The financial contribution of customer-oriented marketing capability

Page 1

ORIGINAL EMPIRICAL RESEARCH

The financial contribution of customer-orientedmarketing capability

Fernando Angulo-Ruiz & Naveen Donthu & Diego Prior &

Josep Rialp

Received: 16 November 2012 /Accepted: 5 August 2013# Academy of Marketing Science 2013

Abstract This article assesses the financial contribution ofmarketing capability. In contrast with previous research, whichconceptualizes marketing capability as the deployment of mar-keting resources to achieve sales, this study conceivesmarketingcapability as the deployment of marketing resources to achievethe ultimate objectives of customer satisfaction and brand equity(i.e., customer-oriented marketing capability [COMC]). Thus,this research disentangles the dynamic relationships amongmarketing resources, sales, customer satisfaction, and brandequity through the use of network Data Envelopment Analysisto capture COMC. According to what the value relevanceperspective proposes, COMC positively influences the growthof Tobin’s q and improves the growth of analysts’ recommen-dations. These findings remain robust and consistent with theuse of additional measures and methods common to the mar-keting and financial literatures. Our study provides tools and aframework for analysis for managers tomaximize their ability touse marketing strategy to drive performance.

Keywords Customer-oriented marketing capability .

Tobin’s q . Analyst recommendations .

Network data envelopment analysis

Howmarketing capabilities influence business performance isone of the key areas of concern for marketing academics andpractitioners, as the Marketing Science Institute’s (2012) re-search priorities demonstrate. A key topic pertaining to thisarea is the way firms deploy resources to serve customersbetter (e.g., Dutta et al. 1999; Morgan 2012). Empirical re-search on marketing capabilities commonly assumes that theprimary objective of marketing resources is to generate sales,though Day (1994) contends that marketing capabilitiesshould center on the customer, such that the objective ofmarketing resources is the creation of customer value in termsof serving customers better and encouraging the market toprefer one brand to another (Bolton 1998; Bolton and Drew1991; Keller 1993; Sheth et al. 2000; Srivastava et al. 2001).This apparent dichotomy between generating sales and creat-ing customer value motivates this research. Our perspective isto integrate sales and customer value in a multi-stage process,where focusing solely on maximizing customer value couldnot necessarily imply maximizing sales. Our assumption isthat firms should employ marketing resources for obtainingthe optimum sales level that leads to maximizing customervalue, and the optimal sales level does not necessarily meanthe maximum sales level. Therefore, marketing resources,sales, and customer value constitute a network process wherethe final purpose (to maximize customer value) determineshowmarketing resources should be employed for reaching theoptimal level of sales that permits the final objective.

In line with the literature on organizational capability andthe marketing chain approach,1 we conceptualize marketingcapabilities from a customer perspective—that is, customer-oriented marketing capability (COMC). In essence, COMCrefers to a firm’s distinctive ability to deploy and allocate

1 The marketing chain approach is a conceptual framework that helpsmeasuremarketing productivity. In particular, it is “a chain-of-effects modelthat relates the specific actions taken by the firm (marketing actions) to theoverall condition and standing of the firm” in terms of customer, market,financial, and firm value impacts (Rust et al. 2004, p. 77).

F. Angulo-RuizSchool of Business, Grant MacEwan University, Room 5-221F,10700-104 Avenue, Edmonton, AB, Canada T5J 4S2e-mail: [email protected]

N. Donthu (*)J. Mack Robinson College of Business, Georgia State University,35 Broad St., Suite 1335, Atlanta, GA 30303, USAe-mail: [email protected]

D. Prior : J. RialpDepartament d’Economia de l’Empresa, Universitat Autònomade Barcelona, Edifici B, Campus UAB, Barcelona, Spain 08193

D. Priore-mail: [email protected]

J. Rialpe-mail: [email protected]

J. of the Acad. Mark. Sci.DOI 10.1007/s11747-013-0353-6

Page 2

marketing resources to achieve the ultimate objectives ofcustomer satisfaction and brand equity (Day 1994; Hunt andMorgan 1995; Keller and Lehmann 2003; Rust et al. 2004;Winter 2000). Customer satisfaction is an indicator of howwell a firm serves its customers, such that well-served cus-tomers are satisfied with the firm’s product (Bolton 1998;Bolton and Drew 1991). Brand equity represents the end resultof getting the customer to prefer one brand to another, suchthat he or she is aware of and has positive associations with thebrand (Keller 1993; Srinivasan et al. 2005). However, not allthe firms are organized around the customer: Day (2006)reports that only 32 % of the companies in his sample wereorganized around the customer, and Gulati (2007, 2009) ar-gues that putting the customer at the center of the organizationis actually a more difficult task to execute than it seems. Thatis why understanding the capabilities of customer-driven or-ganizations and how these capabilities influence financialperformance becomes relevant.

We examine the influence of COMC on firms’ futureearnings directly and through financial analysts . A key aspectof this influence is how stock markets seize marketing capa-bilities information or, in other words, how marketing capa-bilities information is integrated into future earnings. Theliterature recognizes that financial analysts play an importantrole as information intermediaries for investors because theydigest company and market information and provide earningsforecasts and recommendations to buy, hold, or sell a givenstock (Jegadeesh et al. 2004; Ngobo et al. 2012; Womack1996). As such, financial analysts may be a channel throughwhich marketing capabilities information is integrated intoshare prices and, in turn, into future earnings (Luo et al.2010). Therefore, we examine the influence of COMC onanalysts’ stock recommendations. Financial analysts mightconsider COMC because they have better and deeper knowl-edge of the firm capabilities they are analyzing and thus arelikely to use COMC in the elaboration of their recommenda-tions. To date, this aspect has been used in the finance andaccounting literature to link relationships, intangibles, andaccruals to financial performance, though it remains largelyignored in the marketing literature (see Luo et al. 2010; Ngoboet al. 2012). Consequently, this research issue is important forunderstanding howmarketing capabilities influence long-termfinancial performance directly and through analysts’recommendations.

In the next section, we review prior literature on marketingcapabilities and financial performance. Then we develop ourconceptual framework and research hypotheses regardinghow COMC influences financial performance in the longrun. Next, we elaborate a model to capture COMC as wellas models to capture the effect of COMC on performance. Weestimate our empirical models using a merged dataset com-prising firms’ marketing and financial information from Ad-vertising Age and Crain Communications, the American

Customer Satisfaction Index (ACSI), COMPUSTAT, and theInstitutional Brokers Estimate System. We apply networkData Envelopment Analysis (DEA) to measure COMC andpanel data methods to estimate the effect of COMC on finan-cial performance. We also perform several robustness checksof the findings. Finally, we discuss implications for managers,researchers, and marketing theory.

Literature review

In Table 1, we summarize extant literature on marketingcapabilities and its contribution to financial performance andposition our article accordingly. To organize the literature, weground our review on the resource-based view and organiza-tional capability theories. Resource-based theory views a firmas a bundle of tangible and intangible assets, such as physical,human, and organizational resources, that can be used toimplement value-creating strategies to improve efficiencyand effectiveness (Barney 1991; Helfat and Peteraf 2003).Extending resource-based theory, researchers claim that capa-bilities help firms outperform competitors more than re-sources.2 A capability is a combination of resources and isembedded in the organization and its processes (Makadok2001; Teece et al. 1997). Winter (2000, p. 983) conceptualizescapability as a “high-level routine … that, together with itsimplementing input flows, confers upon an organization’smanagement a set of decision options for producing signifi-cant outputs of a particular type.” Similarly, Amit andSchoemaker (1993), Helfat and Peteraf (2003), and Zolloand Winter (2002) assert that a capability refers to the organi-zation’s ability to perform a coordinated set of tasks (with itsorganizational resources) to achieve a particular end result.Therefore, marketing capabilities should feature three mainelements: (1) marketing resources (inputs), (2) marketing-specific end results (outputs), and (3) the process (a coordi-nated set of tasks that connect resources with end results).

Some studies refer to marketing capabilities as the skill ofusing marketing resources, or skills in segmenting andtargeting markets, in advertising and pricing, or in creatingcustomer value (e.g., Krasnikov and Jayachandran 2008;Morgan et al. 2009a; Ramaswami et al. 2009; Song et al.2005, 2007, 2008; Vorhies and Morgan 2005). Other studiesdecompose marketing capabilities into brand managementand customer relationship management capabilities (Morganet al. 2009a; Vorhies et al. 2011). In doing so, these studies“construct marketing capabilities from the bottom up usingskills and organizational routines, as well as micro-level pro-cesses that allow firms to gain competitive advantages” (Morgan2012, p: 106; Vorhies and Morgan 2005). Marketing activities

2 A discussion of the strengths and weaknesses of resource-based andcapability theories is beyond the scope of this article.

J. of the Acad. Mark. Sci.

Page 3

Tab

le1

Com

paring

empiricalliterature

onmarketin

gcapabilitywith

thecurrentresearch

Representative

studies

Conceptualizationof

marketin

gcapability

Resource-based,organizatio

nalcapability

Methodfor

measuring

capability

Impacton

operational

performance

Impacton

forw

ard-looking

performance

Marketin

gresources

Marketin

g-specificendresult

Akdenizetal.

(2010)

Theyfollo

wDuttaetal.’s

(1999)

concept

Advertising;

ME;SRO,C

RSales

DEA,SFE

−−

Duttaetal.

(1999)

“Superiority

inidentifying

custom

ers’needsandin

understanding

thefactorsthatinfluenceconsum

erchoice

behavior”(p.550)

Advertising;

ME;C

RSales

SFE

−To

bin’sq

Krasnikov

and

Jayachandran

(2008)

“Understandandforecastcustom

erneedsbetterthan

competitors

andto

effectivelylin

kofferingsto

custom

ers”

(p.1)

−Marketsensing

andcustom

erlin

king

Meta-analysis

Efficiencyand

market

performance

−

Morganetal.

(2009a)

Theyfollo

wDay’s(1994)

andSrivastavaetal.’s

(1998)

concept

−Marketsensing,branding,

andCRM

competences

Survey

Revenue

grow

th,

margingrow

th,

profitgrow

th

−

Narasim

hanetal.

(2006)

“Excellencewith

which

thefirm

convertsmarketin

gexpenditu

re…

into

such

metrics

assalesor

custom

ersatisfaction”

(p.511)

Advertising;

ME;C

RSales

SFE

Operatin

gincomeon

assets

−

Nathetal.(2010)

“Integrativ

eprocessin

which

afirm

uses

its…

resourcesto

understand

complex

consum

erspecificneeds,achieveproduct

differentiatio

nrelativ

eto

competition,andachievesuperior

brandequity”(p.3)

Advertising;

ME;C

RSales

DEA

Operatin

gincomeon

assets

−

Ram

aswam

ietal.(2009)

Identificationof

custom

ers,creatio

nof

custom

erknow

ledge,

shapingof

custom

erperceptio

nsof

theorganizatio

n’sproducts

andim

age,build

ingcustom

errelatio

nships

−Customer

managem

ent

competences

Survey

Subjectiv

efinancial

performance

−

Song

etal.(2005)

Theyfollo

wDay’s(1994)

concept

−Technology/m

arketin

gcompetences

Survey

Subjectiv

efinancial

performance

−

Song

etal.(2007)

“Knowledgeof

thecompetitionandof

custom

ers,as

wellasskill

insegm

entin

gandtargetingmarkets,inadvertisingandpricing,

andin

integratingmarketin

gactiv

ity”(p.21)

−Technology/IT/m

arketlinking/

marketin

gcompetences

Survey

Earningsbefore

taxes/revenues

−

Song

etal.(2008)

Theyfollo

wSongetal.’s

(2007)

concept

−Technology/IT/m

arketlinking/

marketin

gSu

rvey

−−

Vorhies

and

Morgan(2005)

“Transform

resourcesinto

valuableoutputsbasedon

theclassic

marketin

gmix”(p.82)

−Eight

marketin

gcompetences

Survey

Subjectiv

efinancial

performance

−

Thisstudy

“Process

tocombine

marketin

gresourcesby

developing

and

leveraging

relatio

nala

ndintellectuala

ssetsto

satisfy

custom

ersandattain

brandequity”

Advertising,

prom

otion,

andmarketin

gintangibles

Salesas

interm

ediary

outputand

brandequityandcustom

ersatisfactionas

final

outputs

NetworkDEA

Returnon

assets

Tobin’sqand

analysts’

recommendatio

ns

ITinform

ationtechnology,EBIT

earnings

before

interestsandtaxes,CR

investmenton

custom

errelatio

nship,

CRM

custom

errelatio

nshipmanagem

ent,SR

Oshow

room

occupancy,

ME

marketin

gexpenditu

res,DEAdataenvelopm

entanalysis,andSF

Estochasticfrontierestim

ation

J. of the Acad. Mark. Sci.

Page 4

and marketing objectives are grouped, but without an explicitreference to the process by which activities transform intoobjectives. Therefore, marketing capabilities are not approachedas an integrated process of resources and end results, missing thealignment with organizational capability theory.

