304
***
(3.99) (3.80)
Constant 0.00920 0.0104
(1.45) (1.64)
Panel B: R&D and organization capital
PM_CH*RDSGA_QuintRank 0.0805
***
0.0819
***
(3.05) (3.10)
PM_CH -0.0217
**
-0.0217
**
(-2.07) (-2.07)
ATO_CH 0.00560
**
0.00622
***
(2.54) (2.92)
RDSGA_QuintRank 0.0527
***
0.0515
***
(5.62) (5.52)
Constant -0.00118 -0.000339
(-0.17) (-0.05)
N 111584 111584
Controls Yes Yes
Firm FE & Year FE Yes Yes
This table reports results of Model (5) and (6) estimated by the two alternative measures of intangible intensity. Panel A shows
the results estimated by intangible intensity excluding goodwill (IntanNoGW_QuintRank). Panel B reports the results estimated
by the intensity of R&D and organization capital (RDSGA_QuintRank). IntanNoGW_QuintRank is calculated by yearly ranking
IntanNoGW_Intensity into quintile and scaling to range between 0 and 1. IntanNoGW_Intensity is total intangible capital
excluding goodwill divided by total capital. RDSGA_QuintRank is calculated by yearly ranking RDSGA_Intensity into quintile
and scaling to range between 0 and 1. RDSGA_Intensity is R&D capital plus Organization capital divided by total capital.
All continuous variables are winsorized at 1%. Standard errors are clustered by firm. t statistics in parentheses
*
p < 0.1,
**
p <
0.05,
***
p < 0.01
The definition of the variables, see previous tables.
50
Table 10: The DuPont components and intangible intensity before and after 2000
(1) (2)
RNOA_CHF RNOA_CHF
Panel A: Intangible capital excluding goodwill
PM_CH -0.00682 -0.00656
(-0.33) (-0.32)
IntanNoGW_QuintRank 0.0419
***
0.0402
***
(4.72) (4.55)
PM_CH* IntanNoGW_QuintRank -0.0374 -0.0368
(-0.68) (-0.67)
Post 0.00436 0.00208
(0.52) (0.25)
Post *PM_CH -0.0181 -0.0178
(-0.77) (-0.76)
Post * IntanNoGW_QuintRank -0.0150 -0.0150
(-1.60) (-1.59)
Post *PM_CH* IntanNoGW_QuintRank 0.160
***
0.160
***
(2.59) (2.59)
ATO_CH 0.00570
***
0.00636
***
(2.59) (2.99)
Constant 0.00492 0.00609
(0.79) (0.98)
Panel B: R&D and organization capital
PM_CH -0.00484 -0.00482
(-0.23) (-0.23)
RDSGA_QuintRank 0.0630
***
0.0617
***
(6.09) (5.97)
PM_CH* RDSGA_QuintRank -0.0410 -0.0399
(-0.76) (-0.74)
Post 0.00233 0.0000684
(0.28) (0.01)
Post *PM_CH -0.0219 -0.0219
(-0.91) (-0.92)
Post * RDSGA_QuintRank -0.0153 -0.0152
(-1.61) (-1.60)
Post *PM_CH* RDSGA_QuintRank 0.161
***
0.161
***
(2.62) (2.63)
ATO_CH 0.00556
**
0.00618
***
(2.52) (2.90)
Constant -0.00561 -0.00472
(-0.83) (-0.70)
N 111584 111584
Controls Yes Yes
Year FE & Firm FE Yes Yes
This table shows results of Model (7) and (8) estimated by the two alternative measures of intangible intensity. Panel A shows the
results estimated by intangible intensity excluding goodwill (IntanNoGW_QuintRank). Panel B reports the results estimated by
the intensity of R&D and organization capital (RDSGA_QuintRank). IntanNoGW_QuintRank is calculated by yearly ranking
IntanNoGW_Intensity into quintile and scaling to range between 0 and 1. IntanNoGW_Intensity is total intangible capital
excluding goodwill divided by total capital. RDSGA_QuintRank is calculated by yearly ranking RDSGA_Intensity into quintile
and scaling to range between 0 and 1. RDSGA_Intensity is R&D capital plus Organization capital divided by total capital.
All continuous variables are winsorized at 1%. Standard errors are clustered by firm. t statistics in parentheses
*
p < 0.1,
**
p <
0.05,
***
p < 0.01
The definition of the variables, see previous tables.
