Which Competitive Efforts Lead to Future Abnormal Economic Rents?
Victoria Dickinson
University of Mississippi
and
Gregory A. Sommers
Southern Methodist University
May 2011
Correspondence:
Victoria Dickinson, University of Mississippi – E. H. Patterson School of Accountancy, Conner 204C,
University, MS 38677; Phone: (662) 915-5448; Fax: (662) 915-7483; email: [email protected].
The helpful comments of Mark Bradshaw, Gia Chevis, Asher Curtis, Peter Easton, Dana Hollie, Stephen
Penman, Kirk Philipich, Holly Skaife, Dan Wangerin, Terry Warfield, John Wild, Sunny Yang, in
addition to the workshop participants at Baylor University, Florida State University, Southern Methodist
University, and session participants at the 2007 Institute for Excellence in Corporate Governance
Conference at University of Texas–Dallas, the 2007 University of Wisconsin Doctoral Alumni
Conference, the 2008 American Accounting Association Annual Meeting, the 2008 Financial Economics
and Accounting Conference, and the 2010 Center for Accounting Research and Education Conference are
greatly appreciated.
Which Competitive Efforts Lead to Future Abnormal Economic Rents?
Abstract: Economic theory suggests that firms can impede the normalization of economic rents
relative to industry peers through competitive efforts. Firms‟ rents directly impact profitability as
well as forecasting and valuation assumptions. This paper utilizes realized operating
performance, adjusted by estimates of operational risk, to establish which proxies for competitive
efforts are effective at protecting rents. We demonstrate inclusion of competitive efforts
information significantly improves forecasting of future accounting returns. Several competitive
efforts generalize across industries including power over suppliers and the credible threat of
expected retaliation (Porter, 1980). These efforts are associated with a three percent premium in
risk-adjusted RNOA over industry peers, which is economically significant given that the mean
risk-adjusted RNOA over the sample period is 3.3 percent. Traditional variables used to capture
firms‟ competitive efforts such as proxies for product differentiation, innovation, and capital
requirements, do not result in rents once profitability is risk- and industry-adjusted. Finally,
market participants do not fully impound competitive efforts into price, resulting in associations
of those efforts with future stock returns.
Keywords: forecasting; earnings; competitive efforts; competitive advantage; profitability;
persistence; rates of return; industry- and risk-adjusted.
Data Availability: All data are publicly available from sources identified in the text.
1. INTRODUCTION
Profits and losses signal the existence of excess supply or demand (Mueller, 1986; Stigler,
1963).1 Firms freely respond to these signals: entering/exiting markets or expanding/contracting
operations, leading expected risk-adjusted returns to equalize across firms. However, if firms can
expend efforts to establish competitive advantages (impeding the effects of competition), this
equalization may not obtain, at least in the short run.2 For this reason, competitive efforts directly
impact firm profitability and should be incorporated into analysis, forecasting, and valuation
assumptions.
Our study uses accounting information to operationalize the measurement of competitive
efforts (“efforts”) and abnormal economic rents (“rents”) set forth in prior analytical studies.3
Specifically, we measure the association of rents with accounting proxies for efforts. Our
research goal is twofold: first to identify which competitive efforts are successful in attaining
persistent rents, and second to use that information to improve forecasts of profitability. In the
process of pursuing these goals, we contribute to the literature by providing accounting-based
empirical proxies for competitive efforts and by developing a method to adjust operating
profitability for risk of operations (abnormal rents). Also, we demonstrate that market
participants do not fully impound competitive efforts into price, resulting in associations of those
efforts with future stock returns.
1
2
3
While the economics literature refers to profits and losses, a more precise description is that abnormal profits
and losses arise due to supply and demand conditions.
van Breda (1981) provides a model showing that in the absence of barriers-to-entry, there will be entrants to the
industry until the risk-adjusted return to investments is equalized. Where barriers-to-entry exist, he hypothesizes
that high barriers will be associated with high rates of return.
The term competitive advantage refers to an outcome where a firm is able to realize an economic advantage
over other firms. However, often competitive advantage is also used to refer to the action taken in an attempt to
create the desired outcome. To distinctly identify and separate the action and the outcome, we use competitive
effort to refer to actions taken by firms and abnormal economic rents to refer to the economic outcome.
1
The traditional empirical literature to date has focused on barriers-to-entry, in both a
resource-based view and an industrial organization view, to examine competitive advantage
although no single definition or theory is dominant (McAfee et al., 2004; Carlton, 2004). A
barrier has been defined as an incumbent‟s advantage over potential entrants such that economic
rents can be earned without attracting new entrants into the market (Bain, 1956), while Stigler
(1968) defines a competitive advantage as a cost advantage that has accrued to the incumbent.
Common examples of these competitive advantages include economies of scale, product
differentiation, innovation, and capital requirements.
Porter (1980) proposes an expanded set of competitive advantages (e.g., power over
suppliers and customers, and the credibility of expected threat of retaliation), yet researchers
struggle to operationalize these constructs in empirical research. To measure whether efforts
(both traditional and expanded) result in economic rents, we form accounting-based proxies
capturing the theorized advantages and examine the association with future operating
performance. In doing this, we identify which efforts are effective in establishing persistent
operational rents. Further, extant research has relied on analytical modeling and/or single
industry analyses to study the relation between efforts and rents, whereas we examine potential
efforts across all industries. As a result, we are able to specify which efforts generalize to all
industries, and which efforts beyond those of firms‟ peers are expected to yield increased
profitability (abnormal economic rents). We then measure the improvement in forecasting of
profitability conditional on the identified efforts.
Specifically, we first model operating rents at one-year- and five-years-ahead as a
function of firms‟ expenditures of efforts, and examine whether individual efforts represent
unique and incremental factors contributing to future risk-adjusted operating rents (profitability).
2
Our findings demonstrate that inclusion of proxies for effort significantly increases explanatory
power for future realized operating profitability over benchmark models. We next examine the
persistence of rents conditional on the magnitude of each effort. The ability to sustain rents in the
long run is studied by forming portfolios based on the current level of each effort proxy to
determine whether the effort is associated with positive and sustainable rents in the future. If
convergence to a mean value occurs quickly, then the effort did not result in sustainable rents.
Conversely, evidence of sustained differential operating profitability conditional on increased
magnitude of effort indicates a potential source of rents.4
We find several competitive effort proxies successfully generalize across all industries
including power over suppliers and the ability to credibly signal expected retaliation. These
efforts result in a three percent premium in risk-adjusted return on net operating assets (RNOA)
over industry peers (our measure of abnormal rents), even after five years. This is economically
significant given that the mean risk-adjusted RNOA over a 30 year period is only 3.3 percent.
Effective bargaining power over customers results in modest long-term gains over competitors,
while the remaining effort proxies do not result in economically significant rents. Because
competitive efforts are effective when they delay or prevent convergence from taking place,
knowledge about firms‟ expenditures or efforts is critical for making ex ante forecasts of future
profitability and/or earnings. To that end, we find that the inclusion of competitive efforts
variables improves ex ante forecasts of profitability over benchmarks by over 26 percent.
Finally, we provide evidence that the market does not efficiently price information with respect
4
Convergence is an important property for two reasons. First, because of the underlying economic notion that
firms will enter/exit markets or expand/contract operations until rents are eliminated (i.e., profitability has
converged). Second, an assumption underlying virtually all forecasting and valuation models is that a firm
achieves a steady state (e.g., the terminal value assumption in Lee et al. (1999) and Gebhardt et al. (2001) is that
firms‟ profitability converge to the industry median). If profitability does not converge, then either the forecast
horizon must be extended (i.e., forecast truncation is delayed) and/or growth rates leading up to and used in the
terminal value calculation will be differentially affected.
3
to competitive efforts. Excess market returns exist for nearly each competitive effort proxy
examined throughout the study.
This study contributes to the accounting and economic literature in several important
ways. First, we demonstrate how accounting information captures the distinct efforts underlying
theories of competition. This underscores the basic importance of accounting, which is to assist
users in assessing the amounts, timing, and uncertainty of future cash flows (profitability).
Second, in addition to documenting which efforts lead to persistent rents, we document that the
findings are not simply industry effects. Firms that invest effort beyond their industry peers can
earn above-industry profitability, but this does not hold for every type of competitive effort
examined. Third, our results have managerial implications related to resource allocation in that
managers should allocate scarce resources toward efforts that are expected to result in the
maximum return on operations relative to their peers. Specifically, we find results that refute
conventional wisdom regarding the benefit of several traditional competitive effort candidates
(e.g., product differentiation, innovation, and capital requirements) once the cost of investment
and riskiness of operations are considered.
Fourth, this study contributes to the understanding of firm profitability and forecasting.
Accounting is deemed more useful if it can be used to extract detailed knowledge about
underlying economic events beyond the summary measures presented in the financial
statements.5 Fifth, we demonstrate the importance of focusing on operating profitability (Nissim
and Penman, 2001), which removes the potential bias of financing activities from operating
profitability. We also contribute to the financial statement analysis literature by developing a
5
The accountant‟s understanding of transactional flow and how transactions aggregate to financial statement
amounts provides specialized knowledge to develop financial statement ratios (combinations of amounts that
reach beyond simple summary measures from the financial statements) to capture economic phenomena and
measurements such as competitive efforts and rents.
4
process to control for operating risk. This allows us to adjust operating profitability for the
riskiness of firms‟ operations prior to examination of the effects of competitive efforts.
2. PROFITABILITY, COST OF CAPITAL, AND COMPETITIVE EFFORTS
This section discusses the theories that are central to our study: (i) economic rents (profitability),
(ii) cost of capital for operations, (iii) competitive efforts, and (iv) relation to prior research.
(i) Economic Rents/Profitability
Investors and creditors make better economic decisions when they understand the sources and
persistence of profitability; that is, abnormal economic rents. In this study, we examine two
attributes of economic rents: (1) the economic determinants (source) of the rents, and (2) the
time-series properties (persistence) of the rents. We measure abnormal economic rents using an
ex post measure of operating profitability computed as return on net operating assets (RNOA).
RNOA has been shown to be more relevant for forecasting future profitability than traditional
return on assets or equity (see Fairfield et al., 1996; Nissim and Penman, 2001).6 Additionally,
firms generate sustained value through operations rather than through financing transactions
(Penman, 2010).
Prior research demonstrates that the current level and change in profitability, along with
growth in net operating assets (NOA), explain future profitability (Freeman et al., 1982; Fairfield
and Yohn, 2001). We expand the potential determinants of future profitability to include
traditional competitive effort proxies from the economics literature. Next, we expand the
included set of competitive efforts to accounting-based proxies for constructs unexplored in the
6
RNOA excludes financial assets from the denominator because they are presumed to be valued at their fair
value on the balance sheet. Operating liabilities are subtracted from operating assets because operating
liabilities reflect a source of leverage that can increase profitability of operations.
5
accounting literature (e.g., power over suppliers, power over customers and the credible threat of
expected retaliation) to determine the effect of each variable on future profitability.
Profitability can be biased due to conservative accounting (Feltham and Ohlson, 1995;
Basu, 1997; Zhang, 2000; Liu and Ohlson, 2000; Beaver and Ryan, 2000; Ohlson and JuettnerNauroth, 2005). Examples are research and development and advertising that are required by
U.S. GAAP to be expensed immediately. However, these factors are expected to positively
impact future profitability. The focus of this paper is on explaining future profitability via
competitive efforts as opposed to an exploration of accounting quality (i.e., conservatism). More
importantly, methods of controlling for conservatism such as using the conservative accounting
factor (Penman and Zhang, 2002; Cheng, 2005) use variants of our competitive effort variables
to proxy for conservatism. Therefore, in the study, we maintain the assumption that these
expenditures are a primary driver of future profitability, regardless of their biased accounting
treatment.
(ii) Cost of Capital for Operations
Stigler (1963) reports that profitability displays strong central tendencies over time, but
convergence to common values are incomplete.7 Specifically, he identifies potential impediments
to complete convergence as disturbances related to: (1) shifts in demand, (2) advances in
technology, and (3) macroeconomic factors. An additional reason for non-convergence is
potential differences in the risk of operations across firms. To control for this possibility, we
focus on abnormal economic rents by adjusting profitability for the amount of risk firms incur.
7
Convergence analysis is important because Nissim and Penman (2001) suggest that when profitability is meanreverting, then the task of forecasting future profitability is reduced to forecasting growth in net operating assets
(NOA) and revenue. They demonstrate that future RNOA converges when firms are sorted into portfolios based
on level of current RNOA. However, if competitive efforts prevent convergence, then constant growth rate
assumptions and truncated forecast horizons commonly used in valuation models should be reconsidered.
6
Specifically, firms undertaking greater (lesser) amounts of risk are expected to be compensated
with higher (lower) levels of profitability (van Breda, 1981). Thus, we use ex post risk-adjusted
profitability (RNOARA) to proxy for continuing economic rents.
Penman (2010) points out that it is the cost of capital for operations that is a determinant
of the cost of capital for equity, not vice-versa. In addition, because this study focuses on
operating profitability, the risk adjustment must use a measure of operational risk rather than
firm (equity) risk. The finance literature commonly refers to the risk of operations as the
weighted-average cost of capital (WACC).8 The use of cost of capital for operations in adjusting
operating profitability eliminates risk effects due to leverage or changes in leverage. Importantly,
the level of risk for firm operations does not depend on whether additional debt or equity is used
to fund those operations.