Insofar as resource-based and organizational capabilitytheories conceive marketing capabilities as the process ofconverting marketing resources into marketing-specific endresults (Dutta et al. 1999; Narasimhan et al. 2006), we com-bine these studies and develop our research to consider skillsand processes in the generation of marketing capabilities(Dutta et al. 1999; Vorhies and Morgan 2005). The questionof what the end results of marketing resources should be is nottrivial. Extant literature argues that marketing resourcesshould maximize sales and, thus, that capabilities should beoriented to sales generation (Akdeniz et al. 2010; Dutta et al.1999; Narasimhan et al. 2006; Nath et al. 2010). However,these studies do not explore whether current sales levels aredue to past customer satisfaction or past brand equity levels.At the same time, these studies do not indicate whether saleslevels affect customer satisfaction or brand equity levels. Ourstudy explicitly posits that marketing resources should opti-mize sales for maximizing customer satisfaction and brandequity. Our research also disentangles the potential relation-ships among sales, customer satisfaction, and brand equity.The basic assumption here is that marketing actions areintended first to spark customer response to a product andthen, through customer response, to create customer satisfac-tion and brand equity (Keller and Lehmann 2003; Rust et al.2004). This study also assumes that firms have a strategic needto develop a close relationship with customers (Day 2006;Gulati 2009).

Studies have used different types of data to operationalizemarketing capabilities. Some studies operationalize marketingcapabilities using primary data collected through surveys(e.g., Morgan et al. 2009a, b), whereas other studiesoperationalize marketing capabilities using secondary dataand estimate them with frontier estimations, such as stochasticfrontier estimation (SFE) or DEA (Akdeniz et al. 2010; Duttaet al. 1999). Our article builds on these latter studies; however,studies that employs DEA use basic models to estimate staticmarketing capabilities (e.g., Akdeniz et al. 2010; Nath et al.2010). In contrast, we seize advances in operations researchby employing a novel, refined method to measure COMC,based on network DEA estimations (Färe and Grosskopf2000; Lewis and Sexton 2004; Sexton and Lewis 2003; Prietoand Zofío 2007; Tone and Tsutsui 2009), and thus providedynamic marketing capabilities estimations.

Regarding the influence of marketing capabilities on per-formance, Table 1 shows that the literature relates marketingcapabilities to subjective and objective measures of financialperformance. We construct our article on the basis of thesestudies, in particular on articles that stress objective financial

performance. However, these studies are largely restricted tolinking marketing capabilities to short-term performance. In-stead, our article takes into consideration research on valuerelevance (e.g., Kothari 2001) and examines the impact ofmarketing capabilities on long-term performance. In the fol-lowing sections, we attempt to answer whether and howCOMC influences financial performance.

Conceptual framework: impact of COMC on financialperformance

We summarize our conceptual framework and hypotheses inFig. 1. We examine the relationship between COMC andfinancial performance. In the development of the framework,we proceed as follows: we first define two key constructs ofour framework—that is, COMC and forward-looking perfor-mance—and then explain the logic of each hypothesizedarrow.

Defining COMC

As mentioned previously, our approach to measuring market-ing capability intends to integrate sales and customer-orientedapproaches. We also intend to disentangle the black box ofmarketing capability. When defining COMC, we employ bothorganizational capability theory and resource-based theoryand view marketing as a lead function in managing customerrelationships (Srivastava et al. 1998). In general, we defineCOMC as the process of combining marketing resources byleveraging relational and intellectual assets to satisfy customersand attain brand equity (Day 1994; Keller and Lehmann 2003;Rust et al. 2004; Srivastava et al. 2001; Winter 2000). In thatsense, the three components of COMC, according to organiza-tional capability theory, are marketing resources and the twomarketing-specific end results of customer satisfaction andbrand equity; the process involves developing and leveragingrelational and intellectual assets to connect marketing resources

Forward-Looking Performance

Tobin’s qAnalysts’ recommendations

Control VariablesQuarterly stock returns

Return on assetsIndustry concentration

Firm assets

H1, H2COMC

Fig. 1 Conceptual framework and hypotheses of the impact of COMCon forward-looking performance

J. of the Acad. Mark. Sci.

Page 5

with marketing end results. According to resource-based theoryapplied to marketing, a firm must develop and leverage itsrelational and intellectual market-based assets to satisfy cus-tomers and build brand equity (Srivastava et al. 2001). Rela-tional market-based assets are “relationships with and percep-tions held by external stakeholders,” whereas intellectualmarket-based assets include “the types of knowledge a firmpossesses about its environment” (Srivastava et al. 2001, pp.779–81), which refers to firms’ market knowledge of cus-tomers, competitors, or the market environment in general.By leveraging knowledge of competitors and perceptions ofcustomers, firms can develop marketing strategies (e.g., prod-uct or promotion strategies) that in turn lead to tactical market-ing actions, customer response, and the creation of marketingintangibles.

Figure 2 graphs our conception of marketing capability.The starting point of the process is current levels of marketingintangibles (or marketing-specific end results in period t ). Ourassumption here is that the customer knows a firm’s product orhas already consumed (or experienced) it. Under this assump-tion, there is a current level of brand equity or customersatisfaction in period t ; otherwise, it could be very difficultfor a company to achieve brand equity or customer satisfac-tion. In the case of a new product by a new company, levels ofbrand equity or customer satisfaction are assumed to be non-existent. Brand equity and customer satisfaction pertain to thevalue the customer assigns to a market offering in terms ofawareness (recall and recognition), attributes (product andnon-product related), benefits (functional, experiential, sym-bolic), attitudes, attachment (loyalty), and network effects(word of mouth) (Fornell et al. 1996; Keller 1993;Srivastava et al. 2001). In particular, brand equity is theincremental effect of a brand name on the preference for amarket offering, and the antecedents that affect brand equityare brand awareness and brand associations in terms of attri-butes and benefits (Ailawadi et al. 2003; Keller 1993;Srinivasan et al. 2005). Customer satisfaction is the result of

an “overall evaluation of [the] whole purchase and consump-tion experience with a good or service” (Fornell 1992, p. 11)and captures the attitude of the customer toward the marketoffering; the antecedents of customer satisfaction are customerexpectations and product quality (Anderson and Sullivan1993; Bolton and Drew 1991).

Significant levels of brand equity or customer satisfactionin period t may lead customers to increase purchases of theproduct and thus help companies obtain sales in period t+ 1.In addition, companies also expect to generate sales in periodt+ 1 through marketing actions, such as branding initiatives,advertising campaigns, or other initiatives designed to moti-vate customer purchases (Rust et al. 2004). In this research,marketing actions refer to tactical actions that require market-ing expenditures so that firms can deploy, allocate, and com-bine marketing resources (Dutta et al. 2005; Narasimhan et al.2006; Rust et al. 2004). Marketing resources capture a rangeof activities, including new product development, packaging,advertising, and promotion (Keller and Lehmann 2003).

To increase (or decrease) levels of brand equity or customersatisfaction in period t+ 1, customers need to have purchased(or experienced) the product again in period t+ 1. In addition,and in line with the marketing chain approach, a firm’s mar-keting resources in period t+ 1 can build brand equity andcustomer satisfaction in the same period (Keller and Lehmann2003; Rust et al. 2004). Marketing resources affect brandequity antecedents, such as brand awareness of and brandassociations with attributes and benefits (Keller 1993), andcustomer satisfaction antecedents, such as customer expecta-tions and product quality (e.g., Rust et al. 1995). Advertisingand promotion, for example, build brand awareness and brandassociations (Keller 1993) and “have an effect on customerexpectation formation through information and make productquality easy to evaluate” (Anderson and Sullivan 1993, p.132). Moreover, marketing resources may have carryovereffects, such that advertising and promotion in period t mayaffect brand equity, customer satisfaction, or sales in period

Customer Response

1

2

21Marketing ResourcesAdvertisingPromotion

Sales(t+1)

Marketing IntangiblesBrand EquityCustomer Satisfaction

Marketing IntangiblesBrand Equity(t)

Customer Satisfaction(t)

Sales (t+1)

2

Fig. 2 COMC and network DEA

J. of the Acad. Mark. Sci.

Page 6

t+ 1 through levels of brand equity and customer satisfactionin period t . For example, the extent of awareness, liking, andpositive attitude toward the company’s product is built overthe years through its advertising efforts (Dutta et al. 1999).

In summary, marketing capability is related not only to howcompanies manage the use of resources to generate sales butalso to how the experience with the product may improvemarketing intangibles. Some companies may be able to opti-mize sales and improve intangibles. However, creating sales isonly necessary to improve intangibles because generatingintangibles is a complex task that involves a multi-step pro-cess. In the first step, the marketing intangibles coming fromthe previous period and the marketing resources already spentgenerate sales. Sales are important from a short-term perspec-tive and assume the role of intermediate output because, in thesecond step, the maximization of the contemporary marketingintangibles requires, apart from other input consumption, gen-erating the optimal sales level possible. In other words, salesintermediate the impact of the previous marketing intangiblesand current marketing resources on the generation of newmarketing intangibles. In the way the estimation method hasbeen defined, the network DEA model guarantees that thefinal output is achieved if, and only if, the level of the inter-mediate outputs (the sales level) is aligned with this target. Inother words, the results of the isolated optimization of the twodifferent stages can produce a final result significantly differ-ent from the optimal of the network DEA.

Forward-looking financial performance

We assume that financial performance has both a short-termand a long-term dimension (Srivastava et al. 1998). Opera-tional performance refers to the short-term dimension of per-formance and contains information about contemporaneousearnings reflected on the firms’ profit and loss account.Forward-looking performance reflects the long-term dimen-sion of performance and therefore comprises informationabout future earnings, commonly associated with performancein capital markets (Barth et al. 2001; Kothari 2001). In thisresearch, we concentrate on forward-looking performance. Asmentioned previously, managers and academics are interestedoverall in how marketing influences financial performance inthe long run. However, extant literature neglects the influenceof marketing capability on forward-looking performance (seeTable 1).