51
Table 11: The DuPont component and intangible intensity using Fama-MacBeth regressions
Panel A: STEM vs Non-STEM firms
(1) (2)
RNOA_CHF RNOA_CHF
PM_CH*STEM 0.0891
***
0.0893
***
(4.17) (4.25)
PM_CH -0.0101 -0.0101
(-0.64) (-0.67)
ATO_CH 0.00588
***
0.00814
***
(2.96) (4.10)
STEM -0.0285
***
-0.0298
***
(-4.49) (-4.67)
RNOA -0.205
***
-0.202
***
(-19.08) (-18.93)
RNOA_CHL -0.0375
***
-0.0378
***
(-3.08) (-3.14)
NOA_CH -0.0327
***
(-7.08)
WC_CH -0.235
***
(-7.82)
NCO_CH -0.0947
***
(-5.14)
FIN_CH -0.0503
***
(-3.49)
Constant 0.0219
***
0.0202
***
(7.33) (6.62)
N 111584 111584
adj. R
2
0.1159 0.1159
AB Controls Yes Yes
Panel B: Intangible intensity
(1) (2)
RNOA_CHF RNOA_CHF
PM_CH*Intan_QuintRank 0.102
***
0.103
***
(4.14) (4.25)
PM_CH -0.0193 -0.0193
(-1.14) (-1.19)
ATO_CH 0.00569
***
0.00809
***
(2.81) (4.06)
Intan_QuintRank -0.00709 -0.00789
(-1.17) (-1.32)
RNOA -0.202
***
-0.198
***
(-19.23) (-19.00)
RNOA_CHL -0.0382
***
-0.0387
***
(-3.10) (-3.17)
NOA_CH -0.0334
***
(-7.18)
WC_CH -0.237
***
(-7.90)
NCO_CH -0.0969
***
(-5.37)
FIN_CH -0.0608
***
(-4.09)
Constant 0.0164
***
0.0147
***
(3.93) (3.43)
N 111584 111584
adj. R
2
0.1135 0.1134
AB Controls Yes Yes
52
This table shows results the association between the DuPont components and future changes in RNOA using Fama-MacBeth
regressions. Panel A reports results of Model (3) and (4), whereas Panel B presents results of Model (5) and (6). All continuous
variables are winsorized at 1%. Standard errors are clustered by firm. t statistics in parentheses
*
p < 0.1,
**
p < 0.05,
***
p < 0.01
The definition of the variables, see previous tables.
Abstract (if available)
Abstract
Prior research on DuPont analysis found that changes in asset turnover predicted future profitability whereas profit margin did not. This was attributed to the structural nature of the U.S. economy as focused primarily on manufacturing and asset turnover’s ability to capture physical asset utilization efficiency. The U.S. economy has shifted toward service-and-technology-based industries over the last few years with the surge of STEM (science, technology, engineering, and math) firms. Accordingly, I re-examine the usefulness of DuPont analysis to predict future changes in profitability and whether intangible intensity attributes to the change in the property of DuPont ratio. I find that changes in profit margin are now positively associated with future changes in profitability, consistent with the notion that intangible capital creates barriers to entry and production differentiation. Further, this relation is stronger for STEM firms than non-STEM firms and also more pronounced for high intangible intensity firms than low intangible intensity firms but only in industries with low concentration and high product similarity environment. Taken together, these results highlight how changes in the structural makeup of the economy can change how financial statement analysis is applied and adds to the literature in fundamental analysis.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
“What you see is all there is”: The effects of media co‐coverage on investors’ peer selection
PDF
Essays on the firm and stakeholders relationships: evidence from mergers & acquisitions and labor negotiations
PDF
Picky customers and expected returns
PDF
Understanding the disclosure practices of firms affected by a natural disaster: the case of hurricanes
PDF
Essays in corporate finance
PDF
The changing policy environment and banks' financial decisions
PDF
Internal capital markets and competitive threats
PDF
Insider editing on Wikipedia
PDF
Quality investment and advertising: an empirical analysis of the auto industry
PDF
Three essays in derivatives, trading and liquidity
PDF
Intergenerational succession in family firms in the Philippines
PDF
Real asset liquidity and asset impairments
PDF
Accrual quality and expected returns: the importance of controlling for cash flow shocks
PDF
How large are the coinsurance benefits of mergers?
PDF
Disclosure distance and earnings announcement returns
PDF
Social movements and access to credit
PDF
The term structure of CAPM alphas and betas
PDF
Dissecting the heterogeneity of mouse hematopoietic stem cells in vivo
PDF
Political connections and access to bond capital: Reputation or collusion?
PDF
Integrating the industry sector in STEM learning ecosystems: a multicase study
Asset Metadata
Creator
Pongtepupathum, Suteera
(author)
Core Title
Changing fundamental analysis in the new economy: the case of DuPont analysis and STEM firms
School
Marshall School of Business
Degree
Doctor of Philosophy
Degree Program
Business Administration
Degree Conferral Date
2021-08
Publication Date
07/18/2021
Defense Date
04/26/2021
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
DuPont analysis,Forecasting,fundamental analysis,intangible intensity,OAI-PMH Harvest,profitability,STEM
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Soliman, Mark (
committee chair
), Ahern, Kenneth (
committee member
), Ogneva, Maria (
committee member
)
Creator Email
pongtepu@usc.edu,spongtepupathum@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC15607940
Unique identifier
UC15607940
Legacy Identifier
etd-Pongtepupa-9777
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Pongtepupathum, Suteera
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Repository Email
cisadmin@lib.usc.edu
Tags
DuPont analysis
fundamental analysis
intangible intensity
profitability
STEM