Further, we adjust RNOARA for industry performance to determine whether competitive
efforts delay mean reversion of industry- and risk-adjusted economic rents, allowing us to
generalize our results across all firms, regardless of industry membership. Using industry- and
risk-adjusted measures should negate any common effects within the industry and economy.
Remaining abnormal economic rents are presumed to be the result of firms‟ competitive efforts.
(iii) Competitive Efforts
While the economics literature has developed analytical theory regarding potential competitive
efforts, empirical evidence within accounting or economics with respect to the effectiveness of
these factors is sparse. Overall strategic management is the process that controls and evaluates
business within its participative industries. One aspect of strategy is the development, allocation,
and deployment of resources within firms to generate future profits (i.e., creation and protection
8
This name arises from the common method in which it is calculated based on the costs of capital for debt and
equity.
7
of economic rents). Penrose (1959) pioneered the resource-based view (RBV) school of thought
by describing firms as bundles of productive resources under administrative direction. These
resources include all assets, capabilities, organizational processes, firm attributes, knowledge,
etc. controlled by firms that potentially improve the efficiency and effectiveness of firms‟
operations (Barney, 1991; Daft, 2004). The resource-based view was in sharp contrast to another
prevailing strategic theory, the industrial organization or strategic business unit (SBU) view,
which portrayed firms as dependent on external sources of profitability (i.e., product markets,
industry factors, etc.). The SBU theory was developed by strategists in private industry
(specifically, General Electric) and various consulting groups (e.g., Boston Consulting Group,
McKinsey & Company, etc.) and suffered from the criticism that any advantages at the industry
level are not controlled by individual firms and thus, are subject to competitive arbitrage and
appropriation.
The resource-based view deepened in scope to explain competitive advantages, or the
creation and sustainability of economic rents that accrue to individual firms (Wernerfelt, 1984;
Rumelt, 1984). Barney (1991) explained that a resource must have four attributes to result in a
potentially competitive advantage: (1) it must be valuable by exploiting opportunities or
neutralizing threats, (2) it must be rare among firms‟ current and potential competitors, (3) it
must be imperfectly imitable, and (4) there cannot be equivalent substitutes for the resource.
Further, economic rents can accrue from either resources or capabilities. For example, Amit and
Schoemaker (1993) state that resources are tradable and non-specific to the firm, while
capabilities are firm-specific and are the degree to which firms effectively utilizes resources.
Resources are described as inputs to firms‟ operations used to produce products or
services (Wernerfelt, 1984). Examples of resources include raw materials, capital equipment,
8
skilled human resources, and acquired technology. Barney (1986) suggests that in a competitive
market, economic rents can only stem from resources through luck or because market
participants possess heterogeneous beliefs about a resource‟s value; otherwise the supplier of the
resource extracts the full rent. These acquired resources are contrasted with resources developed
within firms that cannot be bought or sold in factor markets (Dierickx and Cool, 1989) such as
reputation, culture, firm-specific know-how, and values.
Capabilities, on the other hand, are the ability to create coordination and cooperation
between firms and their resources to achieve core competencies (Hamel and Prahalad, 1990).
Capabilities include research and development, excellent customer service, and efficient
manufacturing processes. Hamel and Prahalad (1990) state that competitive advantage is derived
from management‟s ability to consolidate technology and production skills into competencies
that empower individual business units to adapt to a dynamic competitive environment.
Finally, Porter (1980) combines the resource and capabilities views while maintaining
that the importance of product and industry markets affect competitive advantage. Specifically,
he suggests that competition is driven by traditional barriers-to-entry such as economies of scale,
product differentiation and innovation. However, he states that other non-resource factors also
affect competition, namely the availability of substitute products, bargaining power of suppliers
or customers, and the credible threat of expected retaliation. In this study, we begin with the set
of resource-based competitive efforts and expand it considerably to include the additional nonresource factors of competition identified by Porter. Each set of variables is discussed in detail in
Section 3.
(iv) Relation to Prior Research
9
We follow in the spirit of Waring (1996), who examines industry-adjusted persistence of
profitability using several industrial organization variables including economies of scale
(labor/capital ratio in exponential form), sunk costs (R&D and advertising intensity), and excess
capacity (capital intensity). He uses return on total assets as a measure of profitability, instead of
an operating profitability measure such as RNOA, and he does not consider risk of operations in
his return measurement. Further, he limits his analysis to the manufacturing sector. We use rates
of risk-adjusted operating profitability and expand the sample to encompass all industry sectors.
Additionally, we use accounting-based proxies for the traditional advantages used in Waring‟s
study; along with increasing the type and scope of resources and capabilities examined (i.e.,
Porter, 1980).
Villalonga (2004) is similar to our study in that she investigates what firm characteristics
determine the persistence of firm-specific profitability. She also uses the persistence of ex post
profitability to identify the positive effect of competitive advantages on future profitability.
However, she uses Tobin‟s Q to proxy for competitive advantages on the basis that it
distinguishes intangible resources from physical and financial resources (i.e., degree of asset
intangibility is the market value of total assets in excess of their book value, or unrecorded
assets). In our study, we attempt to identify which specific factors make up this broad class of
intangibles, where intangibility is viewed as the difficulty to trade, substitute, or imitate firms‟
resource endowments (Villalonga, 2004). More importantly, we provide accounting-based
proxies to measure specific factors that are reflected in increased future profitability, but may or
may not be included in the market value of firms (an issue we investigate in a later section).
Finally, Cheng (2005) examines barriers-to-entry (one subset of competitive advantage)
in relation to the one-year-ahead persistence of abnormal return on equity (ROE) within a
10
residual income framework. He finds that abnormal ROE increases with industry-level and firmlevel barriers-to-entry. We extend Cheng‟s study of R&D intensity, advertising intensity, and
capital intensity to a larger set of competitive efforts and expand the window to also analyze
five-year-ahead performance. These efforts expend considerable resources, and knowledge of
whether the impact on profitability (where it exists) persists is critical, especially in a
competitive environment. Further, while Cheng (2005) decomposes abnormal ROE (rents) into
industry-level and firm-level components, we adjust rents and efforts to control for industryspecific operating cycles and business models. Finally, while Cheng‟s (2005) risk adjustment
uses the Fama and French (1997) industry-level cost of equity capital, we develop a robust
measure of the cost of capital for operations based on accounting data as discussed in section 3.9
3. RELATION BETWEEN COMPETITIVE EFFORTS AND PROFITABILITY
This section presents the competitive effort proxy variables, control variables and sample
descriptives. The traditional competitive effort variables have been explored in prior literature,
whereas the expanded competitive effort variables are largely unexplored in empirical studies.
(i) Traditional Competitive Effort Proxies
Economies of scale. Economies of scale capture resource capabilities that use technology and
quality manufacturing to convert raw materials into final products. The boost to profitability
(rents) derives from either producing a superior product that will command a price premium; or
by producing efficiently, which minimizes costs while maintaining product performance. Both
elements are reflected in Cost of Sales (CoS), defined as cost of goods sold divided by net sales.
A decrease in CoS captures either an increase in sales while holding costs constant, or a decrease
9
The Fama French industry costs of equity capital are noisy, which is acknowledged by Fama and French‟s
(1997) conclusion that “estimates of cost of equity are distressingly imprecise.”
11
in costs while holding sales constant, either of which is indicative of increased economies of
scale. Therefore, CoS is expected to be negatively related to profitability. Oster (1990) points out
that economies of scale arise from efficient use of assets (i.e., raw materials or production assets)
or from specialization of labor. As such, this variable captures economies of scale that are
incremental to the level of capital deployed.
Product Differentiation. Product differentiation is the ability to establish brand identities,
which represent barriers to new entrants or existing competitors. If firms with differentiated
products continually earn rents, then it must hold that competitors are prevented from developing
close substitutes, which would eliminate the profit advantage of differentiating firms (Caves and
Porter, 1977; Mueller, 1986). Operating profitability is posited to be positively related to product
differentiation. Waring (1996) reports that advertising intensity, as a proxy for product
differentiation, is insignificant for explaining industry-adjusted persistence of profitability,
however it might explain firm-level profitability after consideration of operational risk. The
advertising intensity ratio (AdvInt) is measured as advertising expense divided by net sales.
Innovation. Firms that invest more on innovation (i.e., research and development and
patents) will have higher future profitability if the projects undertaken generate positive net
present value. Innovation (Innov) captures the degree of firms‟ proprietary technology and is
measured as the sum of research and development expense and patent amortization expense
divided by net sales. Waring (1996) reports research and development intensity is positively
associated with industry-adjusted persistence of profitability. Further, Brown and Kimbrough
(2011) find that increases in R&D result in stock returns that possess less commonality
(synchronicity) across firms, which indicates that the benefits of R&D accrue to individual firms.
12
Capital Requirements. When a high level of capital is required in order to compete in the
industry, barriers-to-entry exist. Following Lev (1983), we use capital intensity (CapInt) as a
proxy for capital requirements measured as depreciation expense divided by net sales. 10,11 Firms
with higher levels of capital intensity will have higher rents if the preemptive investment
effectively deters entry (Spence, 1977; 1979). However, if the industry is mature, overinvestment
could result in lower profits stemming from excess capacity (Porter, 1980; Lieberman, 1987).
Therefore, we refrain from making an ex ante prediction regarding the effect of capital intensity
on profitability.
(ii) Expanded Competitive Effort Proxies
Power over Suppliers. Firms are able to establish rents by exerting power over their suppliers in
order to obtain favorable trade terms. We use two measures of power over suppliers: operating
liability leverage and inventory turnover. Operating liability leverage occurs when firms rely on
operating creditors to “float” their operating obligations. Examples include extended credit
payment terms and/or just-in-time inventory practices. The inventory holding costs remain with
suppliers which frees up firms‟ resources for alternative use.
The “quality guarantee” theory of trade credit (Smith, 1987) suggests that more powerful
customers will demand longer credit terms to allow for an inspection of goods before final
payment is made. Additionally, firms with more power over their suppliers are able to arrange
for favorable credit terms and/or delivery arrangements. Both of these factors are captured in
firms‟ operating liability leverage (OLLev), which is measured as operating liabilities divided by
10
11
The proxy chosen in this study represents an “average” commitment to capital and accounts for the
obsolescence of the capital investment. For example, if the bulk of the PPE was purchased early in firms‟ life
cycles, it could be outdated, and new entrants could enter the market with newer and more efficient technology.
As such, the capital intensity measure represents capital maintenance as opposed to strictly capital investment.
Traditionally, the capital intensity ratio uses the sum of depreciation expense and net interest expense in the
numerator. However, interest expense results from firms‟ financing decisions and should not be considered
when focusing on operations. For that reason, only depreciation expense is used in the numerator.
13
net operating assets (Nissim and Penman, 2001). OLLev is expected to be positively related to
profitability.
Conversely, the transaction cost theory of trade credit (Petersen and Rajan, 1997)
suggests that suppliers will offer customers credit terms so as to minimize the quantity of
transactions processed during a period. Thus, high inventory turnover ratios (representing
repeated transactions) indicate that firms have a negotiating advantage over their suppliers.
Normally, inventory turnover is calculated as cost of goods sold divided by inventory. However,
we use the inverse of the inventory turnover in order to eliminate problems arising from small or
zero values of inventory in the denominator. Therefore, because we expect inventory turnover to
be positively related to profitability, our measure, InvTurn (the inverse of the inventory turnover
ratio), should be negatively associated with firm profits.
Power over Customers. Similarly, efforts to exert power over customers also represent a
potential source of competitive gain. We use two measures to capture firms‟ power over
customers: receivables turnover and market share. Porter (1980) defines the bargaining power of
customers as the ability of buyers to erode suppliers‟ profits through actions including bargaining
for higher quality and/or lower prices. However, firms should have bargaining power over their
customers when repeated or frequent transactions exist, which implies limited substitutes for
sellers‟ products (controlling for quality and price). Additionally, higher turnover of receivables
implies shorter credit terms, which limits customers‟ power. For this reason, receivables
turnover, measured as net sales divided by receivables captures power over customers. As with
the inventory turnover, we use the inverse of receivable turnover to mitigate problems with small
or zero values of receivables in the denominator. Therefore, a negative relation between
receivables turnover (ARTurn) and profitability is expected. Our second measure of power over
14
customers is market share (MktShr). Market share is defined as firm-specific sales revenue
divided by total sales revenues for all firms in the industry. Higher market share is another
implication that customers have fewer available substitutes for firms‟ products, and it is expected
to be positively related to profitability.