The influence of COMC on forward-looking performance

We ground the relationship between COMC and forward-looking performance on the value relevance approach. Thisapproach indicates that a firm’s future earnings (e.g., stockreturns) can be reflected in new information contained inaccounting and non-accounting performance measures (e.g.,

Barth et al. 2001; Jacobson and Mizik 2009; Kothari 2001;Mizik and Jacobson 2008). The application of the valuerelevance approach in marketing suggests that non-accountingmeasures, such as marketing intangibles measures,have the ability to affect future earnings (e.g., Anderson et al.2004) and may supplement accounting data to directly affectstock returns (e.g., Jacobson and Mizik 2009; Luo et al. 2010;Srinivasan and Hanssens 2009). Therefore, the literature evi-dences that firms’ intangible assets have the power to predictfuture earnings (Daniel and Titman 2006).

Using the value relevance rationale, we posit that COMCmay influence future earnings because it can provide infor-mation not incorporated in accounting measures. COMC re-flects the ability of a firm to create and take advantage ofmarketing intangibles through the expenditures in marketingactivities and sales generation. Thus, greater COMC maygenerate higher levels of word of mouth and, at the same time,boost customer loyalty, which in turn can boost customeracquisition and customer retention. As a consequence, asVorhies et al. (2011) recognize, the capability of managingthe relationship with customers drives positive returns onmarketing investments, thus helping enhance financialperformance.

Therefore, COMC influences firms’ actual and futuregrowth through (1) the ability to acquire new customers forcurrent offerings (positive word of mouth of existing cus-tomers leads new customers to try existing offerings [Ambleret al. 2002]), (2) the ability to encourage cross-buying fromcurrent customers (brand awareness and positive brand asso-ciations on the one hand and customers’ experience andlengthy relationships with the firm on the other hand influencecustomers’ willingness to engage in cross-buying [Bolton1998; Branson 1998; Verhoef 2001]), and (3) the use ofreduced marketing resources (strong brands may not requireas much continued investment as competitors’ weaker brandsto maintain their success, and current customers in strongrelationships with firms may continue to purchase withoutthe need for further marketing costs [Ambler et al. 2002]).Therefore, by acquiring and retaining customers throughCOMC, the firm ensures actual and future growth and, con-sequently, future expected earnings (Collins and Kothari1989; Rappaport 1998; Srivastava et al. 1998). In essence,COMC affects future earnings positively by providing infor-mation about future growth. Thus, H1 is hypothesized. H1 isnot a new hypothesis but it is necessary for the completenessof our framework. H1 is also necessary because it will allow tocompare our study with the one developed by Dutta et al.(1999). Therefore,

H1: COMC has a positive direct impact on a firm’s value onthe stock market.

In contrast, financial analysts establish their recommenda-tions using all available sources of industry and firm-specific

J. of the Acad. Mark. Sci.

Page 7

information (Howe et al. 2009; Womack 1996). In this sense,analysts’ recommendations often capture intangible aspects ofa firm’s operations that do not appear in financial performanceindicators (Jegadeesh et al. 2004; Womack 1996). Recentresearch shows, for example, that financial analysts incorpo-rate customer satisfaction information into share prices (Luoet al. 2010; Ngobo et al. 2012). Therefore, because COMCcomprises firm-specific information, it may be a potentialsource of analysts’ recommendations, and thus we assumethat financial analysts also incorporate COMC into futureearnings.

The main objective of financial analysts is to make stockrecommendations (Womack 1996). Analysts first collect andprocess a variety of information about different stocks, formtheir beliefs, and predict stock values relative to their currentmarket prices and then rate the investment potential of eachstock in terms of recommendations to buy, hold, or sell(Jegadeesh et al. 2004; Womack 1996). Therefore, analysts’recommendations should capture future earnings. Literaturehas proved that changes in recommendations predict futureearnings because such changes may be capturing qualitativeaspects of a firm’s operations, such as managerial abilities,intangible assets, or other growth opportunities, that do notappear in accountingmeasures (Jegadeesh et al. 2004). Insofaras COMC provides non-accounting information or qualitativeaspects of a firm’s operations, we argue that COMC influencesanalysts’ stock recommendations.

COMC may improve analysts’ recommendations becauseit provides intangible information in terms of customer satis-faction and brand equity, which are significant sources offuture earnings (Anderson et al. 2004; Barth et al. 1998;Jacobson and Mizik 2009; Madden et al. 2006). In addition,marketing intangible assets (i.e., customer satisfaction andbrand equity) “represent a reservoir of cash flow that hasaccumulated from marketing activities but has not yet trans-lated into revenue” because at any point in time firms’ “mar-keting actions will have made changes in customers’ mentalstates, but they may not yet have influenced the firms’ profitand loss account” (Rust et al. 2004, p. 78). Thus, we expect thefollowing:

H2: COMC has a positive impact on analysts’ stockrecommendations.

In summary, we can indicate that overall, these two hypoth-eses are oriented toward the assessment of the value added ofthe intangibles generated by the marketing function. Whenvalue is created, the final intangible assets in time period t +1exceed largely those presented in the previous year. However,when value is destroyed (or not generated as efficiently as itcould be), the final output is far from its optimal level. Onecondition to achieve this desirable impact on the intangibles isto have aligned the intermediate output (the level of sales) toguarantee the achievement of the final goal.

The role of control variables

Future earnings can be influenced by several variables. Stan-dard variables used as control variables in marketing andfinance research are quarterly stock returns, return on assets(ROA), industry concentration, and firm assets (Andersonet al. 2004; Jegadeesh et al. 2004; Luo and Donthu 2006;Luo et al. 2010; Mizik and Jacobson 2008).

Quarterly stock returns We use quarterly stock returns toexplain future earnings following the logic of finance re-search. In particular, Jegadeesh et al. (2004) use past stockreturns as independent variables, arguing that firms withhigher past stock returns earn higher returns over the next12 months. Thus, we expect a positive effect of past quarterlystock returns on future earnings.

ROA Research in marketing and value relevance argues thatwhen examining the influence of a marketing variable onfuture earnings, an accounting variable should be includedas an additional explanatory variable. Specifically, ROA isused as an accounting variable because it captures the short-term or operational performance of firms. Thus, in line withextant research, we expect a positive effect of ROA on futureearnings.

Industry concentration To control for possible industry effects,we include industry concentration in the framework. Accordingto industrial organization theory, industry concentration refersto the number of incumbents in a given industry (Scherer1980). Industries with fewer competitors, and thus lower rivalryamong them, have low levels of concentration, whereas theopposite is true for industries with high levels of concentration.The finance literature suggests that for low-concentrated indus-tries, firms tend to earn higher stock returns because theyengage in more and riskier innovations and suffer more distressrisk (Hou and Robinson 2006). Therefore, we expect a negativeeffect of industry concentration on future earnings.

Firm assets The role of firm assets on future earnings is notclear. Some studies find a positive effect of assets on futureearnings, whereas others find no effect (e.g., Jegadeesh et al.2004; Luo et al. 2010).

Methodology

Modeling the effect of COMC on forward-lookingperformance

Following mainstream financial literature (Damodaran 2002),we employ Tobin’s q and analysts’ stock recommendations as

J. of the Acad. Mark. Sci.

Page 8

measures of forward-looking performance. Our intention is toobserve consistency of the COMC effects on differentforward-looking performance measures.

Effect of COMC on Tobin’s q Tobin’s q is a forward-lookingmeasure of performance, comparable across firms, that capturesinformation about firms’ future potential earnings (e.g.,Anderson et al. 2004; Luo and Donthu 2006; Rao et al. 2004).Tobin’s q value is the ratio of the firm’s market value to thecurrent replacement costs of its assets. A firm that createsmarketvalue that is greater than the replacement cost of its assets alsocreates more firm value (Chung and Pruitt 1994). The differencebetween a firm’s Tobin’s q and unity indicates the degree of afirm’s anticipated future abnormal returns and its level of intan-gibles (Rao et al. 2004; Simon and Sullivan 1993). ThereforeTobin’s q is a surrogate of firm’s value on the stock market.

Because Tobin’s q may suffer from substantial autocorre-lation (Jacobson andMizik 2009), we use its annual growth asthe dependent variable. To test H1, that COMC directly influ-ences Tobin’s q, we specify the effect of COMC. To avoidproblems of spurious correlation, however, we specify theeffect of COMC in variations or changes (ΔCOMC). Workingwith variations also enables us to control for firm-specificinformation that is not modeled. As mentioned previously, weinclude quarterly stock returns, ROA, industry concentration,and firm assets in our model (Anderson et al. 2004; Luo andDonthu 2006; Luo et al. 2010; Mizik and Jacobson 2008). Inaddition, to control for potential bias due to omitted time andrandom effects (Boulding 1990; Jacobson 1990), we specify atime effect λt that is common to all firms and prevents cross-individual correlation, along with the random-error term νit. Inshort, we elaborate the following equation:

ΔTobin’s qit ¼ β0 þ β1ΔCOMCit þ β2SRETiq1 tð Þ þβ3SRETiq2 tð Þ þ β4SRETiq3 tð Þ þ β5SRETiq4 tð Þþβ6ΔROAit þ β7ΔHHIijt þ β8ΔSizeit þ λt þ υit:

ð1Þ

where

ΔTobin’s qit the annual change of Tobin’s qΔCOMCit the annual changes of COMCSRETiq1-4(t) the quarterly stock returns during fiscal year tΔROAit the annual change of ROAΔHHIijt the annual change of industry concentrationΔAssetsit the annual change of firm assets, andλt annual time effects.

Effect of COMCon analysts’stock recommendation Analysts’stock recommendations reflect a message directed to investorsabout whether to sell, hold, or buy stocks. To estimate theeffect of COMC on analysts’ recommendations and, thus, to

test H2, we followHowe et al. (2009), Jegadeesh et al. (2004),and Womack (1996), who suggest including lagged stockreturns in the specification. Jegadeesh et al. (2004) advisethe use of changes in analysts’ recommendations instead ofthe level of analysts’ recommendations as the dependentvariable. We also include the effects of ΔCOMC and controlvariables as follows:

ΔARiq1 tþ1ð Þ ¼ β1ΔCOMCit þ β2SRETiq1 tð Þ þ β3SRETiq2 tð Þ þβ4SRETiq3 tð Þ þ β5SRETiq4 tð Þ þ β6ΔROAitþβ7ΔHHIijt þ β8ΔSizeit þ λt þ υit:

ð2Þwhere ΔARiq1(t + 1) represents the annual change of analysts’recommendations at the end of the first quarter following thefirm’s fiscal year-end t. We use a quarter-ahead measure[q1(t+1)] to ensure that capital market participants have in-corporated new information in their expectations. The rest ofthe variables are defined as previously.

Modeling COMC

Existing literature employs survey-based, stochastic frontiers,or activity analysis to model and estimate marketing capabil-ity. As noted previously, we are investigating a capability thatemphasizes the process of converting resources (inputs) intoend results (outputs). In this case, the DEA method fits withthis approach and therefore serves to model and estimateCOMC. In essence, DEA captures how efficiently firms de-ploy inputs to achieve desired outputs, and therefore moreefficient firms will have greater capability (Dutta et al. 2005).As a mathematical method to compare firms’ productivityusing multiple inputs and multiple outputs, DEA is preferableto simple ratios (e.g., brand equity to advertising). Further-more, DEA builds an efficient frontier that consists of allefficient units, enabling a comparison with best performers;in contrast, regression analysis and canonical regression relyon a comparison with the mean (Donthu et al. 2005; Luo andDonthu 2006). The ability to benchmark a firm’s capabilityrelative to the best performers reflects comparative advantagetheories, which indicate that the most heterogeneous firms (orbest performers) obtain superior competitive advantages(Hunt and Morgan 1995). Compared with SFE, DEA hasgreater flexibility because it does not require an explicit func-tional form imposed on the data (e.g., Coelli et al. 2005). Thus,DEA has gained widespread acceptance and is usually used toexamine the efficiency of price, advertising, service quality,and customer satisfaction, among other topics (e.g., Kamakuraet al. 1988; Luo and Donthu 2005; Mittal et al. 2005;Mukherjee et al. 2003; Pergelova et al. 2010).

Following Luo and Donthu (2005) and Mittal et al. (2005),we employ an output-oriented DEA model with variable

J. of the Acad. Mark. Sci.

Page 9

returns to scale (Banker et al. 1984) to control for possibleeconomies of scale. According to Piercy (1987), this output-oriented model is appropriate because firms usually approvemarketing budgets in advance, with the aim to maximizeoutput (e.g., customer satisfaction) with the available funds(marketing resources).

Because capabilities develop over time, we include a timevariable in the DEA estimation (Tulkens and Vanden Eeckaut1995) by using an intertemporal rather than a contemporane-ous estimation model. The technological assumption of thelatter is a yearly, cross-sectional, technical change, whereasthe intertemporal frontier assumes stability and comparabilityduring a period of analysis. That is, intertemporal DEA as-sumes that the dominant best practices in marketing remainrelatively stable for a six- to seven-year window (Mittal et al.2005). We construct a single intertemporal frontier from theobservations throughout the observation period.

One of the requirements of DEA is that firms included inthe analysis should be comparable. In the assessment ofindustrial productivity, this requirement is met when we com-pare manufacturing firms that form part of the same economicsector. In this article, we are not comparing manufacturing butrather marketing practices, which implies that the requirementfor the economic sector is not applicable. As a result, werelaxed this requirement because, among different industries,marketing practices can be perfectly comparable.

As Fig. 2 illustrates, our proposal for capturing marketingcapability is different from standard DEA efficiency modelsbecause the generation of marketing intangibles is the result oftwo consecutive and interdependent stages. Standard DEAmethods treat the organization under assessment as a black-box , meaning just taking into account the inputs consumptionand the output created. In some cases, and depending on theresearch question raised, this approach can be acceptable. In ourcase, as a more complex problem is defined, the standard DEAappears to be insufficient. Indeed, as a relationship between theintermediate (short-term) output (the level of sales) and the final(long-term) output (the generation of marketing intangibles)has to be established, network DEA is necessary.

Our approach of DEA models is based on network DEA.During stage 1, the existent marketing intangibles (comingfrom period t ), in addition to the marketing resources(expended in t +1), are oriented toward the fulfillment of salesmaximization (in t +1). In stage 2, the optimal level of sales isadded to the previous resources (marketing resources,expended in t +1, and marketing intangibles, existent in t ) tooptimize the level of marketing intangibles generation (int +1). A firm oriented toward the maximization of intangiblesgeneration should manage efficiently both steps in order toguarantee the achievement of the final goal. Any problem inone of the steps will make it impossible to achieve the maximallevel of intangibles generation. Thus far, we have providedan intuition of our network DEA proposal. The Appendix

provides the mathematical development to estimate the effi-ciency coefficients.

Data and operationalization of variables

Sample

The unit of analysis is the firm. The marketing, financial, andcontrol data variables cover seven consecutive years, from 2000to 2006. After merging data from various sources, we gather264 complete observations in levels during the period of anal-ysis. However, after computing network DEA to estimate mar-keting capability, we have 212 observations in levels to be usedin hypotheses testing. We use 160 observations in changes (Δ)to test H1 and H2. This reduction in the number of observationsis common when several secondary sources are merged. It isimportant to mention, though, that the main bottleneck of datacollection resides in the marketing data.We used twomarketingdata sources: Advertising Age and Crain Communications, andthe ACSI. In particular, the Advertising Age and Crain Com-munications marketing source created a drastic reduction in thenumber of companies under analysis. For collecting financialdata we use COMPUSTAT and the Institutional Brokers Esti-mate System for analysts’ recommendations information; thisalso created a reduction in the number of companies underanalysis, since not all companies included either in ACSI orAdvertising Age are followed by stock analysts.

Our sample consists of firms from the following industries:limited service restaurants, automobiles and light vehicles,fixed-line and wireless telephone service, personal care andcleaning products, specialty retail stores, food manufacturing,breweries, department and discount stores, personal com-puters, supermarkets, and soft drinks, among others. Theindustry composition matches the type of industries used inextant literature (e.g., Anderson et al. 2004). Table 2 shows thesample distribution by industry.

Table 3 provides the variables, their operationalization, andthe sources of data. Financial and control data mainly camefrom COMPUSTAT and the Institutional Brokers EstimatesSystem (I/B/E/S). Marketing data came from different sourcesas specified subsequently.

Operationalization of the dependent variables

Tobin’s q We measure an approximate Tobin’s q using Chungand Pruitt’s (1994) approach, which is standard in marketingresearch. We use the following formula:

Approximate Tobin’s q ¼ MVEþ LPSþ DEBTð Þ.BTA; ð3Þ

where MVE is the annual close stock price multiplied by theoutstanding annual common shares, LPS is the liquidating

J. of the Acad. Mark. Sci.

Page 10

value of the firm’s preferred stock, DEBT is (short-termliabilities—short-term assets)+(long-term debt), and BTAstands for the book value of total assets.

AR We use more than 10,000 I/B/E/S analysts’ recommenda-tions issued for the firms under analysis during our study andreverse them to indicate 1 (“strong sell”) to 5 (“strong buy”),following the financial literature (e.g., Howe et al. 2009;Jegadeesh et al. 2004). The AR is computed for each firm-year and is the average of the analysts’ recommendationsissued during the first quarter following the firms’ fiscalyear-end. For example, for firms whose fiscal year ends inDecember, we estimate the AR using recommendations issuedbetween January andMarch of the following year.We treat theAR variable as quantitative; prior research has shown thattreating AR as quantitative or ordinal produces the sameresults (Jegadeesh et al. 2004). After computing the averageanalyst recommendation for each firm, we calculate the stan-dard deviation of recommendations. In the end, we integrateboth the average recommendation and its standard deviationin one measure. We follow the ratio according to Sharpe’s(1966) rationale, or the so-called ratio of return to volatility.Sharpe suggests that the average profitability of a mutual fundshould be adjusted to its risk (the standard deviation). Apply-ing Sharpe’s idea, we calculate the ratio of average recom-mendations divided by its standard deviation. This ratio indi-cates analysts’ recommendations adjusted by dispersion,where a higher ratio represents better recommendations witha high degree of consensus (less dispersion). We apply

logarithm to this ratio to obtain the actual measure of AR usedas the dependent variable in Eq. 2.

Operationalization of COMC

We employ a network DEA model to estimate COMC. Tooperationalize it, we use advertising and promotion expendi-tures as marketing resources, sales as intermediary output, andbrand equity and customer satisfaction as marketing-specificend results (see Fig. 2). We focus on advertising and promo-tion resources because they demand the largest share of mar-keting expenditures (Ambler 2000) and are relevant inputflows of the marketing process (Keller and Lehmann2003; Rust et al. 2004). We focus on sales because theyrepresent customer response (Ailawadi et al. 2003). Weconcentrate on brand equity and customer satisfaction inparticular, because they have roots in market orientation,place the customer at the center of the firm, and arewidely examined in marketing literature (Day 1994; Kohliand Jaworski 1990; Rust et al. 2004).

Advertising and promotion We employ total advertising ex-penditures and total promotion expenditures. We obtain ad-vertising and promotion data from the “100 Leading NationalAdvertisers” 2001–2007 reports of Advertising Age and CrainCommunications.

Sales We use firms’ annual sales from COMPUSTAT.

Table 2 Sample distribution byindustry 2-digit

SICSIC major groups (2-digit) Examples of sampled companies Number of

observations

20 Food and Kindred Products Anheuser Busch Companies,Inc./Coca-Cola Company/KelloggCompany

62

21 Tobacco Products Altria Group Inc. 6

28 Chemicals and Allied Products Colgate-Palmolive Company 15

30 Rubber and Miscellaneous Plastic Products Nike 6

35 Industrial Machinery and Equipment Dell Corporation 12

36 Electrical, Other Electrical Equipment,Excluding Computers

General Electric Co. 6

37 Transportation Equipment Toyota Motor Corporation 45

42 Motor Freight Transportation, Warehouse United Parcel Service Inc. 2

48 Communications AT&T Inc. 13

52 Building Materials Lowe’s Companies, Inc. 10

53 General Merchandise Stores Kohl’s Corporation 34

54 Food Stores Safeway Inc. 15

57 Furniture and Home Furnishings Stores Best Buy Co., Inc. 6

58 Eating and Drinking Places Burger King Holdings, Inc. 20

73 Business Services Microsoft Corp. 6

78 Motion Pictures Times Warner Inc. 6

- Total − 264

J. of the Acad. Mark. Sci.

Page 11

Customer satisfaction We use the ACSI, in which customersatisfaction is a latent variable that results from perceivedquality, perceived value, and customer expectations (for de-tails, see Fornell et al. 1996).

Brand equity Brand equity refers to the incremental effect of abrand name on the preference for a market offering (Ailawadiet al. 2003; Keller 1993). We follow the revenue premium ratio-nale proposed by Ailawadi et al. (2003), which suggests thatthe difference between a firm’s revenues and its competitor’sunbranded revenues (i.e., revenue premium) represents brandequity. Following Luo and Donthu (2006) and Slotegraaf andPauwels (2008), we elaborate Eq. 4 and use its residual as anapproximate measure of brand equity. In essence, this equa-tion suggests that after accounting for past sales level, delayedmarketing effects, firm effects (size and firm dummies), in-dustry concentration effects (HHI), and economy-wide factors(year dummies), we can obtain the residual of sales, whichcan be an approximate measure of brand equity (BEit). Thismeasure captures un-modeled brand effects.

BEit ¼ Salesit−Predicted Salesit;where Predicted Salesit¼ β0 þ β1Salesi t−1ð Þ þ β2ADi t−1ð Þ þ β3PROMOi t−1ð Þþ

β4ΔAssetsit þ β5HHIij þ ηi þ λt þ υi t;

ð4Þ

where

BEit the approximate annual brand equity offirm i at the end of year t.

Salesit the log of the annual sales level of firm iat the end of year t.

ADi(t – 1) andPROMOi(t – 1)

the one-year lags of the annual advertis-ing and promotion expenditures of firm i,respectively. We used lag variables tocontrol for the carryover effects of mar-keting resources.

ΔAssetsit the annual growth of assets of firm i atthe end of year t.

HHIij the effect of industry concentration onfirm i in industry j, which is timeinvariant in the period under analysis.

λt the time effect common to all firmsand prevents cross-individual corre-lation (for which we include year-dummy effects).

ηi a permanent but unobservable firm-specific effect to control for firm-specificinformation that we do not model.

υit. the random-error term.

Operationalization of control variables

SRET The stock return is the log of the quarterly growth of themarket value of equity (MVE), or log(1+ΔMVEiqt), whereΔMVEiqt is [(MVEiqt – MVEiq – 1(t))/MVEiq – 1(t)].

Table 3 Variables and sources of data

Variable Operationalization Sources of information

Tobin’s q Approximate Tobin’s q=(MVE+LPS+DEBT)/BTA (see equation 6) COMPUSTAT

AR The average of analysts’ recommendations during the first quarter after the firms’ fiscalyear-end adjusted by its standard deviation. We apply log.

I/B/E/S

COMC The result of the output-oriented network DEA based on the ratio of multiple outputs(customer satisfaction and brand equity) and intermediary outputs (sales) to multipleinputs (advertising and promotion). We apply log.