Credible Threat of Expected Retaliation. Incumbent firms are expected to retaliate against
a new entrant or the expansion of an existing competitor, and this propensity for retaliation
potentially creates rents by deterring entry/expansion. However, the threat of retaliation is
credible only if incumbents can actually execute the retaliation.12 This requires that firms have
sufficient financial flexibility to engage in retaliatory actions. Examples of retaliation include
launching protracted price wars, legal actions, or threatening to expand into rivals‟ geographical
or product markets.13
Firms can credibly signal a threat to retaliate through demonstrated borrowing capacity
and/or the availability of excess funds. We use the inverse of borrowing capacity, financial
leverage (FLev) to proxy for borrowing capacity. Financial leverage is measured as net financial
obligations divided by common stockholders‟ equity. FLev is expected to be negatively related
to future profitability if borrowing capacity deters either entry or expansion of firms‟
competitors. However, it is possible that financial leverage is related to higher future rents for
firms that possess positive net present value growth opportunities (i.e., can boost profits through
12
13
Note that, by definition, it is not possible to observe when a company is successful in deterring attempts at
entry. The existence of the credible threat is expected to be sufficient to prevent entry and thus result in rents. If
the entrant attempts entry, rational firms are expected to retaliate if resources are available.
An example of retaliation includes Dell Computer‟s ability to launch aggressive price cuts to thwart existing
competition. These price cuts are sustainable, at least in the short run, considering Dell‟s cash hoard in excess of
$8.9 billion (Ricadela, 2008; Vance, 2008). Firms can also suffer from competitors‟ abilities to thwart profitlimiting regulation. For example, Whole Foods spent considerable resources to file countersuits against the
Federal Trade Commission‟s blockage of their proposed merger with Wild Oats (Wilke and Kesmodel, 2008).
If the merger was successful, competitors of the two grocers would likely have experienced a decline in future
profitability.
15
leverage). A positive operating spread, defined as RNOA in excess of firms‟ net borrowing costs,
coupled with excess demand for firms‟ goods, can lead to higher future operating profitability.
For this reason, we refrain from making an ex ante directional prediction with respect to the
association between FLev and future operating profitability.
Excessive cash is evidence of firms‟ financial flexibility, and thereby a resource for
potential retaliation. We capture this resource as excess funds (ExFunds), computed as net
financial assets (truncated at zero if firms have net financial obligations) scaled by net operating
assets (NOA), which is expected to be positively associated with operating profits.14 Conversely,
free cash flow theory suggests that a firm will build up excess funds, generally as cash or
marketable securities (both financial assets), when it has exhausted its positive net present value
investment opportunities (Jensen, 1986). For that reason, the effect of excess funds on future
profitability is indeterminable, ex ante.
(iii) Firm Level Controls
Firm age and the size of operations are included as control variables to capture potential
confounding or scale effects. Age is expected to be positively associated with profitability as it
captures firms‟ positions on their learning curve (Spence, 1981), yet age is not within a firm‟s
control. Because our focus is on the purposeful expenditure of competitive efforts to increase
rents, we include age in the models to control for the positive effect on profitability that arises
from learning (i.e., the mere passage of time). Age is calculated as the log of the number of years
since a firm‟s first appearance in the CRSP database.
14
The positive relation between excess funds and profitability would be mechanical if profitability were measured
using net income due to the expected positive return on the net financial assets. However, no relation between
operating profitability and net financial assets is expected unless their existence serves to create rents.
16
Monsen and Downs (1965) suggest that the bureaucratic structure of large firms results in
suboptimal performance. If so, size will have a negative effect on profitability. Conversely, if
size captures market dominance or presence, it will be positively related to future profitability.
Size of operations (OpSize) is computed as the log of the market value of firms‟ operations.
(iv) Industry Adjustment
Existing economics literature examines competitive efforts at the industry level, however, Oster
(1990) points out that firms can free-ride on the actions of industry peers. At the same time, if
firms within an industry earn rents relative to their peers, they must have access to a resource,
technology, or managerial talent that prevents other firms from imitating their actions (Mueller,
1986). For that reason, it is important to examine performance within industries (i.e., controlling
for industry characteristics). We define industry using the Fama-French 48 industry
classifications (Fama and French, 1997), and subtract the industry median profitability to capture
whether firms who employ competitive efforts are able to generate sustained rents relative to
industry peers. We industry-adjust the competitive effort variables using a similar technique to
control for commonality in industry operating cycles, business models, resource requirements,
growth, and technology.
(v) Cost of Capital for Operations
To compute the cost of capital for operations, researchers normally use estimates of the cost of
capital for equity. In this study, we modify the method in Easton and Sommers (2007), in which
they estimate the cost of capital for equity based on realized accounting data, to directly estimate
cost of capital for operations. We begin with a residual operating income framework (Penman,
2010), modified to begin growth from current year residual operating income rather than forming
an expectation of future residual operating income.
17
V
Oper
jt
 NOAjt
OI

jt

 r jOper  NOAjt 1 1  g j 
r
Oper
j
gj

(1)
Notice that in this model, the perpetual growth starts from current residual operating income. The
growth rate implies a residual operating income stream such that the present value of this stream
is equal to the difference between the value of operations and the book value of net operating
assets.
To estimate the cost of capital for operations, equation (1) is transformed into the
following regression relation:
OI jt
NOAjt 1
  0  1
V jtOper  NOAjt
NOAjt 1
 e jt
(2)
where  0  r Oper , 1  r Oper  g  1  g  . As Easton and Sommers (2007) point out, this
regression can be estimated for any portfolio to simultaneously obtain estimates of the ex-ante
cost of capital (in this case for operations) and expected growth rate for the portfolio. In our
implementation, we estimate cost of capital for operations for portfolios formed on quintiles of
size and book-to-market yielding 25 estimates per year.
Because our study focuses on firm operations, the cutoffs for size are formed on market
value of operations, defined as market capitalization of equity plus NFO. The cutoffs for bookto-market are formed on the operations equivalent measure, which is NOA deflated by market
value of operations. These calculations implicitly make the common assumption that the book
value of NFO approximates its market value. Estimates are winsorized at a lower bound of 3
percent. Risk-adjusted RNOA is computed as RNOA less the estimated cost of capital for
operations in a firm‟s size and book-to-market portfolio.
18
(vi) Sample Details
The sample includes NYSE, AMEX and NASDAQ firms, excluding ADRs, with necessary data
on Compustat from 1972 to 2003.15 Variables are defined in the appendix. Firms with sales
revenue, lagged net operating assets (NOA), cost of goods sold, market value of equity, or
common stockholders‟ equity of less than $1 million are excluded to prevent skewness
associated with small denominator effects.16 We replace missing data with zero for advertising
expense, R&D expense, patent amortization expense, depreciation expense, inventory and
receivables.17 The empirical tests are performed on a final sample of 65,220 firm-year
observations. Finally, the top and bottom one percent of the explanatory variables each year were
winsorized to reduce effects of extreme values.
(vii) Summary Statistics
Summary statistics are presented in Table 1. Median RNOA is 10.4 percent before risk
adjustment which is consistent with Nissim and Penman‟s (2001) reported median RNOA of
10.0 percent. Median risk-adjusted RNOA is 3.3 percent which implies the average cost of
capital for operations (WACC) for sample firms is 7.1 percent.
[Insert Table 1 about here]
Table 2 presents a set of benchmark values across the Fama-French 48 industry
classifications (Fama and French, 1997). One of the stated goals of Nissim and Penman (2001) is
15
16
17
Data on intangibles is not available on Compustat until 1970 and two prior years of data are needed to estimate
the cost of capital. This means that the first year of data available for analysis is 1972.
In calculating NOA, a minimal amount of cash, defined as 0.5 percent of sales, is considered necessary to
support operations with the balance considered as a financial asset (Penman, 2010). Our results are invariant to
this computation.
It is unknown whether these variables are actually zero or missing from the Compustat database. However, this
constraint biases against detecting an effect for the variables that are computed from these fields. Zeros were
imputed for the following percentages of the sample observations: advertising expense, 67 percent; R&D and
patent amortization, 43 percent; depreciation expense, one percent; inventory, ten percent; and receivables, two
percent.
19
to provide “a historical analysis of ratios that yields such benchmarks for the equity researcher
…” and Table 2 serves a similar purpose with respect to competitive effort proxies across
industries. The primary observation is that there is considerable variability in the level of each
competitive effort across industries. Secondly, there is also substantial variability in rents, riskadjusted RNOA, across industries.18 Untabulated correlations indicate that multicollinearity
across competitive effort variables is not a concern, meaning that each variable captures a
distinct economic characteristic.
[Insert Table 2 about here]
4. EMPIRICAL RESULTS
We first model profitability (over one-year- and five-year-ahead horizons) as a function of
competitive effort. Next, we sort on the current level of each effort proxy and examine
profitability over five subsequent years to determine whether profitability converges to the mean.
We then investigate whether inclusion of the competitive effort proxies improves forecasts of
one-year-ahead profitability levels and changes. Finally, we examine excess stock returns to
determine if the market understands the role of competitive efforts in future market performance.
(i) Multivariate Analysis
This section provides multivariate evidence as to whether each competitive effort variable has
incremental explanatory power over the remaining variables in determining the level of future
industry- and risk-adjusted RNOA (rents).19 To examine whether variables are incrementally
18
19
The industries with the highest risk-adjusted RNOA are tobacco (10.1 percent), trading (9.4 percent), insurance
(8.3 percent), defense (7.2 percent), printing (5.8 percent), and consumer goods (5.2 percent). The poorest
performing industries are coal (0.9 percent), medical equipment (1.1 percent), precious metals (1.1 percent),
communications (1.2 percent), real estate (1.3 percent), agriculture (1.5 percent), petroleum (1.7 percent), lab
equipment (1.7 percent) and candy (1.7 percent).
Untabulated results for a changes specification are qualitatively similar to the levels results.
20
informative in explaining operating profitability, one-year-ahead industry- and risk-adjusted
RNOA is regressed on a base model of the current level and change in industry- and riskadjusted RNOA and industry-adjusted growth in NOA. Current profitability is known to be
serially correlated with future profitability (Fairfield and Yohn, 2001; Penman and Zhang, 2006).
Additionally, changes in operating profitability can also occur due to a denominator effect, or
growth in NOA, GNOA (Fairfield and Yohn, 2001; Penman and Zhang, 2006). Thus, an industryadjusted NOA growth variable is included in the model to ensure that the effects on profitability
are not driven solely by changes in investment. Firm age and size of operations are included as
control variables. Thus, the base model (Model 1) is:
RA
RNOAtRA
  2 RNOAtRA   3GtNOA   4 Aget   5 OpSizet  
1   0  1 RNOAt
(3)
The coefficient on current profitability is expected to be positive, while the change in
profitability and growth in NOA is expected to be negative, because profitability is meanreverting (Freeman et al., 1982; Fairfield and Yohn, 2001). Next, industry-adjusted values of the
traditional competitive effort variables are introduced to test for incremental effect on
profitability over Model 1. Thus, the traditional competitive effort model (Model 2) is:
RA
RNOAtRA
  2 RNOAtRA   3GtNOA   4 Aget   5 OpSizet   6 CoS t 
1   0   1 RNOAt
 7 AdvIntt   8 Innovt   9 CapInt t  
(4)
Next, industry-adjusted values of the expanded competitive effort variables are added to
the model (Model 3):
RA
RNOAtRA
  2 RNOAtRA   3 GtNOA   4 Aget   5 OpSizet   6 CoS t 
1   0   1 RNOAt
 7 AdvIntt   8 Innovt   9 CapInt t   10OLLevt   11InvTurnt   12 ARTurnt  (5)
 13 MktShrt   14 FLev t   15 ExFunds t  
In the final model, we include both industry-adjusted levels and changes of the
competitive efforts to determine whether any of the efforts are prone to diminishing returns. In
21
other words, if increasing a variable does not increase rents without bound, the sign of the
coefficient on the change variable should display the opposite sign of the coefficient of the level
variable. Therefore, the final regression allowing for diminishing returns (Model 4) is as follows:
RA
RNOAtRA
  2 RNOAtRA   3 GtNOA   4 Age t   5 OpSizet   6 CoS t 
1   0   1 RNOAt
 7 CoS t   8 AdvIntt   9 AdvIntt   10 Innovt   11Innovt   12CapInt t 
(6)
 13CapInt t   14OLLevt   15 OLLevt   16 InvTurnt   17 InvTurnt 
 18 ARTurnt   19 ARTurnt   20 MktShrt   21MktShrt   22 FLev t 
 23FLev t   24 ExFunds t   25 ExFunds t  
The results from estimating the four regression models using one-year-ahead rents are
found in Table 3. The regressions are performed annually, with the coefficients reported as the
mean of the annual coefficients and the t-statistics are based on the standard errors of the time
series to mitigate time-period effects (Fama and MacBeth 1973).20 All of the profitability
variables in the base model are significant and of the predicted sign. Neither of the control
variables (age and size) have a significant effect on one-year-ahead operating profitability.
[Insert Table 3 about here]
When the traditional competitive efforts proxies are added (Model 2), the explanatory
power for one-year-ahead industry- and risk-adjusted RNOA increases by 6.72 percent.21
However, only economies of scale (industry-adjusted CoS) is shown to be associated with
profitability in the expected direction (t = 4.9). However, when firms spend in excess of their
20
21
Results are substantially unchanged when we control for clustering by firm and year (i.g., Gow et al., 2010);
however, we use the Fama-MacBeth method in our presentation because we rely on differences in adjusted Rsquared to make comparisons of explanatory power between models. When using robust standard errors
adjusted for clustering, the adjusted R-squared is computed on the entire sample period (i.e., a pooled adjusted
R-squared for each model). This is problematic when dealing with a long time series (i.e., 1972 to 2003) if the
effects related to the competitive effort proxies are concentrated in certain time periods. By using annual
regressions and reporting our adjusted R-squared as the mean of the annual adjusted R-squareds, we are
preventing time-specific results from unduly influencing the overall sample period results.