Self-estimated

Advertising Total sum of advertising expenditures in television, radio, print, outdoor, and Internet “100 Leading National Advertisers”by Advertising Age and CrainCommunications

Promotion Total sum of expenditures in direct marketing, sales promotion, co-op spending,coupons, catalogs, product placement, and special events

“100 Leading National Advertisers”by Advertising Age and CrainCommunications

Sales A firm’s annual sales level COMPUSTAT

Customersatisfaction

A firm’s American Customer Satisfaction Index National Quality Research Centerat the University of Michigan

Brand equity The residual of sales (see Eq. 4) COMPUSTAT

SRET Stock return is the log of the quarterly growth of market value of equity (MVE):log (1+ ΔMVEiqt), where ΔMVEiqt: [(MVEiqt−MVEiq−1(t))/MVEiq−1(t)].

COMPUSTAT

ROA The unanticipated ROA is the residual of the following equation:(ROAit−ROAi(t−1))=β0+β1(ROAi(t−1)−ROAi(t−2)). ROA is the ratio of earningsbefore taxes and extraordinary items to total assets. ROA has been mean centered.

COMPUSTAT

HHI Hirschman–Herfindahl index based on four-digit Standard Industrial Classification COMPUSTAT

Assets We measure firm assets as the natural log of total assets. COMPUSTAT

J. of the Acad. Mark. Sci.

Page 12

ROA The ROA is the ratio of earnings before taxes andextraordinary items (EBT) to total assets. To isolate fixedeffects, we mean-center the ROA and obtain a measure ofunanticipated ROA (UΔROA) on the basis of the followingequation: (ROAit−ROAi(t−1))=β0+β1(ROAi(t−1)−ROAi(t−2)).We estimate UΔROA in the spirit of Mizik and Jacobson(2008) and Kormendi and Lipe (1987).

HHI We calculate the Hirschman–Herfindahl index of theindustry in which the firm operates on the basis of itsfour-digit Standard Industrial Classification (Schmalensee1977).

Assets We use the natural log of a firm’s total assets.

Time-specific variable We use dummy variables for each yearunder analysis.

Empirical findings

Descriptive statistics

After running the network DEA model, we obtain three effi-ciency scores. Stage 1 efficiency score represents the market-ing capability of using advertising, promotion, and past mar-keting intangibles to maximize sales. Stage 2 efficiency scorerefers to the marketing capability of employing sales, adver-tising, promotion, and past marketing intangibles to maximizemarketing intangibles. Stage 3 efficiency score represents theoverall marketing capability, or what we conceptualize andmodel as COMC (in technical terms, the optimization ofstages 1 and 2 is produced simultaneously ). We obtainoutput-oriented efficiency scores in terms of Shepard’s dis-tances (Wilson 2008), for which scores are less than or equalto 1. Intuitively, a score equal to 1 means that the firm is on thefrontier, is efficient, and forms part of a sub-sample of firmsthat optimize the transformation of inputs into outputs. Ascore of less than 1 indicates that the firm is not on the frontierand is less capable than efficient firms of optimallytransforming inputs into outputs. Descriptive statistics showthat the mean of stage 1 capability is 0.63, the mean of stage 2capability is 0.97, and the mean of COMC is 0.79. Figure 3shows the histogram of COMC along with the stage 1 and 2capabilities from the network DEA.

We present the descriptive statistics and the correlationmatrix of the main and control variables (in variations) inTables 4 and 5, respectively. Table 5 shows that ΔCOMC ispositively correlated with ΔAR (r=.20) and with ΔTobin’s q (r=.03). These bivariate estimations offer the first indication of theconnection between COMC and long-term financialperformance.

The effect of COMC on AR and Tobin’s q: Testing H1 and H2

Estimation method for Eqs. 1 and 2 The nature of Eqs. 1 and 2demands that we control for dynamic serial correlation(Arellano 2003; Roodman 2006). The Arellano–Bond test sug-gests that our estimations do not reveal first-order serial corre-lation in differences, which means that they do not suffer fromdynamic panel bias (see Roodman 2006). However, they maysuffer from potential problems of contemporaneous autocorre-lation, and White’s special test recommends correcting forheteroskedasticity in the estimations (Wooldridge 2000). There-fore, for Eqs. 1 and 2, we use panel-corrected standard errors(PCSE) and correct for both contemporaneous autocorrelationand heteroskedasticity (PCSE [het, ar]).3

Effect of COMC on Tobin’s q The estimations of Eq. 1 inTable 6 reveal that ΔCOMC has a significant and positiveeffect on ΔTobin’s q (.22, p <.05). We correct this estima-tion for heteroskedasticity and autocorrelation and find noevidence of multicollinearity. The fit indicates a moderateand significant R-square (.41, p <.01). Control variablesalso affect Tobin’s q. For example, quarterly stock returnsshow a positive and significant effect on Tobin’s q. Firmassets show a negative and significant effect on Tobin’s q,in accordance with extant empirical research. In essence,COMC has a significant effect on Tobin’s q after wecontrol for the effect of additional independent variables,in support of H1.

Effect of COMC on AR As we detail in Table 6, ΔCOMChas a significant and positive effect onΔAR (.81, p <.05). Thiseffect occurs after we account for the effect of past quarterlystock returns, ROA, HHI, and assets. The R-square of thismodel is significant (.27, p <.01). As we expected, the growthof COMC improves analysts’ recommendations, in support ofH2.4

Disentangling the effect of stage 1 and stage 2 marketingcapabilities We decompose the effect of COMC into twosub-effects, taking into consideration efficiencies from stage1 and stage 2 of the network DEA.5 Table 6 reveals that stage

3 Recent marketing research has employed a similar methodology toobtain unbiased estimates (Mizik and Jacobson 2008).4 We further analyze the relationship between COMC and analysts’ stockrecommendations, taking into consideration the moderator effect of ROA(measured as UΔROA), industry concentration (HHI), and total assets(Assets). The results reveal that the effect of COMC on analysts’ stockrecommendations is independent of industry concentration and firmassets. However, ROA moderates the impact of COMC on analysts’recommendations, such that COMC has a larger impact on analysts’recommendations when firms’ ROAs are lower than the median. Thisimplies that financial analysts take COMC into consideration more whenROA is lower. We thank an anonymous reviewer for suggesting thisanalysis. Detailed results are available on request5 We thank an anonymous reviewer for suggesting this additional analysis

J. of the Acad. Mark. Sci.

Page 13

1 capability has a significant and positive impact only onΔTobin’s q (.11, p <.01). However, stage 2 capability showsa positive and significant effect on ΔTobin’s q (.88, p <.10)and ΔAR (4.64, p <.01). In testing the relative strength ofthese two capabilities, the beta coefficient of the effect of stage1 capability on ΔTobin’s q (.13) is higher than that of stage 2capability (.06). In contrast, the beta coefficient of stage 2

capability on ΔAR is higher (.34) than that of stage 1 capabil-ity (−.17). Stage 2 marketing capability seems to be consistentin its effect on both Tobin’s q and analysts’ recommendations.From these results, a marketing capability that includes intan-gibles seems to provide companies good prospects in the longrun.

Robustness check

We perform several additional analyses to provide validity androbustness of our main results.6 In particular, we test constructvalidity of COMC, relax the assumption of time stability ofthe period of analysis using contemporaneous network DEA,check the potential endogeneity of COMC, verify potentialidentification issues of the main estimations, evaluate thevariation of our results controlling for 2-digit SIC effects,verify the mediator role of analysts’ stock recommendationsin the impact of COMC on Tobin’s q, assess the effect ofCOMC on a different dependent variable, and test whetherCOMC is superior to the conceptualization of marketing

6 We thank all five anonymous reviewers for raising important concernsand suggesting relevant additional analysis

Fig. 3 Histogram of COMC and sub-efficiencies obtained from the network DEA model

Table 4 Descriptive statistics

Variables Mean SD Min Q1 Median Q3 Max

1. ΔARiq1(t+1) −.06 .54 −1.44 −.47 −.06 .32 1.89

2. ΔTobin’s qit −.09 .57 −6.19 −.19 −.02 .14 1.46

3. ΔCOMC it −.00 .22 −.79 −.12 .00 .06 .96

4. SRETiq1(t) .01 .16 −.44 −.07 .00 .08 .92

5. SRETiq2(t) .01 .23 −2.78 −.06 .02 .10 .58

6. SRETiq3(t) −.03 .18 −.67 −.12 −.00 .06 .84

7. SRETiq4(t) .04 .19 −.91 −.05 .03 .12 .96

8. UΔROAit .00 .05 −.36 −.01 .00 .01 .36

9. ΔHHIit .00 .02 −.11 −.00 −.00 .01 .08

10. ΔAssetsit .09 .26 −.93 .00 .06 .13 2.96

160 observations in variations (Δ) were used

J. of the Acad. Mark. Sci.

Page 14

capability developed in extant literature. (Detailed results areavailable on request.)

To validate the construction of COMC, we use SFE. Inparticular, to capture stage 1 of the network DEA throughSFE, we use the natural logarithm of sales (LnSales) as theoutput variable and the natural logarithm of advertising(LnAdvertising) and promotional spending (LnPromotion)from period (t ) and customer satisfaction (LnCS) and brandequity (LnBE) from period (t–1) as the input variables. Weemploy panel data stochastic frontier models followingBattese and Coelli (1988). We obtain expected positive signsfor all right-hand side variables included in the SFE. Theseresults provide robustness to the input and output variablesused in the network DEA model.

As indicated in the “Methodology” section, we estimate anintertemporal output-oriented network DEA. We decided todevelop an intertemporal estimation under the assumption ofstability and comparability during the period of analysis(2000–2006). We relaxed this assumption because, in thatperiod of analysis, several firms changed how they weremarketing their products and incorporated the Internet intotheir marketing program, and many firms began to employmobile marketing to gain customers toward the end of thatperiod. We ran a contemporaneous output-oriented networkDEA as a way to check the robustness of the DEA analysisincluded in this research. We followed the illustration provid-ed in Fig. 2. The results indicate that the impact of COMC onTobin’s q and analysts’ recommendations remains significantand positive (p <.05).

To capture potential endogeneity of COMC not modeled inEqs. 1 and 2, we use Arellano–Bond general methods ofmoments (GMM). Several unobserved factors (e.g., man-agers’ expectations of change in financial performance) mayhave a correlation with both COMC and performance. Weperform dynamic panel data estimations using robust differ-ence GMM with a collapsed number of instruments. Wemodel COMC as an endogenous variable following the stan-dard approach of lags 2, as Roodman (2006) suggests. The

coefficients of the impact of COMC on Tobin’s q and ana-lysts’ recommendations remain significant (p <.01).

To test for potential identification issues due to having thedependent variable of Eqs. 1 and 2 aggregated annually andsome of the independent variables expressed in quarterlyterms, we remove year fixed effects from both equations.The results suggest that the impact of COMC on Tobin’s qand analysts’ recommendations remains significant.

We re-run Eqs. 1 and 2 adding an additional control—theeffect of 2-digit SIC. Results reveal that effects of COMC onTobin’s q and on AR remain significant and positive. Also, theeffects of stage 1 and stage 2 capabilities remain significant.

Previous research has studied the mediating effect of ana-lysts’ recommendations in the impact of a marketing constructon firm value (e.g. Luo et al. 2010). We verify the mediatingrole of AR , following Luo et al. (2010). In fact, our resultssuggest that AR is a full mediator because when introducinganalysts’ recommendations in Eq. 1, the impact of COMC onTobin’s q is no longer significant (from p <.05 to p >.10).Also, entering AR leaves the effect of stage 2 capability onTobin’s q no longer significant.

We also assess the robustness of our findings using anadditional dependent variable. We use firms’ raw stock returnsas an additional measure of forward-looking performance.Stock return is the log of the annual growth of the marketvalue of equity {logΔMVEit=log[(MVEit – MVEi(t−1))/MVEi(t−1)]}, and we use it to recalculate Eq. 1, followingMizik and Jacobson’s (2008) method. We obtain similar re-sults as those in Table 6. The annual growth of COMC revealsa significant effect on stock returns (.08, p <.01), which indi-cates that financial markets consider information contained inΔCOMC an effective signal of future potential earnings. Theeffect of control variables is similar to that on Tobin’s q.

To provide empirical evidence of the superior contributionof COMCon performance, we perform additional tests, takinginto consideration the role of extant capability models in theliterature. In particular, we test alternative models that includea capability based only on sales maximization. To measure

Table 5 Correlation matrix

160 observations in variations (Δ)were used

Variables 1 2 3 4 5 6 7 8 9 10

1. ΔARiq1(t+1) 1

2. ΔTobin’s qit .24 1

3. ΔCOMC it .20 .03 1

4. SRETiq1(t) .06 .14 .16 1

5. SRETiq2(t) .17 .03 .06 −.22 1

6. SRETiq3(t) −.15 .31 −.07 .14 −.44 1

7. SRETiq4(t) −.05 .15 .00 −.12 .29 −.22 1

8. UΔROAit .19 .21 −.09 −.11 .13 −.01 .06 1

9. ΔHHIit −.04 .01 .25 .17 .07 .05 −.01 −.04 1

10. ΔAssetsit −.10 −.26 .23 .39 .22 −.16 .13 −.33 .16 1

J. of the Acad. Mark. Sci.

Page 15

this capability, we use intertemporal output-oriented DEAwith variable returns to scale. We employ advertising andpromotion expenditures as inputs and sales as the output ofthe capability measurement. The results suggest that the sales-maximizing capability does not have a significant effect onTobin’s q and analysts’ recommendations. Therefore, it ismore appropriate to consider the measure of marketing capa-bility—that is, COMC—which we suggest in this article. Wealso perform models that include sales-maximizing capabilityand customer satisfaction and brand equity as control vari-ables in Eqs. 1 and 2. Estimations indicate that sales-maximizing capability along with brand equity does not havea significant impact on Tobin’s q and analysts’ recommenda-tions. Only customer satisfaction shows a significant andpositive effect on Tobin’s q (p <.10) and analysts’ recom-mendation (p <.05), which is in line with theory andextant empirical research examining the financial impli-cations of customer satisfaction (Anderson et al. 2004;Luo et al. 2010; Ngobo et al. 2012). To provide additionalrobustness, we also estimate the effect of COMC, includ-ing the impact of customer satisfaction and brand equityas control variables. COMC shows a significant and pos-itive effect on Tobin’s q and analysts’ recommendations(p <.05) beyond the independent effect of customer satis-faction. Thus, our measure of marketing capability seemsto have a better fit than the sales-maximizing capability,even with the addition of customer satisfaction and brandequity.

Discussion and conclusion

This research investigates the effect of COMC on forward-looking performance. We reveal the following key findings:First, COMChas a significant and positive effect on Tobin’s q.Second, COMC prompts analysts to issue better stock recom-mendations on average while reducing the dispersion of theirrecommendations. Third, the influence of COMC on Tobin’sq and analysts’ recommendations is significant after we con-trol for additional independent variables and perform severalrobustness checks.

One intriguing finding of this study relates to the strongerimpact of stage 1 capability on Tobin’s q, and the strongerimpact of stage 2 capability on analysts’ recommendations.Accounting and finance research helps understand this inter-esting finding better. In particular, accounting and financeresearch provides rationale about the behavior of the small/average investors and large (sometimes institutional or stockanalysts) investors (Bhattacharya 2001; Jegadeesh et al. 2004;Mikhail et al. 2007; Womack 1996). Tobin’s q can representthe stock response of the average investor, and it seems logicalto assume that this investor considers the most visible ele-ments of firms’ performance, such as sales and sales growth,in his valuation. The average investor reacts more favorably tosales information contained in stage 1 capability than stage 2capability because this information is easier for the averageinvestor to assimilate. However, stock analysts react morefavorably to stage 2 capability than to stage 1 capability

Table 6 COMC and forward-looking performance

Independent variables Hypotheses Model ΔTobin’s qit Model ΔTobin’s qit Model ΔARiq1(t+1) Model ΔARiq1(t+1)

Eq. 1 Eq. 1 Eq. 2 Eq. 2Estimates[standardized estimates]

Estimates[standardized estimates]

Estimates[standardized estimates]

Estimates[standardized estimates]

ΔCOMCit H1, H2 .22 [.08]* .81 [.33]*

ΔStage1it .11 [.13]** −.14 [−.17]† a

ΔStage2it .88 [.06]†a 4.64 [.34]**

SRETiq1t .80 [.22]*** .92 [.26]*** 1.81 [.54]** 1.51 [.45]**

SRETiq2t .77 [.31]*** 1.01 [.41]*** .34 [.14] .83 [.35]*

SRETiq3t .99 [.31]*** 1.13 [.36]*** −1.27 [−.42]* −.64 [−.21]

SRETiq4t .74 [.25]*** .74 [.25]*** −.28 [−.10] .05 [.02]

UΔROAit .46 [.04] .18 [.02] .94 [.09] 3.15 [.29]***

ΔHHIij −2.01 [−.07] −1.30 [−.05] −7.17 [−.27]† −6.49 [−.24]

Assets it −.83 [−.38]*** -.89 [−.41]*** −.74 [−.36] −.96 [−.46]

Year effects (λ t) Included Included Included Included

Constant −.06 .00 .05 .12

R-squared .41*** .45*** .27** .42***

Wald chi2 73.99 80.85 29.92 177.91

Year effects refer to the range from 2000 to 2006. 160 observations in variations (Δ) were used in the analysis. a refers to one-tailed test. We performedPCSE corrected for contemporaneous autocorrelation and heteroskedasticity

†p <.10, *p <.05, **p<.01, *** p<.001

J. of the Acad. Mark. Sci.

Page 16

because stage 2 capability provides intangible information,and analysts know how to manage and incorporate intangibleinformation to predict future benefits; this type of informationis usually more difficult to incorporate in predictions(Jegadeesh et al. 2004; Womack 1996).

Another intriguing finding from this study refers to theseemingly odd negative effect of stage 1 capability on ana-lysts’ recommendations. The contrarian model hypothesis(Lakonishok et al. 1994) helps explain this negative effect.Lakonishok et al. (1994) suggest that some investors follownaïve strategies such as extrapolating past sales growth too farinto the future. For example, naïve investors get overexcitedabout buying stocks from firms with high past sales growth sothat these stocks become overpriced, or these investorsoverreact to stocks of firms with low past sales growth,oversell them, and these stocks become underpriced. Thecontrarian model hypothesis (Lakonishok et al. 1994) positsthat investors who bet against naïve investors will have stocksthat outperform naïve investors’ stocks because contrarianstrategies invest disproportionately in stocks that areunderpriced and underinvest in stocks that are overpriced.Lakonishok et al. (1994, p. 1558) find that the effect of pastgrowth sales on future earnings is negative and Jegadeeshet al. (2004, p. 1097) corroborate such negative impact usinganalysts’ recommendations; these findings support the con-trarian model hypothesis. In our research, because one com-ponent of stage 1 capability is sales, this capability provides tothe market information about past-realized sales; therefore, ithas a negative impact on analysts’ recommendations (andtherefore on future earnings), in accordance with previouswork. Stage 2 capability, on the contrary, has to do with futuregrowth and therefore it positively impacts future earnings (aspredicted by analysts’ recommendations).

This study provides new insights to the literature on mar-keting capabilities. For instance, Dutta et al. (1999) study theimpact of marketing capability on Tobin’s q in the hightechnology industry. Dutta et al. (1999) find the effect ofmarketing capability on Tobin’s q to be .4791 (at p<.01).Our study reveals that the impact of stage 1 capability onΔTobin’s q is .11 (at p <.01), and the impact of stage 2capability on ΔTobin’s q is .88 (at p <.10). Coefficients ofDutta et al. (1999) and our paper are elasticities and thereforecan be compared. This study’s coefficient of stage 2 capabilityis larger than Dutta’s et al. coefficient of marketing capability.Interestingly, this research provides new insights about theimportant role of combining marketing resources, marketingintangibles, and sales into a multi-stage and dynamic process.Also, by changing sales for customer satisfaction and brandequity as the ultimate objectives within a marketing capabilitymulti-stage and dynamic process, companies will enjoy moregeneration of earnings in the future.

This study also helps clarify the mixed results found inMorgan et al. (2009a,) that indicate customer relationship

capabilities have a positive impact on margin growth; whilebranding capabilities have a negative impact on margingrowth. Our study reveals that a marketing capability thatcombines customer and branding capabilities (COMC) in anetwork process is the one that improves the bottom line andfuture earnings of organizations.

Implications for managers

The results have several implications for practitioners. Inparticular, this research contributes to practitioners’ decisionmaking processes. Managers need to incorporate into theirmarketing decision making process the connections amongdifferent marketing constructs—marketing resources, market-ing intangibles, consumer response, financial performance—in different periods of time. As we have demonstrated, theseconstructs are connected through a network of processes. Thisstudy demonstrates that the traditional connection, marketingresources linked to sales, is not enough to improve the bottomline of companies. Also, managers simply using marketingintangibles—customer satisfaction, brand equity—alone isnot sufficient to grow the bottom line of their organizationseither. Managers need to know how to incorporate the dynam-ics and network processes of marketing constructs to maketheir managerial practice really impactful.

COMC influences financial performance in the long run,which conforms to the beliefs of expert practitioners whoconsider brands and customers primary among the firm assetsthat can affect growth and profitability (Opinion ResearchCorporation 2008). As mentioned earlier, practitioners need tohave a network process mindset such that (1) the generation ofsales is dependent on current marketing efforts and intangiblesgenerated in the past and (2) the creation of marketing intangi-bles (e.g., brand equity, customer satisfaction) is dependent oncurrent marketing efforts, sales generation, and intangiblesgenerated in the past; the connection of these two sub-network processes (COMC) will influence the firm’s bottomline in the long run. Managers should not focus their marketingefforts on the short-term only (sub-process 1), because the futuresustainability of the firm depends on the long-term view (sub-process 2). Therefore, focusing marketing activities on currentsales could negatively affect the intangible developments, whichconstitute the grounds of future sustainability. Furthermore,behind the implementation of the network DEA, we find theconcept of global efficiency. Firms must be efficient in a globalsense in the entire marketing chain, rather than being efficient ineach shackle of the chain, because, for example, being the mostefficient only in selling could prohibit the generation of intan-gibles. Thus, we recommend that managers not forget the finalpurpose (intangible development) when seeking efficiency inthe different steps of the process.

We also suggest the inclusion of COMC in 10-K reportsbecause COMC has a net incremental effect on Tobin’s q,

J. of the Acad. Mark. Sci.

Page 17

beyond the effect explained by other control variables. Inclu-sion of COMC provides additional information to capitalmarkets and informs investors about the future potentialprofits of a firm. COMC and its customer value components(including brand equity and ACSI) would be beneficial addi-tions because these metrics offer more confidence to share-holders and potential investors.

This study further demonstrates that market analysts accountfor COMC in their recommendations, and as long as COMCandstage 2 capability improve, so do analysts’ recommendations.This finding may help firms address the challenge of attractingand retaining investors (Opinion Research Corporation 2008). Iffirms concentrate their marketing efforts on improving COMCand informing the market about this improvement, analysts willnotice and issue better stock recommendations, which shouldmotivate investors to stay with or purchase the stocks.