The incremental explanatory power of the adjusted R-squared was statistically evaluated across pair-wise
comparisons of the models using the Vuong (1989) test in this and all regressions reported later in the paper.
Each model outperforms the prior specifications (p-values on each Z-Statistic are < 0.01).
22
competitors on advertising, innovation, or capital, the effect on future operating profitability is
negative and significant for innovation and capital (advertising becomes significantly negative
once the expanded variables are added in Models 3 and 4).22 This suggests that in combination,
the traditional competitive efforts possess explanatory power for profitability, but only achieving
economies of scale relative to industry peers will result in increased future operating
profitability.
Model 3 includes the expanded set of competitive effort proxies, which results in a 25.15
percent improvement in explanatory power for operating profitability over the base model.
Additionally, the expanded variables (power over suppliers, power over customers, and expected
threat of retaliation) result in an increase in explanatory power of 17.27 percent over the
traditional model. Power over suppliers (OLLev) relative to industry peers has a positive and
significant effect on one-year-ahead profitability (t = 4.6), but the sign of InvTurn is contrary to
predictions (t = 2.7).23 It is possible that the lower InvTurn is related to higher future profitability
if firms are able to expropriate cost savings that stem from suppliers‟ reductions of transaction
costs (e.g., order quantity discounts). Power over customers also contributes to higher rents but
only through the ARTurn variable (t = 2.8).
Credible threats of expected retaliation are captured by FLev and ExFunds. FLev (the
opposite of borrowing capacity) is significantly, but positively associated with higher rents (t =
4.5). This suggests that one-year-ahead operating profitability improves with an increase in
borrowing, indicating that on average, firms have positive net present value investment
22
23
The results are unchanged when the analysis is performed on a sample that is not industry-adjusted.
The inventory turnover (INVturn) and receivables turnover (ARturn) variables are defined as the inverse of the
traditional measures. This means that the sign of the coefficient should be interpreted as –1 multiplied by the
coefficient. In other words, a significant negative coefficient means the traditional measure is associated with
higher future profitability.
23
opportunities. There is also a positive and significant coefficient on ExFunds (t = 3.0) consistent
with financial reserves acting as a credible signal of expected retaliation and inconsistent with
the free cash flow hypothesis.
The final specification (Model 4) incorporates changes in the competitive effort variables
to capture the notion of diminishing returns. Model 4 increases explanatory power of the model
by 32.53 percent over the base model and by 5.9 percent over Model 3 (the expanded model). Of
the variables that contribute to current operating profitability from Model 3, none display
diminishing returns, however the ΔInvTurn variable takes on the predicted sign and is significant
(t = 4.0). This suggests that increases in inventory turnover from the current level results in
increased power over suppliers which translate into increased rents.
Table 4 repeats the regression analysis from Table 3 but with substituting five-year-ahead
industry- and risk-adjusted RNOA for the dependent variable in the regression models. Results
are expected to differ between the one-year- and five-year-ahead models only if the benefits of
establishing competitive efforts do not persist beyond one year or, conversely, if the rents lag
expenditures beyond one year. As seen in Table 3, the variables in the base model are significant
and of the predicted sign. Also, age now contributes to positive and significant future operating
profitability (t = 2.8). Learning is a cumulative function and accumulated knowledge increases
with firm age (Agarwal and Gort, 2002). When the traditional competitive effort proxies are
added to the model (Model 2), the explanatory power for five-year-ahead industry- and riskadjusted RNOA increases by 35.68 percent. Again, only economies of scale (industry-adjusted
CoS) is shown to be associated with future operating profitability (t = 3.5) in the expected
direction. The coefficients on innovation and capital are significantly negative, similar to the
one-year-ahead model in Table 3.
24
[Insert Table 4 about here]
Model 3 (expanded set of competitive effort proxies) results in a 78.12 percent
improvement in explanatory power over the base model for future operating profitability.
Additionally, the expanded variables (power over suppliers, power over customers, and expected
threat of retaliation) result in an increase in explanatory power of 31.28 percent over the model
including only the traditional variables (economies of scale, product differentiation, innovation,
and capital intensity). In all cases, the increases in adjusted R-squared from adding the
competitive effort variables are much larger as the horizon changes from one-year- to five-yearahead operating profitability (the increase in Model 3 over Model 1 is 78.12 percent for the fiveyear versus 25.15 percent for the one-year specification).
With respect to the individual variables, the results for OLLev (positive and significant at
t = 4.5) are similar to the one-year-ahead results in Table 3. Again, the coefficient on FLev is
positive and weakly significant (t = 1.8), which confirms that deploying funds obtained in
previous periods is related to future industry- and risk-adjusted RNOA. ARTurn and ExFunds
are no longer significant, which suggests that the positive profitability effects of power over
customers and credible threat of retaliation dissipate over time.
In the diminishing returns specification (Model 4) the increase in explanatory power over
the base model is 108.86 percent over the base model and 17.26 percent over Model 3 (the
expanded model). Even though the diminishing returns model increases explanatory power, none
of the variables exhibit significant diminishing returns with the exception of ΔExFunds (t =
2.1).
To recap the results, economies of scale (CoS), power over suppliers (OLLev and
ΔInvTurn), power over customers (ARTurn), and expected threat of retaliation (ExFunds) all
25
lead to increased operating profitability in the subsequent year, but only economies of scale
(CoS) and power over suppliers (OLLev) deliver rents that persist through year t + 5. The
traditional competitive effort variables of product differentiation, innovation, and capital
requirements do not lead to positive future operating profitability within either time horizon. This
is an important result because it shows that once returns are industry- and risk-adjusted, investing
in these efforts beyond competitors will not increase future profitability.
(ii) Convergence Analysis
The next step in our analysis examines convergence to determine whether the current level of
competitive effort results in a sustainable effect on rents such that it will improve forecasting.
This analysis allows: (1) the competitive effort variables a protracted period in which to build
economic benefits, and (2) an examination of the over-time mean-reversion properties of rents
relative to the magnitude of the competitive effort. Previous research documents that profitability
mean-reverts, though not completely, in the cross-section of firms (Freeman et al., 1982;
Fairfield et al., 1996; Fama and French, 2000; Nissim and Penman, 2001). In general, if a
competitive effort is effective, convergence of operating profitability should at least be delayed,
if not prevented entirely.
To assure that our sample is representative of data used in prior research, we first
replicate Nissim and Penman‟s (2001) convergence graph of RNOA over time by forming
portfolios based on the magnitude of current RNOA. For each portfolio, the median RNOA is
computed in the classification year and in each of the following five years.24 Each base year
24
To avoid dependence among observations, Nissim and Penman (2001) used the average of multiple nonoverlapping five-year time series. However, we find that the medians based on non-overlapping intervals are
sensitive to business cycle effects. Therefore, we used overlapping portfolios in our graphs to smooth out
macroeconomic driven effects, but in all cases, the ordering of the portfolios is preserved when using the mean
of non-overlapping intervals and spreads are similar.
26
observation is compared with observations for years t + 1 through t + 5.25 The analysis is then
repeated for risk-adjusted RNOA. The remaining spread between the highest and lowest
portfolios at t + 5 captures the degree to which convergence has not occurred by the end of the
five-year window. Finally, we examine industry- and risk-adjusted RNOA convergence to
investigate whether firms can, on average, maintain excess rents above their peers throughout the
five-year window.
[Insert Figure 1 about here]
Figure 1, Panel A, depicts the convergence of unadjusted RNOA, while Panels B and C
present risk-adjusted RNOA and industry- and risk-adjusted RNOA, respectively. The data in
Panel A of Figure 1 appears similar to that displayed in Nissim and Penman (2001). However,
risk-adjusted RNOA in Panel B of Figure 1 converges to a tighter range by year five than the
non-risk-adjusted RNOA (4.2 percent versus 8.4 percent). This stresses the importance of
adjusting for risk before consideration of the effects of competitive effort on rents. The five-yearahead spread of risk-adjusted RNOA is consistent with long-run growth rates in abnormal
earnings used in valuation research (Claus and Thomas, 2001; Easton et al., 2002). In Panel C,
the range of industry- and risk-adjusted RNOA converges to 4.0 percent after five years and the
distance between the top and second quintiles of firm-year observations is greater than in the
non-industry-adjusted graph. This illustrates the presence of dominant leaders within industries
by maintaining higher operating profitability than their industry peers; that is, persistent
abnormal economic rents exist.
25
Firms are classified by competitive effort portfolio during the base year and convergence patterns are analyzed
over time. As in Nissim and Penman (2001), firms that do not survive the entire time series are dropped when
their associated data no longer appear in Compustat. While this imparts an inherent survival bias in later years,
similar patterns are found for a subsample of firms surviving the entire five year window.
27
Figure 2 examines the convergence patterns of portfolios formed on the magnitude of
each industry-adjusted competitive effort proxy. For each set of industry-adjusted competitive
effort portfolios, the median industry- and risk-adjusted RNOA is computed in the classification
year and in each of the following five years. We first discuss the traditional competitive effort
variables then examine the expanded set of variables.
[Insert Figure 2 about here]
Firms with low versus high CoS relative to industry peers enjoy an initial advantage of
3.6 percent, but by the fifth year, this advantage has decreased to a 0.4 percent spread in
industry- and risk-adjusted operating profitability, which indicates that economies of scale do not
result in large long-run rents (Figure 2, Panel A). Panel B of Figure 2 demonstrates that firms
with higher advertising intensity relative to industry peers see a slight increase in industry- and
risk-adjusted RNOA after five years (spread between high and low portfolio increases from 0.03
percent to 0.41 percent).26,27
Firms with high innovation expenditures relative to industry peers are initially at a
substantial operating profitability disadvantage (at year 0 the spread between high and low is 3.0
percent) but the negative rents partially dissipate by year five (spread equals 1.3 percent) (Figure
2, Panel C). This suggests that the five year window used in this analysis may not be long
enough to capture the full effects of innovation on operating profitability.28 Likewise, firms with
the highest capital intensity relative to industry peers (Figure 2, Panel D) have lower profitability
26
27
28
Untabulated results suggest that the 5-year spread in AdvInt portfolios using unadjusted RNOA is over five
percent. This underscores the importance of the industry- and risk-adjustment to RNOA in making inferences
with respect to persistence of the product differentiation.
For the convergence analysis based on AdvInt, only three groups are reported due to the number observations
with zero advertising expense. In later analyses, firms with zero values of Innov or ExFunds are similarly
grouped.
Lev and Sougiannis (1996) find that useful lives on R&D ranged from five years for the scientific instrument
industry and nine years for the chemical industry which is consistent with a longer window being necessary to
fully realize the benefits of innovative expenditures.
28
(difference of 2.6 percent) but the spread shrinks to 1.4 percent by the fifth year. This result casts
doubt on traditional economic thinking that capital requirements pose a barrier-to-entry. Instead,
the risk of excess capacity appears to dampen operating profitability, even after five years.
Power over suppliers results are presented in Panels E and F of Figure 2. Firms with
higher OLLev relative to industry peers enjoy positive rents and this result continues to be
substantial after five years (the spread between the highest and lowest portfolio is 2.7 percent).
InvTurn provides a consistent benefit throughout the five years, albeit more modest than OLLev
(range of high relative to low InvTurn firms after five years is 0.8 percent). This suggests that the
quality guarantee theory of trade credit (OLLev) is more descriptive than the transaction theory
(InvTurn) with regard to maximization of firm rents. Power over customers relative to industry
peers, represented by ARTurn (Figure 2, Panel G) and MktShr (Figure 2, Panel H) display
modest operating profitability gains over the five year interval, resulting in a year five spread of
1.3 percent and 0.7 percent, respectively.
The last group of competitive effort variables relate to firms‟ credible threats of expected
retaliation relative to industry peers. Borrowing capacity (the negative of FLev) (Figure 2, Panel
I) and ExFunds (Figure 2, Panel J) result in persistent rents, even in the long-run. Firms with low
FLev continue to outperform those with high FLev by 2.3 percent after five years. Likewise,
firms with higher excess funds continue to outperform those with low reserves by 3.1 percent
after five years yielding economically significant rents.
To summarize, power over suppliers (OLLev) and the credible threat of expected
retaliation (negative of FLev and ExFunds) provide the greatest long-run benefit to operating
profitability with high portfolios outperforming low portfolios by anywhere from 2.3 to 3.1
percent even after five years. Economies of scale (CoS) and power over customers (ARTurn and
29
MktShr) result in increased future operating profitability, but the spread in rents are less dramatic
after five years (0.4 to 0.7 percent). Product differentiation (AdvInt) results in modest
profitability gains after five years (approximately 0.4 percent). In contrast, a five year horizon is
not sufficiently long to observe positive operating profitability from innovation.
(iii) Forecasting Analysis
A central goal of this study is investigating whether inclusion of information regarding
competitive efforts results in forecasting innovations relative to benchmark models of future
profitability. We adopt the base model from Equation 3 to forecast future RNOA (Fairfield and
Yohn, 2001) as the benchmark to evaluate forecast improvement. Three additional forecasting
models are examined: (1) the traditional model, (2) the expanded model, and (3) the diminishing
returns model.
Following prior research (Fairfield and Yohn, 2001), observations are deleted from the
forecasting analysis if any of the following conditions are met: (1) the absolute value of industryand risk-adjusted RNOA is greater than 1.0, or (2) the absolute values of any of the change
variables is greater than 50 percent. In all cases, forecasts are computed using a hold-out sample.
Specifically, each forecast model is estimated using a five-year rolling window (years t6 to
t1), with the resulting coefficients applied against the accounting variables for year t. 29 We
forecast both the level of (Table 5, Panel A) and change in (Table 5, Panel B) industry- and riskadjusted RNOA (forecasted ΔRNOAt+1). With respect to the change in forecasted RNOA, the
forecasted change is added to actual RNOAt. The result is a forecast for RNOAt+1 which is
compared to the actual RNOAt+1 for each firm-year.
[Insert Table 5 about here]
29
To minimize variation due to business cycle effects, we restrict our out-of-sample forecast period to the years
2001 through 2003. This means that the estimation period ranges from 1995 through 2002.
30
A firm-year forecast error is computed for each forecasting model as the square of
(Actual RNOAt+1 less Forecasted RNOAt+1), or the mean squared forecast error (MSFE). Next,
the firm-year MSFE of each model is subtracted from the corresponding firm-year MSFE of the
benchmark model to compute firm-year pairwise differences in MSFE, presented in Table 5.
Panel A presents the MSFE differences between the base model and the three alternative models
for level of industry- and risk-adjusted RNOA (forecasted ΔRNOAt+1). All three models lead to
significant forecast improvements over the base model for forecasting future level of RNOA.
Panel B examines the change model. The expanded and diminishing returns models represent
significant forecasting improvements over the base model.
Inclusion of the competitive effort proxies represents a statistical and economically
significant forecasting improvement over the base model. The base model of levels (changes) has
a MSFE of 0.0042 (0.0060). The improvement of 0.0011(0.0009) for the levels (changes)
represents a 26.19 percent (15.00 percent) reduction in MSFE. This clearly demonstrates that
information about contemporaneous competitive efforts results in superior forecasts of future
profitability.
(iv)Predicting Future Stock Returns
Previous sections demonstrate that our competitive effort proxies successfully capture
differences across firms that result in persistent profitability into future periods. Results, thus far,
indicate that the highest level of profitability is attained (and persists) in the power over suppliers
(OLLev) and credible threat of expected retaliation variables (Figure 2, Panels E, I, and J).
However, investors could potentially undervalue firms‟ competitive efforts if they do not fully
recognize the valuation implications (i.e., recognize the future cash flow impacts related to
current competitive efforts via the financial statement signal). If this potential undervaluation
31
arises, then the various competitive effort proxies should be associated with future excess
returns. This finding would indicate that the market inefficiently underestimates the persistence
of RNOA as a result of firms‟ expenditures on competitive efforts. On the other hand, if
investors recognize the economic fundamentals related to competitive efforts, this signal should
be impounded into price and future excess returns related to the competitive effort proxies will
be indistinguishable from zero (i.e., the market is efficient with respect to the information
contained in effort expenditures).
We examine whether investors recognize the persistent profitability related to firms‟
competitive efforts by computing one-year-ahead buy-and-hold size- and book-to-marketadjusted returns. We form portfolio returns based on both market capitalization decile and bookto-market decile (resulting in 100 portfolios returns) each month and subtract the appropriate
portfolio return from a firm‟s raw monthly return to compute the excess return. We accumulate
excess returns from the beginning of the fourth month of year t+1 through the third month of
year t+2. This procedure ensures that published financial statement data is available to investors
prior to portfolio formation so they can assess firms‟ competitive effort proxies. Delisting returns
are used whenever possible and delisted firms without a corresponding delisting return in the
CRSP database are assumed to have a return of zero (Piotroski, 2000) unless the delisting
occurred for performance-related reasons, in which case the missing delisting return is set to
negative 100 percent.30
We regress excess returns on the competitive effort proxies and report the results from
estimating the regression with reported standard errors robust to clustering by firm and year
30
Prior research has used -30 percent for NYSE and AMEX firms (Shumway, 1997) and -55 percent for NASDAQ
firms (Shumway and Warther, 1999) that are delisted for performance. Sloan (1996) uses -100 percent for
performance-related delistings. Because missing delisting returns for mergers are likely to be understated when
setting to zero, I make no upward correction for merger-related missing returns to bias against finding results.
32
(Gow et al., 2010) in Table 6, Panel A. The regressions show that overall investors do not fully
appreciate the persistent profitability related to competitive efforts as evidenced by eight proxies
having significant explanatory for future returns (cost of sales, innovation, capital intensity,
operating leverage, inventory turnover, market share, financial leverage, and excess funds).
However, the signs of the coefficients for cost of sales (CoS) and excess funds (ExFunds) are
contrary to expectations.31
[Insert Table 6 about here]
In Panel B of Table 6, we repeat the prior test using returns and proxies that have been
industry-adjusted by subtracting the annual industry median using Fama and French‟s (1997) 48
industry portfolios. These results would differ from the results in Panel A if the investors‟
recognized within-industry differences in performance across firms. Panel B shows that there are
several changes from the initial results when industry-adjustment is performed. The competitive
effort proxies that continue to have significant explanatory power for future excess returns are
cost of sales, innovation, capital intensity, market share, and financial leverage. In addition,
receivables turnover is now significant.
These results indicate that investors fail to understand the implications of firms‟ efforts to
establish competitive positions relative to these factors even beyond any industry level effects.
Therefore, competitive effort proxies represent an implementable tool to identify stocks that are,
at least temporarily, mispriced.
5. CONCLUSION
31
This result is consistent with Oler and Picconi (2010) that demonstrate that firms that hoard cash are penalized by
the market.
33
Firms have long sought to protect economic rents via competitive efforts. Since these efforts, if
successful, impede the ex post normalization of risk-adjusted returns, they directly impact a
firm‟s future profitability and therefore the assumptions incorporated ex ante into forecasting and
valuation models. We operationalize competitive efforts using accounting-based proxies,
underscoring the importance of information set provided by financial reporting. We then
examine ex-post realizations of performance to establish which expenditures (of resources and
effort) protect rents against potential and existing competitors. Our analyses adjust operating
profitability by estimates of operational risk and industry effects to allow for study of relation of
industry- and risk-adjusted returns (i.e., abnormal economic rents) to competitive efforts.
We demonstrate that the inclusion of competitive effort proxies in future profitability
models significantly increases explanatory power over benchmark models. Further, several
competitive effort variables successfully generalize across the entire sample yielding abnormal
economic rents including power over suppliers and the ability of a firm to credibly signal the
threat of expected retaliation. These proxies result in an over three percent premium in industryand risk-adjusted return on net operating assets (RNOA) over competitors. Thus, they provide an
economically significant advantage given that the mean risk-adjusted RNOA over a 30 year
period is 3.3 percent. Power over suppliers and the ability of a firm to credibly signal the threat
of expected retaliation have been traditionally overlooked in the literature; however, we
demonstrate that the proxies used in this paper are effective capturing their underlying economic
characteristics.
Further, the evidence suggests that many traditional proxies used to capture competitive
efforts such as product differentiation, innovation, and capital requirements do not result in
persistent economic rents for the majority of firms when adjusted for risk and the industry
34
normal level. Expenditures on these variables do not protect operating profitability from
converging to an industry-wide risk-adjusted mean, at least over the five year window examined
in the paper. We also document that the inclusion of the competitive effort proxies used in this
study significantly reduces errors in forecasts of future RNOA by 26.19 percent. Finally, we find
that the market is inefficient in processing information about competitive efforts such that excess
returns are earned in the year following the financial statement signal. Thus, competitive efforts,
as captured by financial accounting data, are important information signals for analysis,
forecasting, and valuation.
35
APPENDIX: VARIABLE DEFINITIONS
Profitability variables follow the definition from Nissim and Penman (2001). Variables are
presented in alphabetical order:
Accounts Receivable Turnover (ARTurn) = Accounts receivable (Compustat #2) / Net sales
(Compustat #12)
Advertising Intensity (AdvInt) = Advertising expense (Compustat #45) / Net sales (Compustat
#12)
Age (Age) = Log of the number of years since a firm first appears in CRSP.
Capital Intensity (CapInt) = Depreciation expense (Compustat #14) / Net sales (Compustat #12)
Common Equity (CSE) = Total common equity (Compustat #60) plus preferred treasury stock
(Compustat #227) minus preferred dividends in arrears (Compustat #242)
Comprehensive Net Income (CNI) = net income (loss) (Compustat #172) minus preferred
dividends (Compustat #19) plus the change in the marketable securities adjustment (Δ in
Compustat #238) plus the change in the cumulative translation adjustment in retained
earnings (Δ in Compustat #230)
Cost Margin (CoS) = Cost of goods sold (Compustat #41) / Net sales (Compustat #12)
Excess Funds (ExFunds) = Financial Assets (FA) / Net operating assets (NOA) if FA > NOA,
else equal to zero
Financial Assets (FA) = Cash and short-term investments (Compustat #1) plus Long-term
receivables, investments and advances to affiliated companies (Compustat #32)
Financial Leverage (FLev) = Net Financial Obligation (NFO)/Common Equity (CSE)
Financial Obligations (FO) = Debt in current liabilities (Compustat #34) plus total long-term
debt (Compustat #9) plus preferred stock (Compustat #130) minus preferred treasury
stock (Compustat #227) plus preferred dividends in arrears (Compustat #242)
Growth in Net Operating Assets (GNOA) = (Net Operating Assets (NOAt)/ Lagged Net Operating
Assets (NOAt-1)) – 1
Innovation Intensity (Innov) = [Research and development expense (Compustat #46) plus Patent
amortization expense (Compustat #65)] / Net sales (Compustat #12)
Inventory Turnover (InvTurn) = Inventory (Compustat #3)/ Cost of goods sold (Compustat #41)
36
Marginal Tax Rate = Applicable highest federal tax rate + .02 to approximate state taxes. This
definition is taken from Nissim and Penman (2001). The federal tax rates applicable to
this sample are 34% for years 1989 – 1992 and 35% for 1993 – 2002.
Market Share (MktShr) = Net sales (Compustat #12) / Total net sales in the Fama and French‟s
48-industry portfolio.
Net Financial Expense (NFE) = (Interest expense (Compustat #15) * (1 minus the marginal tax
rate)) plus preferred dividends (Compustat #19) minus (Interest income (Compustat #62)
* (1 minus the marginal tax rate)) plus lagged marketable securities adjustment
(Compustat #238 t-1) minus current marketable securities adjustment (Compustat #238 t)
Net Financial Obligation (NFO) = Financial Obligations (FO) minus Financial Assets (FA)
Net Operating Assets (NOA) = Net Financial Obligation (NFO) plus Common Equity (CSE)
plus minority interest (Compustat #38) [This definition is used rather than the more
common expression Operating Assets (OA) minus Operating Liabilities (OL) to be
consistent with prior research and due to incomplete data in the Compustat variables
related to operating liabilities.]
Operating Assets (OA) = Total assets (Compustat #6) minus Financial Assets (FA)
Operating Income (OI) = Comprehensive Net Income (CNI) plus Net Financial Expense (NFE)
Operating Liabilities (OL) = Operating Assets (OA) minus Net Operating Assets (NOA)
Operating Liability Leverage (OLLev) = Operating Liabilities (OL)/ Net Operating Assets
(NOA)
Return on Net Operating Assets (RNOA) = Operating Income (OI t)/ Average Net Operating
Assets (NOA)
Risk-Adjusted Return on Net Operating Assets (RNOARA) = RNOA minus the weighted average
cost of capital calculated at the beginning of the year as outlined in Section 2
Size of Operations (OpSize) = log of the sum of market value of equity plus the net financial
obligations.
37
REFERENCES
Agarwal, R., and M. Gort (2002), „Firm and Product Life Cycles and Firm Survival‟, American
Economic Review, Vol. 92, No. 2, pp. 184-190.
Amit, R. and P. Schoemaker (1993), „Strategic Assets and Organizational Rent‟, Strategic
Management Journal, Vol. 14, No. 1, pp. 33-46.
Bain, J. (1956), Barriers to New Competition, Their Character and Consequences in
Manufacturing Industries (Cambridge, MA: Harvard University Press).
Barney, J. (1986), „Strategic Factor Markets: Expectations, Luck and Business Strategy‟,
Management Science, Vol. 32, No. 10, pp. 1231-41.
_______ (1991), „Firm Resources and Sustainable Competitive Advantage‟, Journal of
Management, Vol. 17, No. 1, pp. 99-120.
Basu, S. (1997), „The Conservatism Principle and the Asymmetric Timeliness of Earnings‟,
Journal of Accounting and Economics, Vol. 24, No. 1, pp. 3-37.