Implications for marketing theory and research

Our findings also offer theoretical implications. We first in-troduce COMC to the marketing literature and then theorizeand empirically test a model based on the process associatedwith translating marketing resources into ultimate objectivesof customer satisfaction and brand equity. Our introduction ofCOMC complements and empirically validates the marketingchain approach of Keller and Lehmann (2003) and Rust et al.(2004). Our study also contributes to the literature on market-ing capabilities (Dutta et al. 1999; Vorhies and Morgan 2005;Morgan 2012). The main issue this study intends to explain isrelated to the integration of specialized and cross-functionalcapabilities as well as with capability enhancement. AsMorgan(2012) points out, “most work in this area focuses on a firm’soverall marketing capabilities and fails to identify the specificcapabilities that make up the overall capability” (p. 114). Ourstudy disentangles marketing capabilities into sub-processesand connects these sub-processes, considering the dynamicnature of the relationship among marketing resources, consum-er response, and marketing intangibles. Also, our study incor-porates considerations of rivalry and competitors’ strategies inthe concept of COMC.

According to the finance literature, analysts possess deepknowledge about the firms they recommend because theyobserve those firms’ performance measures and intangibleinformation, such as marketing, innovation, human resources,strategic alliances, and technology. We demonstrate specifi-cally that analysts consider information about marketing ca-pabilities in issuing their recommendations. This finding addsto the marketing literature on the impact of marketing con-structs on analysts’ prediction of earnings and recommenda-tions (e.g. Luo et al. 2010; Ngobo et al. 2012).

This research contributes to the resource-based view andorganizational capability theories by conceptualizing market-ing capabilities from network and dynamic perspectives. This

study disentangles marketing capabilities into sub-processesand connects these sub-processes through dynamic relation-ships. This paper connects marketing constructs such as mar-keting resources, customer response, and marketing intangi-bles—which have been previously studied in isolation—anddevelops a framework to elucidate the impact of a customer-oriented marketing capability on Tobin’s q and analysts’ stockrecommendations. Our findings are consistent with the man-agement literature; intangibles and capabilities play crucialroles in sustaining the firm’s competitive advantage and cre-ating firm value (Amit and Schoemaker 1993; Barney 1991;Daniel and Titman 2006; Day 1994; Helfat and Peteraf 2003;Srivastava et al. 2001; Winter 2000; Zollo and Winter 2002).

Methodologically, our research provides a novel methodol-ogy to open the black box of marketing processes/capability—through network DEA. Network DEAmodels define a specificset of relations among inputs, intermediate outputs, and finaloutputs. Although becoming a complex mathematical problem,the resolution of each network DEA guarantees that the finaloutput (in our case the generation of marketing intangibles) isachieved if, and only if, the level of the intermediate outputs isaligned with this target. In other words, the results of theisolated optimization of the two different stages can produce afinal result significantly different from the optimal of the net-work DEA. Network DEA provides a global efficiency scorefor each firm. In essence, this methodology helps us understandthe world better and capture marketing capability phenomenadeeper. Although the basic DEA measures the input/outputprocess in a static fashion, network DEA overcomes this issueand can capture a process that is normally unobservable anddynamic. Therefore, network DEA allows for the measurementof deployments of different resources to serve customers betterin a dynamic fashion.

Limitations and opportunities for further research

This study has several limitations that provide worthwhileopportunities for further research. First, the sample firms areall large firms, and therefore research should examine small-and medium-sized firms. Second, we gathered secondaryinformation about brand equity and customer satisfactionfrom different sources, whereas primary research (e.g., sur-veys) could amplify our understanding of COMC and itseffect on financial performance. Third, our measure of mar-keting capability might extend to include other marketingresources, such as product development, as well as othermetrics of marketing-specific end results, such as channelequity and customer service (Srivastava et al. 2001). Fourth,measuring COMC using SFEs and computing brand equityusing Simon and Sullivan’s (1993) approach may supplypotential opportunities to provide methodological contribu-tions. Fifth, further research should investigate how to im-prove COMC, perhaps by turning to the knowledge-based

J. of the Acad. Mark. Sci.

Page 18

view of the firm. Marketing dynamics and organizationallearning might help reveal why some firms gain more finan-cial earnings from their marketing capability than others.

Acknowledgments The authors gratefully acknowledge the financialsupport from the Commissioner for Research and Universities of theCatalan Ministry for Innovation, Universities and Enterprise; the Euro-pean Social Fund; and the Spanish Ministry of Science and Education(projects: ECO2010-18967and SEJ2007-67895-C04-02). The authorsthank the editor and all five anonymous reviewers for providing importantinputs into the further development of this paper.

Appendix

Network DEA estimation

We estimate the network output-oriented DEA model withvariable returns to scale as follows:

min:βt;s:t: :

Xk¼1

K

λk ⋅xjkt ≤ x0jt; j ¼ 1;…; J ;

Xk¼1

K

λk ⋅yikt ≥ yit; i ¼ 1;…; I ;

Xk¼1

K

λk ¼ 1

Xk¼1

K

μk ⋅xjkt ≤ x0jt; j ¼ 1;…; J ;

Xk¼1

K

μk ⋅yikt ≤ yit; i ¼ 1;…; I ;

Xk¼1

K

μk ⋅yfkt ≥ yoft

.βt; f ¼ 1;…; F;

Xk¼1

K

μk ¼ 1;

λk ≥ 0; μk ≥ 0;

where β t is the network distance function for the unit underanalysis in period t . For the purposes of this study β t repre-sents the firm’s level of COMC. The term β t =1 indicates thatthe DMU (decision-making unit) under analysis is efficient,andβt<1 indicates that DMU is inefficient (the smaller the β t,the more inefficient the DMU is in generating intangibleresources). The term y it is the intermediate output vector ofthe DMU under analysis in period t (in our case, there is onlyone intermediate output: sales), xjt

o is the observed inputsvector of the DMU under analysis in period t (in our case,

the marketing resources spent in the current period and themarketing intangibles coming from the previous period), andy ft is the final output vector of the DMU under analysis inperiod t (in our case, the marketing intangibles at the end ofthe period under analysis). Finally, y ikt and x jkt refer to outputsand inputs vectors for the k (k=1, …, K ) DMUs forming thetotal sample, and λ and μ indicate the activity vector.

As Fig. 2 illustrates, this program has two steps, which aresolved simultaneously. Step 1 coincides with the restrictionsformed with the λ vector, and step 2 includes the remainingrestrictions, built with the μ vector as activity vector.

Previous works in the field of network DEA include Färeand Grosskopf (1996, 2000), Sexton and Lewis (2003), Lewisand Sexton (2004), Prieto and Zofío (2007), and Tone andTsutsui (2009). Our proposal extends the existent proposals inthe sense that (1) original inputs are taken into account notonly for the optimization of the intermediate but also for thefinal output and (2) the optimization of steps 1 and 2 isproduced simultaneously to maximize the final output, as theisolated optimization of step 1 does not guarantee the achieve-ment of the maximal output in step 2.

References

Ailawadi, K. L., Lehmann, D. R., &Neslin, S. A. (2003). Revenue premiumas an outcome measure of brand equity. Journal of Marketing, 67, 1–17.

Akdeniz, M. B., Gonzalez-Padron, T., & Calantone, R. J. (2010). Anintegrated marketing capability benchmarking approach to dealerperformance through parametric and nonparametric analyses.Industrial Marketing Management, 39(1), 150–160.

Ambler, T. (2000). Marketing metrics. Business Strategy Review, 11(2),59–66.

Ambler, T., Bhattacharya, C. B., Edell, J., Keller, K. L., Lemon, K. N., &Mittal, V. (2002). Relating brand and customer perspectives onmarketing management. Journal of Service Research, 5(1), 13–25.

Amit, R., & Schoemaker, P. J. H. (1993). Strategic assets and organiza-tional rent. Strategic Management Journal, 14(1), 33–46.

Anderson, E. W., Fornell, C., & Mazvancheryl, S. K. (2004). Customersatisfaction and shareholder value. Journal of Marketing,68(October), 172–185.

Anderson, E. W., & Sullivan, M. W. (1993). Antecedents and conse-quences of customer satisfaction for firms. Marketing Science,12(2), 125–143.

Arellano, M. (2003). Panel data econometrics. Oxford: Oxford UniversityPress.

Banker, R. D., Charnes, A., & Cooper, W. W. (1984). Some models forestimating technical and scale efficiencies in data envelopmentanalysis. Management Science, 30(9), 1078–1092.

Barney, J. (1991). Firm resources and sustained competitive advantage.Journal of Management, 17(1), 99–120.

Barth,M. E., Beaver,W. H., & Landsman,W. R. (2001). The relevance ofthe value relevance literature for financial accounting standard set-ting: another view. Journal of Accounting and Economics, 31(1–3),77–104.

Barth, M. E., Clement, M. B., Foster, G., & Kasznik, R. (1998). Brandvalues and capital market valuation. Review of Accounting Studies,3(1–2), 41–68.

J. of the Acad. Mark. Sci.

Page 19

Battese, G. E., & Coelli, T. J. (1988). Prediction of firm-level technicalefficiencies: with a generalized frontier production function andpanel data. Journal of Econometrics, 38, 387–399.

Bhattacharya, N. (2001). Investors’ trade size and trading responsesaround earnings announcements: an empirical investigation. TheAccounting Review, 76(2), 221–244.

Bolton, R. N. (1998). A dynamic model of the duration of the customer’srelationship with a continuous service provider: Rhe role of satis-faction. Marketing Science, 17(1), 45–65.

Bolton, R. N., & Drew, J. H. (1991). A longitudinal analysis of the impactof service changes on customer attitude. Journal of Marketing,55(1), 1–9.

Boulding, W. (1990). Unobservable and business performance: do fixedeffects matter? Marketing Science, 9(1), 88–91.

Branson, R. (1998). Making brand extensions work. Sales andMarketingManagement, 150(11), 84.

Chung, K., & Pruitt, S. W. (1994). A simple approximation of Tobin’s q.Financial Management, 23(3), 70–74.

Coelli, T. J., Prasada-Rao, D. S., O’Donnell, C. J., & Battese, G. E.(2005). An introduction to efficiency and productivity analysis .New York: Springer.

Collins, D. W., & Kothari, S. P. (1989). An analysis of intertemporal andcross-sectional determinants of earnings response coefficients.Journal of Accounting and Economics, 11(2–3), 143–181.

Damodaran, A. (2002). Investment valuation: Tools and techniques fordetermining the value of any asset . New York: Wiley Finance.

Daniel, K., & Titman, S. (2006). Market reactions to tangible and intan-gible information. Journal of Finance, 61(4), 1605–1643.

Day, G. S. (1994). The capabilities of market-driven organizations.Journal of Marketing, 58(October), 37–52.

Day, G. S. (2006). Aligning the organization with the market.MIT SloanManagement Review, 48(1), 41–49.

Donthu, N., Hershberger, E. K., & Osmonbekov, T. (2005).Benchmarkingmarketing productivity using data envelopment anal-ysis. Journal of Business Research, 58(11), 1474–1482.

Dutta, S., Narasimhan, O., & Rajiv, S. (1999). Success in high-technology markets: is marketing capability critical? MarketingScience, 18(4), 547–568.

Dutta, S., Narasimhan, O., & Rajiv, S. (2005). Conceptualizing andmeasuring capabilities. Strategic Management Journal, 26 (3),277–285.

Färe, R., & Grosskopf, S. (1996). Productivity and intermediate products:a frontier approach. Economics Letters, 50(1), 65–70.

Färe, R., & Grosskopf, S. (2000). Network DEA. Socio-EconomicPlanning Sciences, 34(1), 35–49.

Fornell, C. (1992). A national customer satisfaction barometer: theSwedish experience. Journal of Marketing, 56(January), 6–21.

Fornell, C., Johnson, M. D., Anderson, E. W., Cha, J., & Bryant, B. E.(1996). The American customer satisfaction index: nature, purposeand findings. Journal of Marketing, 60(October), 7–18.