Beaver, W. and S. Ryan (2000), „Biases and Lags in Book Value and their Effects on the Ability
of the Book-to-Market Ratio to Predict Book Return on Equity‟, Journal of Accounting
Research,Vol. 38, No. 1, pp. 127-48.
Brown, N. and M. Kimbrough (2011), „Intangible Investment and the Importance of FirmSpecific Factors in the Determination of Earnings‟, Review of Accounting Studies,
forthcoming.
Carlton, D. (2004), „Why Barriers to Entry are Barriers to Understanding‟, American Economic
Review,Vol. 94, No. 2, pp. 466-70.
Caves, R., and M. Porter (1977), „From Entry Barriers to Mobility Barriers: Conjectural
Decisions and Contrived Deterrence to New Competition‟, Quarterly Journal of
Economics, Vol. 91, No. 2, pp. 241-62.
Cheng, Q. (2005) „What Determines Residual Income?‟, The Accounting Review, Vol. 80, No. 1,
pp. 85-112.
Claus, J., and J. Thomas (2001), „Equity Premia as Low as Three Percent? Evidence from
Analysts‟ Earnings Forecasts for Domestic and International Stock Markets‟, The Journal
of Finance, Vol. 56, No. 5, pp. 1629-66.
Daft, R. (2004), Organization Theory and Design (Mason, OH: South-Western Cengage
Learning).
Dierickx, I. and K. Cool (1989) „Asset Stock Accumulation and Sustainability of Competitive
Advantage‟, Management Science, Vol. 35, No. 12, pp. 1501-11.
Easton, P., and G. Sommers (2007) „Effect of Analysts‟ Optimism on Estimates of the Expected
Rate of Return Implied by Earnings Forecasts‟, Journal of Accounting Research, Vol. 45,
No. 5, pp. 983-1015.
, G. Taylor, P. Shroff and T. Sougiannis (2002), „Using Forecasts of Earnings to
Simultaneously Estimate Growth and the Rate of Return on Equity Investment‟, Journal
of Accounting Research, Vol. 40, No. 3, pp. 657-76.
Fairfield, P., R. Sweeney, and T. Yohn (1996) „Accounting Classification and the Predictive
Content of Earnings‟, The Accounting Review, Vol. 71, No. 3, pp. 337-55.
, and T. Yohn (2001), „Using Asset Turnover and Profit Margin to Forecast Changes in
Profitability‟, Review of Accounting Studies, Vol. 6, No. 4, pp. 371-85.
Fama, E., and K. French (1997), „Industry Costs of Equity‟, Journal of Financial Economics,
Vol. 43, No. 2, pp. 153-93.
38
, and K. French (2000), „Forecasting Profitability and Earnings‟, Journal of Business,
Vol. 73, No. 2, pp. 161-75.
, and J. MacBeth (1973), „Risk, Return, and Equilibrium: Empirical Tests‟, Journal of
Political Economy, Vol. 81, No. 3, pp. 607-36.
Feltham, G. and J. Ohlson (1995), „Valuation and Clean Surplus Accounting for Operating and
Financial Activities‟, Contemporary Accounting Research, Vol. 11, No. 2, pp. 689-731.
Freeman, R., J. Ohlson, and S. Penman (1982), „Book Rate-of-Return and Prediction of Earnings
Changes: An Empirical Investigation‟, Journal of Accounting Research, Vol. 20, No. 2,
Part II, pp. 639-53.
Gebhardt, W., C. Lee and B. Swaminathan (2001) „Toward an Implied Cost of Capital‟, Journal
of Accounting Research, Vol. 39, No. 1, pp. 135-76.
Gow, I., G. Ormazabal, and D. Taylor (2010), „Correcting for Cross-Sectional and Time-Series
Dependence in Accounting Research‟, The Accounting Review, Vol. 85, No. 2, pp. 483512.
Hamel, G. and C. Prahalad (1990), „Core Competence of the Corporation‟, Harvard Business
Review, Vol. 68, No. 3, pp. 79-91. (May-June)
Jensen, M. (1986), „The Agency Costs of Free Cash Flows, Corporate Finance and Takeovers‟,
American Economic Review, Vol. 76, No. 2, pp. 323-29.
Lee, C., J. Myers and B. Swaminathan (1999), „What is the Intrinsic Value of the Dow?‟, The
Journal of Finance, Vol. 54, No. 5, pp. 1693-741.
Lev, B. (1983), „Some Economic Determinants of Time-Series Properties of Earnings‟, Journal
of Accounting and Economics, Vol. 5, No. 1, pp. 31-48.
, and T. Sougiannis (1996) „The Capitalization, Amortization, and Value-Relevance of
R&D‟, Journal of Accounting and Economics, Vol. 21, No. 1, pp. 107-38.
Lieberman, M. (1987), „Strategies for Capacity Expansion‟, Sloan Management Review, Vol. 28,
No. 4, pp. 19-27.
Liu, J. and J. Ohlson (2000). „The Feltham-Ohlson (1995) Model: Empirical Implications‟,
Journal of Accounting, Auditing and Finance, Vol. 15, No. 3, pp. 321-31.
McAfee, R., H. Mialon and M. Williams (2004), „What is a Barrier to Entry?‟, American
Economic Review, Vol. 94, No. 2, pp. 461-5.
Monsen Jr., R., and A. Downs (1965) „A Theory of Large Managerial Firms‟, Journal of
Political Economy, Vol. 73, No. 3, pp. 221-36.
Mueller, D. (1986), Profits in the Long Run (Cambridge, MA: Cambridge University Press).
Nissim, D., and S. Penman (2001), „Ratio Analysis and Equity Valuation: From Research to
Practice‟, Review of Accounting Studies, Vol. 6, No. 1, pp. 109-54.
Ohlson, J. and B. Juettner-Nauroth (2005), „Expected EPS and EPS Growth as Determinants of
Value‟, Review of Accounting Studies, Vol. 10, No. 2/3, pp. 349-65.
Oler, D. and M. Picconi (2010), „Implications of Insufficient and Excess Cash for Future
Performance‟, Working Paper (Texas Tech University and Indiana University).
Oster, S. (1990), Modern Competitive Analysis (New York, NY: Oxford University Press).
Penman, S. (2010), Financial Statement Analysis and Security Valuation (New York, NY:
McGraw-Hill/Irwin).
______, and X.-J. Zhang (2002), „Accounting Conservatism, the Quality of Earnings, and Stock
Returns‟, The Accounting Review, Vol. 77, No. 2, pp. 237-64.
39
______, and X. Zhang (2006), „Modeling Sustainable Earnings and P/E Ratios with Financial
Statement Analysis‟, Working Paper (Columbia University and University of California,
Berkeley).
Penrose, E. (1959), The Theory of the Growth of the Firm (New York, NY: Oxford University
Press).
Petersen, M., and R. Rajan (1997), „Trade credit: Theories and Evidence‟, Review of Financial
Studies, Vol. 10, No. 4, pp. 661-92.
Piotroski, J. (2000), „Value Investing: The Use of Historical Financial Statement Information to
Separate Winners from Losers‟, Journal of Accounting Research, Vol. 38, Supplement,
pp. 1-41.
Porter, M. (1980), Competitive Strategy: Techniques for Analyzing Industries and Competitors
(New York, NY: The Free Press).
Ricadela, A. (2008), „Dell‟s Quarter Saved by Cost Cuts‟, Business Week (Online) (November
21).
Rumelt, D. (1984), „Toward a Strategic Theory of the Firm‟, in R. Lamb (ed.), Competitive
Strategic Management (Englewood Cliffs, NJ: Prentice Hall), pp. 556-70.
Shumway, T. (1997), „The Delisting Bias in CRSP Data‟, Journal of Finance, Vol. 52, No. 1, pp.
327-40.
and V. Warther (1999), „The Delisting Bias in CRSP‟s Nasdaq Data and its Implications
for the Size Effect‟, Journal of Finance, Vol. 54, No. 6, pp. 2361-89.
Sloan, R. (1996), „Do Stock Prices Fully Reflect Information in Accruals and Cash Flows about
Earnings?‟, The Accounting Review, Vol. 71, No. 3, pp. 289-315.
Smith, J. (1987), „Trade Credit and Informational Asymmetry‟, Journal of Finance, Vol. 42, No.
4, pp. 863-72.
Spence, M. (1977), „Entry, Capacity, Investment, and Oligopolistic Pricing‟, Bell Journal of
Economics, Vol. 8, No. 2, pp. 534-44.
(1979), „Investment Strategy and Growth in a New Market‟, Bell Journal of Economics,
Vol. 10, No. 1, pp. 1-19.
(1981), „The Learning Curve and Competition‟, Bell Journal of Economics, Vol. 12, No.
1, pp. 49-70.
Stigler, G. (1963), Capital and Rates of Return in Manufacturing Industries (Princeton, NJ:
Princeton University Press).
(1968), The Organization of Industry (Chicago, IL: University of Chicago Press).
van Breda, M. (1981), „The Prediction of Corporate Earnings‟, (Ann Arbor, MI: UMI Research
Press).
Vance, A. (2008), „Cost Cutting Helps Dell Profit Exceed Forecasts.‟, New York Times
(November 21).
Villalonga, B. (2004), „Intangible Resources, Tobin's q, and Sustainability of Performance
Differences‟, Journal of Economic Behavior and Organization, Vol. 54, No. 2, pp. 20531.
Vuong, Q. (1989), „Likelihood Ratio Tests for Model Selection and Non-Nested Hypotheses‟,
Econometrica, Vol. 57, No. 2, pp. 307-33.
Waring, G. (1996), „Industry Differences in the Persistence of Firm-Specific Returns‟, American
Economic Review, Vol. 86, No. 5, pp. 1253-65.
Wernerfelt, W. (1984), „A Resource-Based View of the Firm‟, Strategic Management Journal,
Vol. 5, No. 2, pp. 171-80.
40
Wilke, J. and D. Kesmodel (2008), „Whole Foods Returns FTC‟s Fire – Grocer Files Rare Suit
Against U.S. Agency in Fight Over Wild Oats Merger‟, Wall Street Journal (December
9).
Zhang, X.-J. (2000), „Conservative Accounting and Equity Valuation‟, Journal of Accounting
and Economics, Vol. 29, No. 1, pp. 125-49.
41
TABLE 1
Summary Descriptive Statistics
Profitability
Controls
Traditional
Proxies
Expanded
Proxies
Mean Std Dev 25th Pct Median 75th Pct
RNOA
0.120
0.143
0.053
0.104
0.173
RNOARA 0.034
0.134
-0.033
0.033
0.094
NOA
G
0.128
0.261
-0.022
0.076
0.222
Age
2.504
0.833
1.872
2.499
3.067
OpSize
5.216
1.931
3.807
5.092
6.522
CoS
0.660
0.162
0.562
0.682
0.778
AdvInt
0.010
0.022
0.000
0.000
0.010
Innov
0.018
0.028
0.000
0.002
0.027
CapInt
0.049
0.052
0.021
0.034
0.057
OLLev
0.451
0.306
0.250
0.365
0.544
InvTurn 0.231
0.199
0.079
0.189
0.323
ARTurn 0.173
0.118
0.108
0.154
0.209
MktShr
0.012
0.032
0.001
0.002
0.009
FLev
0.559
0.996
-0.043
0.334
0.851
ExFunds 0.138
0.314
0.000
0.000
0.085
The table presents means of annual amounts for 65,220 firm-year observations over the period 1972 through 2003.
RNOA is operating income divided by net operating assets at the beginning of the year. RNOARA is the return on net
operating assets less the weighted average cost of capital calculated at the beginning of the year. The weighted
average cost of capital is calculated annually for portfolios of firms based on quintiles of book-to-market value of
operations and market value of operations following Easton and Sommers [2007] and assuming net financial
obligations are at market value. GNOA is the percentage increase in net operating assets from the prior to current year.
Age is the age of a firm calculated as the log of the number of years since a firm first appears in CRSP. OpSize is the
size of operations calculated as the log of the sum of market value of equity plus the net financial obligations. CoS is
the cost margin calculated as cost of goods sold divided by net sales. AdvInt is the advertising intensity calculated as
advertising expense divided by net sales. Innov is the innovation intensity calculated as the sum of research and
development expense and patent expense divided by net sales. CapInt is the capital intensity calculated as
depreciation expense divided by net sales. OLLev is the operating leverage calculated as operating assets less net
operating assets divided by net operating assets. InvTurn is the inverse of the inventory turnover ratio calculated as
inventory divided by cost of goods sold for the year. ARTurn is the inverse of the accounts receivable turnover ratio
calculated as accounts receivable divided by net sales for the year. MktShr is firm market share calculated as the
firm net sales divided by the total sales in the Fama and French‟s 48 industry portfolio. FLev is the financial
leverage calculated as net financial obligations divided by common stockholders‟ equity. ExFunds is the amount of
excess funds calculated as net financial assets, if any, divided by net operating assets, else 0. Values of zero were
assumed for observations with missing data for advertising expense, R&D expense, patent amortization expense,
depreciation expense, inventory or receivables. All independent variables were winsorized at the 1st and 99th
percentile.
42
TABLE 2
Competitive Effort Proxies by Industry
Industry
Profitability
RNOARA GNOA
Controls
Age
OpSize
CoS
0.076
2.499
5.092
0.682
0.000
0.002
0.034
0.365
0.189
0.154
0.002
0.334
0.000
0.015
0.095
2.310
5.116
0.718
0.001
0.014
0.042
0.317
0.264
0.137
0.018
0.189
0.051
0.124
0.044
0.066
2.831
5.432
0.704
0.009
0.000
0.025
0.374
0.170
0.083
0.003
0.451
0.000
5.03
0.092
0.017
0.096
2.524
6.278
0.614
0.013
0.006
0.042
0.310
0.120
0.096
0.083
1.204
0.000
9.63
0.122
0.044
0.063
2.646
6.377
0.558
0.068
0.002
0.037
0.378
0.270
0.104
0.035
0.432
0.000
# of Obs
RNOA
2,033.1
0.104
0.033
6.84
0.134
51.59
Candy/Soda
Beer/Liquor
Pooled
Agriculture
Food Products
Tobacco
Traditional Proxies
AdvInt
Innov CapInt
OLLev
InvTurn
Expanded Proxies
ARTurn MktShr
FLev
ExFunds
3.47
0.198
0.101
0.062
3.888
7.331
0.508
0.031
0.004
0.025
0.447
0.532
0.077
0.040
0.694
0.001
Recreation
17.53
0.103
0.032
0.061
2.473
4.189
0.622
0.030
0.013
0.025
0.326
0.291
0.189
0.005
0.351
0.000
Entertainment
26.50
0.111
0.032
0.102
2.178
5.408
0.648
0.020
0.000
0.058
0.275
0.038
0.081
0.008
0.663
0.000
Printing
27.09
0.137
0.058
0.066
2.666
6.175
0.544
0.011
0.002
0.040
0.425
0.105
0.145
0.017
0.291
0.001
Consumer Goods
49.75
0.127
0.052
0.066
2.592
4.867
0.640
0.015
0.009
0.023
0.367
0.252
0.159
0.003
0.229
0.000
Apparel
39.69
0.106
0.044
0.061
2.648
4.283
0.681
0.013
0.000
0.016
0.307
0.305
0.148
0.005
0.281
0.000
Healthcare
27.03
0.107
0.029
0.151
1.928
4.571
0.720
0.000
0.001
0.036
0.252
0.041
0.188
0.009
0.758
0.000
Medical Equip.
56.13
0.113
0.011
0.120
2.205
4.535
0.523
0.004
0.052
0.033
0.298
0.407
0.201
0.006
0.087
0.027
Pharmaceuticals
54.94
0.147
0.029
0.116
2.476
5.963
0.460
0.022
0.072
0.038
0.340
0.396
0.180
0.008
-0.055
0.091
Chemicals
51.72
0.116
0.047
0.067
2.863
6.154
0.661
0.000
0.023
0.038
0.381
0.227
0.167
0.004
0.438
0.000
Plastics
30.56
0.109
0.038
0.074
2.537
4.142
0.704
0.001
0.005
0.035
0.314
0.205
0.150
0.007
0.395
0.000
Textiles
20.53
0.081
0.029
0.053
2.605
4.804
0.776
0.000
0.001
0.032
0.295
0.226
0.155
0.018
0.543
0.000
Bldg. Mat.
73.34
0.107
0.041
0.056
2.792
4.878
0.721
0.000
0.004
0.031
0.321
0.225
0.145
0.003
0.385
0.000
Construction
28.00
0.112
0.041
0.086
2.360
5.085
0.820
0.000
0.000
0.014
0.382
0.233
0.128
0.008
0.565
0.000
Steel
43.72
0.095
0.027
0.050
2.864
5.676
0.795
0.000
0.003
0.037
0.342
0.205
0.141
0.005
0.451
0.000
Fabrication
13.06
0.093
0.029
0.058
2.772
4.221
0.730
0.001
0.002
0.031
0.348
0.215
0.153
0.026
0.451
0.000
Machinery
98.53
0.105
0.030
0.069
2.525
4.687
0.670
0.000
0.021
0.030
0.380
0.313
0.185
0.001
0.265
0.000
Electrical
44.28
0.116
0.032
0.067
2.648
4.400
0.675
0.000
0.022
0.028
0.331
0.303
0.166
0.001
0.245
0.000
Autos
43.34
0.116
0.044
0.081
2.769
5.416
0.781
0.000
0.012
0.028
0.428
0.188
0.147
0.001
0.405
0.000
Aircraft
19.88
0.110
0.041
0.063
3.194
5.824
0.747
0.000
0.019
0.032
0.469
0.294
0.162
0.009
0.419
0.000
6.13
0.120
0.045
0.055
2.899
5.677
0.824
0.000
0.006
0.030
0.675
0.191
0.141
0.049
0.428
0.007
Ships/RR
43
TABLE 2 (Continued)
Competitive Effort Proxies by Industry
Industry
Defense
Precious Metals
Mining
Coal
Petroleum
Utilities
# of Obs
RNOA
Profitability
RNOARA GNOA
Controls
Age
OpSize
CoS
Traditional Proxies
AdvInt
Innov CapInt
OLLev
InvTurn
Expanded Proxies
ARTurn MktShr
FLev
ExFunds
4.38
0.160
0.072
0.108
3.252
6.060
0.786
0.004
0.019
0.030
0.603
0.193
0.149
0.103
0.344
0.038
11.84
0.118
0.011
0.112
2.676
5.745
0.676
0.000
0.000
0.114
0.253
0.242
0.067
0.022
-0.137
0.269
9.91
0.096
0.032
0.060
2.971
5.860
0.735
0.000
0.002
0.068
0.277
0.220
0.143
0.017
0.300
0.000
2.28
0.081
0.009
0.112
2.237
5.906
0.787
0.000
0.003
0.084
0.502
0.107
0.139
0.191
1.092
0.000
94.72
0.096
0.017
0.089
2.614
5.875
0.596
0.000
0.000
0.162
0.336
0.077
0.183
0.000
0.431
0.000
138.59
0.086
0.020
0.061
3.071
6.710
0.720
0.000
0.000
0.075
0.348
0.081
0.130
0.002
1.293
0.000
Communication
41.41
0.088
0.012
0.076
2.416
6.718
0.557
0.003
0.004
0.128
0.307
0.051
0.165
0.002
0.829
0.000
Personal Svcs.
17.19
0.113
0.030
0.097
2.349
4.739
0.670
0.002
0.002
0.041
0.334
0.142
0.137
0.025
0.310
0.001
Business Svcs.
141.03
0.120
0.035
0.098
2.082
4.551
0.662
0.000
0.005
0.035
0.484
0.035
0.196
0.001
0.082
0.042
Computers
69.34
0.118
0.021
0.109
2.164
4.655
0.563
0.002
0.069
0.036
0.454
0.303
0.211
0.001
-0.005
0.150
Electronics
127.44
0.109
0.021
0.110
2.425
4.245
0.638
0.000
0.052
0.039
0.355
0.311
0.182
0.001
0.035
0.056
Lab Equip.
61.16
0.101
0.017
0.082
2.404
4.070
0.550
0.003
0.067
0.031
0.338
0.404
0.216
0.004
0.094
0.050
Paper
52.31
0.112
0.043
0.064
2.770
5.798
0.714
0.000
0.005
0.037
0.344
0.168
0.128
0.005
0.435
0.000
Ship Containers
11.28
0.101
0.038
0.065
2.555
6.277
0.768
0.000
0.004
0.043
0.344
0.164
0.118
0.037
0.687
0.000
Transportation
52.13
0.099
0.034
0.075
2.402
5.493
0.787
0.000
0.000
0.063
0.426
0.027
0.119
0.002
0.610
0.000
Wholesale
88.81
0.098
0.032
0.085
2.419
4.633
0.776
0.000
0.000
0.011
0.405
0.193
0.133
0.001
0.500
0.000
117.50
0.116
0.045
0.102
2.348
5.230
0.699
0.014
0.000
0.017
0.471
0.230
0.022
0.001
0.326
0.000
Rest./Hospitality
37.81
0.109
0.032
0.100
2.158
4.912
0.784
0.020
0.000
0.043
0.249
0.028
0.027
0.006
0.471
0.000
Banking
36.75
0.077
0.018
0.105
2.251
5.975
0.580
0.002
0.000
0.015
0.877
0.068
0.596
0.002
1.609
0.004
Insurance
9.84
0.222
0.083
0.101
2.166
5.391
0.755
0.002
0.002
0.030
0.796
0.040
0.238
0.001
-0.098
0.205
Real Estate
14.25
0.076
0.013
0.075
2.570
4.612
0.622
0.007
0.000
0.049
0.266
0.277
0.140
0.010
0.899
0.000
Trading
44.78
0.174
0.094
0.085
2.223
5.259
0.608
0.002
0.000
0.013
0.348
0.067
0.180
0.004
0.316
0.179
Retail
The table reports means of annual medians for each competitive effort proxy based on Fama and French‟s 48 industry portfolios. For this reason, the number of
observations is reported as the mean number of annual observations, n = 2,033. All variables are as defined in the Appendix and in Table 1.
44
TABLE 3
Regression of One-Year-Ahead Industry- and Risk-Adjusted Return on Net Operating
Assets on Current Industry-Adjusted Competitive Effort Proxies
Pred
±
+
–
+
+
±
–
Intercept
RNOARA
RNOARA
GNOA
Age
OpSize
CoS
CoS
AdvInt
+
AdvInt
Innov
+
Innov
CapInt
±
CapInt
OLLev
+
OLLev
InvTurn
–
InvTurn
ARTurn
–
ARTurn
MktShr
+
MktShr
FLev
±
FLev
ExFunds
+
ExFunds
Adj R2
Percentage increase
in Adj R2 over:
Base:
Coeff
t-stat
0.000
(0.0)
(16.4)
1.169
(-6.6)
-0.229
(-13.0)
-0.194
0.001
(0.4)
0.001
(0.3)
16.57%
Base model
Traditional model
Expanded model
Traditional:
Coeff
t-stat
0.012
(3.5)
(17.8)
1.021
(-5.3)
-0.191
(-13.1)
-0.191
-0.001
(-0.4)
0.002
(1.2)
(-4.9)
-0.260
Expanded:
Coeff
t-stat
0.000
(0.0)
(11.8)
0.885
(-3.9)
-0.147
(-11.8)
-0.151
-0.003
(-0.9)
0.003
(1.3)
(-5.9)
-0.286
-0.194
(-1.7)
-0.299
(-2.6)
-1.246
(-4.8)
-1.274
(-5.8)
-0.869
(-4.7)
-0.854
(-4.6)
0.075
(4.6)
0.036
(2.7)
-0.070
(-2.8)
-0.132
(-1.6)
0.012
(4.5)
0.078
(3.0)
17.69%
6.72%
20.74%
25.15%
17.27%
Diminishing Returns:
Coeff
t-stat
0.001
(0.3)
(11.5)
0.906
(-3.5)
-0.141
(-7.0)
-0.122
-0.002
(-0.5)
0.002
(1.0)
(-5.5)
-0.279
0.032
(0.3)
(-2.3)
-0.286
0.013
(0.0)
(-5.7)
-1.282
0.291
(0.6)
(-4.3)
-0.814
-0.133
(-0.6)
(4.1)
0.063
(4.3)
0.114
(3.0)
0.046
(-4.0)
-0.136
(-3.0)
-0.072
0.053
(1.1)
-0.099
(-1.2)
0.156
(0.3)
(3.3)
0.007
(3.9)
0.044
(2.7)
0.070
(2.6)
0.071
21.96%
32.53%
24.19%
5.90%
Table 3 reports means of annual regression estimates based on the following regression model:
RA
RNOAtRA
  2 RNOAtRA   3GtNOA   4 Aget   5OpSizet   6CoSt 
1   0  1 RNOAt
 7 CoSt   8 AdvIntt   9 AdvIntt  10Innovt  11Innovt  12CapIntt 
13CapIntt  14OLLevt  15OLLevt  16InvTurnt  17InvTurnt 
18 ARTurnt  19ARTurnt   20MktShrt   21MktShrt   22FLevt   23FLevt 
 24ExFunds t   25ExFunds t  
(6)
Coefficients that are significant at p < 0.05 (two-tailed) are highlighted in bold. t-Statistics are computed using Fama and
MacBeth (1973) standard errors. The results present means of annual amounts for 65,220 firm-year observations over the period
1972 through 2003. All variables are as defined in Table 1 and have been adjusted by subtraction of annual industry median using
Fama and French‟s 48 industry portfolios.