Gulati, R. (2007). Silo busting: how to execute on the promise ofcustomer focus. Harvard Business Review, 85(May), 98–108.

Gulati, R. (2009). Re-organize for resilience: Putting customers at thecenter of your business . Boston: Harvard Business SchoolPublishing.

Helfat, C. E., & Peteraf, M. A. (2003). The dynamic resource-based view:capability lifecycles. StrategicManagement Journal, 24(1), 997–1010.

Hou, K., & Robinson, D. T. (2006). Industry concentration and averagestock return. Journal of Finance, 61(4), 1927–1956.

Howe, J. S., Unlu, E., & Yan, X. (2009). The predictive content ofaggregate analyst recommendation. Journal of AccountingResearch, 47(3), 799–821.

Hunt, S. D., &Morgan, R. M. (1995). The comparative advantage theoryof competition. Journal of Marketing, 59(April), 1–15.

Jacobson, R. (1990). Unobservable effects and business performance.Marketing Science, 9(1), 74–85.

Jacobson, R., & Mizik, N. (2009). The financial markets and customersatisfaction: re-examining the value implications of customer satis-faction from the efficient markets perspective. Marketing Science,28(5), 810–819.

Jegadeesh, N., Kim, J., Krische, S. D., & Lee, C.M. C. (2004). Analyzingthe analysts: when do recommendations add value? Journal ofFinance, 59(3), 1083–1124.

Kamakura, W. A., Ratchford, B. T., & Agrawal, J. (1988). Measuringmarket efficiency and welfare loss. Journal of Consumer Research,21(December), 289–302.

Keller, K. L. (1993). Conceptualizing, measuring, and managingcustomer-based brand equity. Journal of Marketing, 57(January),1–22.

Keller, K. L., & Lehmann, D. R. (2003). How do brands create value?Marketing Management, 12(3), 26–31.

Kohli, A. K., & Jaworski, B. J. (1990). Market orientation: the construct,research propositions, and managerial implications. Journal ofMarketing, 54(April), 1–18.

Kormendi, R.,&Lipe, R. (1987). Earnings innovations, earnings persistence,and stock returns. Journal of Business Research, 60(3), 323–345.

Kothari, S. P. (2001). Capital markets research in accounting. Journal ofAccounting and Economics, 31(1–3), 105–231.

Krasnikov, A., & Jayachandran, S. (2008). The relative impact of mar-keting, research and development, and operations capabilities onfirm performance. Journal of Marketing, 72(July), 1–11.

Lakonishok, J., Shleifer, A., & Vishny, R. W. (1994). Contrarian invest-ment, extrapolation, and risk. Journal of Finannce, 49(5), 1541–1578.

Lewis, H. F., & Sexton, T. R. (2004). Network DEA: efficiency analysisof organizations with complex internal structure. Computers &Operations Research, 31(9), 1365–1410.

Luo, X., & Donthu, N. (2005). Assessing advertising media spendinginefficiencies in generating sales. Journal of Business Research,58(1), 28–36.

Luo, X., & Donthu, N. (2006). Marketing’s credibility: a longitudinalinvestigation of marketing communication productivity and share-holder value. Journal of Marketing, 70(October), 70–91.

Luo, X., Homburg, C., & Wieseke, J. (2010). Customer satisfaction,analyst stock recommendation, and firm value. Journal ofMarketing Research, 47(6), 1041–1058.

Madden, T. J., Fehle, F., & Fournier, S. (2006). Brands matter: an empiricaldemonstration of the creation of shareholder value through branding.Journal of the Academy of Marketing Science, 34(2), 224–235.

Makadok, R. (2001). Toward a synthesis of the resource-based anddynamic capability views of rent creation. Strategic ManagementJournal, 22, 387–401.

Marketing Science Institute. (2012). Research priorities 2012–2014 .Cambridge: Marketing Science Institute.

Mikhail, M. B., Walther, B. R., & Willis, R. H. (2007). When securityanalysts’ talk, who listens? Accounting Review, 82(5), 1227–1253.

Mittal, V., Anderson, E. W., Sayrak, A., & Tadikamalla, P. (2005). Dualemphasis and the long-term financial impact of customer satisfac-tion. Marketing Science, 24(4), 544–555.

Mizik, N., & Jacobson, R. (2008). The financial value impact of percep-tual brand attributes. Journal of Marketing Research, 45(February),15–32.

Morgan, N. A. (2012). Marketing and business performance. Journal ofthe Academy of Marketing Science, 40(1), 102–119.

Morgan, N. A., Slotegraaf, R. J., & Vorhies, D. W. (2009a). Linkingmarketing capabilities with profit growth. International Journal ofResearch in Marketing, 26(4), 284–293.

Morgan, N. A., Vorhies, D. W., & Mason, C. H. (2009b). Marketorientation, marketing capabilities, and firm performance.Strategic Management Journal, 30, 909–920.

Mukherjee, A., Nath, P., & Pal, M. (2003). Resource, service quality andperformance triad: a framework for measuring efficiency of bankingservices. Journal of theOperational Research Society, 54(7), 723–735.

J. of the Acad. Mark. Sci.

Page 20

Narasimhan, O., Rajiv, S., & Dutta, S. (2006). Absorptive capacity inhigh-technology markets: the competitive advantage of the haves.Marketing Science, 25(5), 510–524.

Nath, P., Nachiappan, S., & Ramanathan, R. (2010). The importance ofmarketing capability, operations capability, and diversification strat-egy on performance: a resource-based view. Industrial MarketingManagement, 32(2), 317–329.

Ngobo, P. V., Casta, J. F., & Ramond, O. (2012). Is customer satisfaction arelevant metric for financial analysts? Journal of the Academy ofMarketing Science, forthcoming, 40(3), 480–508.

Opinion Research Corporation (2008). NYSE euronext ceo report 2009 .Available from: http://www.nyse.com/pdfs/2009_CEO_Report.pdf(retrieved October 12, 2009).

Pergelova, A., Prior, D., & Rialp, J. (2010). Assessing advertising efficien-cy: does the internet play a role? Journal of Advertising, 39(3), 39–54.

Piercy, N. F. (1987). The marketing budgeting process: marketing man-agement implications. Journal of Marketing, 51(October), 45–59.

Prieto, A. M., & Zofío, J. L. (2007). Network DEA efficiency in input–output models: with an application to OECD countries. EuropeanJournal of Operational Research, 178(1), 292–304.

Ramaswami, S., Srivastava, R. K., & Bhargava, M. (2009). Market-basedcapabilities and financial performance of firms: Insights into mar-keting contribution to firm value. Journal of the Academy ofMarketing Science, 37(2), 97–116.

Rao, V. R., Agarwal, M. K., & Dahlhoff, D. (2004). How is manifestbranding strategy related to the intangible value of a corporation?Journal of Marketing, 68(October), 126–141.

Rappaport, A. (1998).Creating shareholder value: A guide for managersand investors . New York: The Free Press.

Roodman, D. (2006). How to do xtabond2: An introduction to differenceand systemGMM in stata. Working Paper N°103, Center for GlobalDevelopment.

Rust, R. T., Ambler, T., Carpenter, G. S., Kumar, V., & Srivastava, R. K.(2004). Measuring marketing productivity: current knowledge andfuture directions. Journal of Marketing, 68(October), 76–89.

Rust, R. T., Zahorik, A., & Keiningham, T. (1995). Return on quality(ROQ): making service quality financially accountable. Journal ofMarketing, 59(April), 58–70.

Scherer, F. M. (1980). Industrial market structure and economic perfor-mance . Boston: Houghton Mifflin.

Schmalensee, R. (1977). Using the H-index of concentration with pub-lished data. Review of Economics and Statistics, 59(2), 186–193.

Sexton, T. R., & Lewis, H. F. (2003). Two-stage DEA: an application tomajor league baseball. Journal of Productivity Analysis, 19(2–3),227–249.

Sharpe, W. F. (1966). Mutual fund performance. Journal of Business,39(1), 119–138.

Sheth, J. N., Sisodia, R. S., & Sharma, A. (2000). The antecedences andconsequences of customer-centric marketing. Journal of theAcademy of Marketing Science, 28(1), 55–66.

Simon, C. J., & Sullivan, M. W. (1993). The measurement and determi-nants of brand equity: A financial approach. Marketing Science,12(1), 28–52.

Slotegraaf, R. J., & Pauwels, K. (2008). The impact of brand equity andinnovation on the long-term effectiveness of promotions. Journal ofMarketing Research, 45(June), 293–306.

Song, M. C., Di Benedetto, A., & Nason, R. W. (2007). Capabilitiesand financial performance: the moderating effect of strategictype. Journal of the Academy of Marketing Science, 35 (1),18–34.

Song, M. C., Droge, C., Hanvanich, S., & Calantone, R. (2005).Marketing and technology resource complementarity: an analysisof their interaction effect in two environmental context. StrategicManagement Journal, 26(3), 259–276.

Song, M. C., Nason, R. W., & Di Benedetto, C. A. (2008). Distinctivemarketing and information technology capabilities and strategictype: a cross-national investigation. Journal of InternationalMarketing, 16(1), 4–38.

Srinivasan, S., &Hanssens, D. (2009).Marketing and firm value: metrics,methods, findings, and future directions. Journal of MarketingResearch, 46(3), 293–312.

Srinivasan, V., Park, C. S., & Chang, D. R. (2005). An approach to themeasurement, analysis, and prediction of brand equity and itssources. Management Science, 51(9), 1433–1448.

Srivastava,R.K., Fahey,L.,&Christensen,H.K. (2001). The resource-based view and marketing: the role of market-based assets ingaining competitive advantage. Journal of Management, 27 (6),777–802.

Srivastava, R. K., Shervani, T. A., & Fahey, L. (1998). Market-basedassets and shareholder value: a framework for analysis. Journal ofMarketing, 62(January), 2–18.

Teece, D. J., Pisano, G., & Shuen, A. (1997). Dynamic capabilities andstrategic management. Strategic Management Journal, 18(7), 509–533.

Tone, K., & Tsutsui, M. (2009). Network DEA: a slacks-based measureapproach. European Journal of Operational Research, 197(1), 243–252.

Tulkens, H., & Vanden Eeckaut, P. (1995). Non-parametric effi-ciency, progress and regress measures for panel data: method-ological aspects. European Journal of Operational Research,80 (3), 474–499.

Verhoef, P. C. (2001). Analyzing customer relationships: Linking rela-tional constructs and marketing instruments to customer behavior.Doctoral Dissertation, Tilburg Institute, Rotterdam.

Vorhies, D. W., & Morgan, N. A. (2005). Benchmarking marketingcapabilities for sustainable competitive advantage. Journal ofMarketing, 69(January), 80–94.

Vorhies, D. W., Orr, L. M., & Bush, V. D. (2011). Improving customer-focused marketing capabilities and firm financial via marketingexploration and exploitation. Journal of the Academy of MarketingScience, 39, 736–756.

Wilson, P. W. (2008). FEAR 1 .11 user ’s guide. Retrieved August 10,2009 from http://business.clemson.edu/Economic/faculty/wilson/Software/FEAR/fear.html

Winter, S. J. (2000). The satisficing principle in capability learning.Strategic Management Journal, 21(10–11), 981–996.

Womack, K. L. (1996). Do brokerage analysts’ recommendations haveinvestment value? Journal of Finance, 51(1), 137–167.

Wooldridge, J. (2000). Introductory econometrics: A modern approach .New York: South-Western College Publishing.

Zollo,M., &Winter, S. J. (2002). Deliberate learning and the evolution ofdynamic capabilities. Organization Science, 13(3), 339–353.

J. of the Acad. Mark. Sci.