45
TABLE 4
Regression of Five-Year-Ahead Industry- and Risk-Adjusted Return on Net Operating
Assets on Current Industry-Adjusted Competitive Effort Proxies
Pred
±
+
–
+
+
±
–
Intercept
RNOARA
RNOARA
GNOA
Age
OpSize
CoS
CoS
AdvInt
+
AdvInt
Innov
+
Innov
CapInt
±
CapInt
OLLev
+
OLLev
InvTurn
–
InvTurn
ARTurn
–
ARTurn
MktShr
+
MktShr
FLev
±
FLev
ExFunds
+
ExFunds
Adj R2
Percentage increase
in Adj R2 over:
Base:
Coeff
t-stat
0.000
(0.2)
(8.9)
0.405
(-3.6)
-0.114
(-5.2)
-0.069
(2.8)
0.008
-0.001
(-0.4)
3.17%
Base model
Traditional model
Expanded model
Traditional:
Coeff
t-stat
0.006
(2.4)
(8.9)
0.316
(-2.8)
-0.091
(-5.3)
-0.068
(2.4)
0.007
0.000
(0.2)
(-3.5)
-0.160
Expanded:
Coeff
t-stat
0.001
(0.3)
(5.6)
0.247
(-2.4)
-0.068
(-3.7)
-0.050
0.005
(1.8)
-0.001
(-0.6)
(-4.3)
-0.182
0.121
(1.0)
0.083
(0.7)
-0.841
(-4.2)
-0.840
(-4.6)
-0.545
(-3.8)
-0.526
(-3.8)
0.048
(4.5)
0.003
(0.2)
-0.014
(-0.8)
0.048
(0.8)
0.004
(1.8)
0.036
(1.3)
4.30%
35.68%
5.65%
78.12%
31.28%
Diminishing Returns:
Coeff
t-stat
0.001
(0.5)
(6.1)
0.266
-0.053
(-1.5)
(-4.9)
-0.083
0.004
(1.4)
0.000
(-0.8)
(-4.4)
-0.186
0.103
(0.8)
0.112
(0.9)
-0.684
(-1.3)
(-4.4)
-0.788
-0.780
(-1.6)
(-4.0)
-0.551
0.403
(1.6)
(5.2)
0.051
-0.014
(-0.5)
0.004
(0.2)
-0.006
(-0.1)
-0.015
(-0.9)
-0.012
(-0.2)
0.019
(0.3)
-0.067
(-0.2)
0.003
(1.3)
(4.7)
0.026
0.042
(1.5)
(-2.1)
-0.051
6.62%
108.86%
53.94%
17.26%
Table 4 reports means of annual regression estimates based on the following regression model:
RA
RNOAtRA
  2 RNOAtRA   3GtNOA   4 Aget   5OpSizet   6CoSt 
 5   0  1 RNOAt
 7 CoSt   8 AdvIntt   9 AdvIntt  10Innovt  11Innovt  12CapIntt 
(6)
13CapIntt  14OLLevt  15OLLevt  16InvTurnt  17InvTurnt 
18 ARTurnt  19ARTurnt   20MktShrt   21MktShrt   22FLevt   23FLevt 
 24ExFunds t   25ExFunds t  
Coefficients that are significant at p < 0.05 (two-tailed) are highlighted in bold. t-Statistics are computed using Fama and
MacBeth (1973) standard errors. The results present means of annual amounts for 54,030 firm-year observations over the period
1972 through 2003. All variables are as defined in Table 1 and have been adjusted by subtraction of annual industry median using
Fama and French‟s 48 industry portfolios.
46
TABLE 5
Analysis of Mean Squared Forecast Errors
Panel A – Pooled Sample – Predicting One-Year-Ahead Level of Industry- and RiskAdjusted RNOA:
Mean Squared Forecast Error (MSFE) of Base Model Minus MSFE of:
Reduction
Model
in MSFE t-stat
Traditional
0.0002
2.6
Expanded
0.0010
7.6
Diminishing Returns
0.0011
8.1
Panel B – Pooled Sample – Predicting One-Year-Ahead Change in Industry- and RiskAdjusted RNOA:
Mean Squared Forecast Error (MSFE) of Base Model Minus MSFE of:
Model
Traditional
Expanded
Diminishing Returns
Reduction
in MSFE t-stat
-0.0001
-0.5
0.0005
2.3
0.0009
4.0
N = 6,685 firm-year observations. Forecasts are out-of-sample based on coefficients estimated over rolling five-year
interval prior to year t. Models are applied to year t variables to arrive at forecasted RNOAt+1 or at forecasted
ΔRNOAt+1 which is then added to RNOAt to compute the forecasted RNOA t+1. The predicted RNOAt+1 is compared
with actual RNOAt-1 to compute the forecast error. Observations are deleted if the absolute value of RNOA is greater
than 1 or if the absolute values of ΔRNOA, or any other change variable is greater than 0.50.
47
TABLE 6
Regression of One-Year-Ahead Buy-and-Hold
Annual Excess Returns on Current Competitive Effort Proxies
Panel A – Predicting One-Year-Ahead Size and Book-to-Market Decile Adjusted Returns
Using Current Competitive Effort Proxies:
Intercept
RNOARA
RNOARA
GNOA
Age
OpSize
CoS
AdvInt
Innov
CapInt
OLLev
InvTurn
ARTurn
MktShr
FLev
ExFunds
Adj R2
Coeff
0.174
-0.063
-0.002
-0.150
0.012
-0.042
0.060
-0.025
0.535
0.125
0.031
-0.042
-0.026
0.359
0.026
-0.039
3.39%
t-stat
(9.1)
(-2.9)
(-1.1)
(-15.9)
(4.4)
(-30.4)
(3.1)
(-0.3)
(4.9)
(2.4)
(3.7)
(-3.3)
(-1.4)
(7.0)
(8.8)
(-4.3)
48
TABLE 6 (Continued)
Regression of One-Year-Ahead Buy-and-Hold
Annual Excess Returns on Current Competitive Effort Proxies
Panel B – Predicting One-Year-Ahead Industry-Adjusted Size and Book-to-Market Decile
Adjusted Returns Using Industry-Adjusted Current Competitive Effort Proxies:
Intercept
RNOARA IA
RNOARA IA
GNOA IA
Age IA
OpSize IA
CoS IA
AdvInt IA
Innov IA
CapInt IA
OLLev IA
InvTurn IA
ARTurn IA
MktShr IA
FLev IA
ExFunds IA
Adj R2
Coeff
0.067
-0.013
-0.004
-0.148
0.014
-0.048
0.101
0.147
0.489
0.150
0.004
-0.011
-0.056
0.721
0.028
-0.017
3.52%
t-stat
(27.3)
(-0.5)
(-0.2)
(-15.8)
(5.2)
(-30.4)
(5.1)
(1.4)
(3.5)
(2.4)
(0.5)
(-0.8)
(-2.8)
(12.7)
(9.4)
(-1.8)
Table 6 reports annual regression estimates based on the following pooled regression models with standard errors
clustered by year and firm:
Panel A:
exret t 1   0  1 RNOAtRA   2 RNOAtRA   3GtNOA   4 Aget   5OpSizet   6CoSt 
 7 AdvIntt   8 Innovt   9CapIntt  10OLLevt  11InvTurnt  12 ARTurnt 
13MktShrt  14 FLevt  15 ExFunds t  
Panel B:
exret tIA1   0  1 RNOAtRAIA   2 RNOAtRAIA   3GtNOAIA   4 AgetIA   5OpSizetIA   6CoStIA 
 7 AdvInttIA   8 InnovtIA   9CapInttIA  10OLLevtIA  11InvTurntIA  12 ARTurntIA 
13MktShrt IA  14 FLevtIA  15 ExFunds tIA  
The results present estimates from pooled regressions of 61,212 firm-year observations over the period 1972 through
2003. Coefficients that are significant at p < 0.05 (two-tailed) are highlighted in bold. Excess returns for each firm
are computed as the firm‟s buy-and-hold annual return minus the buy-and-hold annual return of an equally-weighted
benchmark portfolio matched on size and book-to-market deciles. Returns are accumulated from the beginning of
the fourth month of year t + 1 through the third month of year t + 2. Delisting returns are used when included in the
CRSP database. Delisted firms without a corresponding delisting return are assumed to have a return of -100 percent
in the delisting period when delisted for performance-related reasons (Delisting codes between 200 and 299); zero
percent otherwise (Sloan 1996; Piotroski 2000). All other variables are as defined in Table 1. Industry-adjusting in
Panel B utilizes subtraction of annual industry median using Fama and French‟s 48 industry portfolios.
49
FIGURE 1
Evolution of Return on Net Operating Assets Portfolios
Panel A: Return on Net Operating Assets (RNOA).
0.30
0.25
Return on Net Operating Assets
0.20
0.15
0.10
0.05
0.00
-0.05
-0.10
-0.15
-0.20
0
1
2
3
4
5
Year Relative to Portfolio Formation Year
Low
Mid-Low
Mid
Mid-High
High
Panel B: Risk-Adjusted Return on Net Operating Assets (RNOARA).
0.30
Risk-Adjusted Return on Net Operating Assets
0.25
0.20
0.15
0.10
0.05
0.00
-0.05
-0.10
-0.15
-0.20
0
1
2
3
4
5
Year Relative to Portfolio Formation Year
Low
Mid-Low
Mid
Mid-High
High
50
FIGURE 1 (Continued)
Evolution of Return on Net Operating Assets Portfolios
Panel C: Industry- and Risk-Adjusted Return on Net Operating Assets (RNOARA).
0.30
Risk-Adjusted Return on Net Operating Assets
0.25
0.20
0.15
0.10
0.05
0.00
-0.05
-0.10
-0.15
-0.20
0
1
2
3
4
5
Year Relative to Portfolio Formation Year
Low
Mid-Low
Mid
Mid-High
High
The panels show the evolution of quintiles of RNOA (Panel A), RNOARA (Panel B), and industry-adjusted RNOARA
(Panel C) formed annually over the following five years. Variables are as defined in Table 1 and industry-adjusted
indicates subtraction of annual industry median using Fama and French‟s 48 industry portfolios.
51
FIGURE 2
Evolution of Industry- and Risk-Adjusted Return on Net Operating Assets
by Quintile of Industry-Adjusted Competitive Effort Proxies
Panel A: Cost of Sales (CoS).
0.08
Risk-Adjusted Return on Net Operating Assets
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
-0.01
-0.02
-0.03
0
1
2
3
4
5
4
5
Year Relative to Portfolio Formation Year
Low
Mid-Low
Mid
Mid-High
High
Panel B: Advertising Intensity (AdvInt).
0.08
Risk-Adjusted Return on Net Operating Assets
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
-0.01
-0.02
-0.03
0
1
2
3
Year Relative to Portfolio Formation Year
Low
Mid
High
52
FIGURE 2 (Continued)
Evolution of Industry- and Risk-Adjusted Return on Net Operating Assets
by Quintile of Industry-Adjusted Competitive Effort Proxies
Panel C: Innovation Intensity (Innov).
0.08
Risk-Adjusted Return on Net Operating Assets
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
-0.01
-0.02
-0.03
0
1
2
3
4
5
4
5
Year Relative to Portfolio Formation Year
Low to Mid-Low
Mid
Mid-High
High
Panel D: Capital Intensity (CapInt).
0.08
Risk-Adjusted Return on Net Operating Assets
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
-0.01
-0.02
-0.03
0
1
2
3
Year Relative to Portfolio Formation Year
Low
Mid-Low
Mid
Mid-High
High
53
FIGURE 2 (Continued)
Evolution of Industry- and Risk-Adjusted Return on Net Operating Assets
by Quintile of Industry-Adjusted Competitive Effort Proxies
Panel E: Operating Leverage (OLLev).
0.08
Risk-Adjusted Return on Net Operating Assets
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
-0.01
-0.02
-0.03
0
1
2
3
4
5
Year Relative to Portfolio Formation Year

Panel F: Inventory Turnover 1
Low
Mid-Low
.
Mid
Mid-High
High
InvTurn
0.08
Risk-Adjusted Return on Net Operating Assets
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
-0.01
-0.02
-0.03
0
1
2
3
4
5
Year Relative to Portfolio Formation Year
Low
Mid-Low
Mid
Mid-High
High
54
FIGURE 2 (Continued)
Evolution of Industry- and Risk-Adjusted Return on Net Operating Assets
by Quintile of Industry-Adjusted Competitive Effort Proxies

Panel G: Receivables Turnover 1
ARTurn
.
0.08
Risk-Adjusted Return on Net Operating Assets
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
-0.01
-0.02
-0.03
0
1
2
3
4
5
4
5
Year Relative to Portfolio Formation Year
Low
Mid-Low
Mid
Mid-High
High
Panel H: Market Share (MktShr).
0.08
Risk-Adjusted Return on Net Operating Assets
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
-0.01
-0.02
-0.03
0
1
2
3
Year Relative to Portfolio Formation Year
Low
Mid-Low
Mid
Mid-High
High
55
FIGURE 2 (Continued)
Evolution of Industry- and Risk-Adjusted Return on Net Operating Assets
by Quintile of Industry-Adjusted Competitive Effort Proxies
Panel I: Financial Leverage (FLev).
0.08
Risk-Adjusted Return on Net Operating Assets
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
-0.01
-0.02
-0.03
0
1
2
3
4
5
Year Relative to Portfolio Formation Year
Low
Mid-Low
Mid
Mid-High
High
Panel J: Excess Funds (ExFunds).
0.08
Risk-Adjusted Return on Net Operating Assets
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
-0.01
-0.02
-0.03
0
1
2
3
4
5
Year Relative to Portfolio Formation Year
Low
Mid
High
The panels show the evolution of industry-adjusted RNOARA for quintiles of competitive effort proxies formed
annually over the following five years. All variables are as defined in Table 1 and have been adjusted by subtraction
of annual industry median using Fama and French‟s 48 industry portfolios.
56