Allen M. Weiss, Erin Anderson. & Deborah J. Maclnnis Reputation Management as a Motivation for Sales Structure Decisions The authors examine whether reputation concerns affect how manufacturers structure their sales organization. Us- ing reputation theory, they examine whether reputation-related perceptions and beliefs affect whether a manufac- turer that currently uses an outside selling organization {i.e., a "rep") intends to vertically integrate the selling func- tion or switch to a new rep. In particular, they propose that a manufacturer's intentions to replace its current manufacturers' rep with a company sales force or a different rep is a function of its perceptions of the reputation of itself and the rep and its beliefs about how high-reputation manufacturers in the industry typically organize their sell- ing function. Survey data support the plausibility of these reputation-based arguments as factors that influence sales organization structure decisions. These results provide some important extensions to reputation theory. The authors discuss the study's implications for both managerial behavior and the literature on channels and organiza- tional governance.

n important topic in marketing involves how manu- they appear consistent with their preconceived intentions to facturers choose the structure of their sales organiza- change. Thus, the decision to vertically integrate the selling Ation (e.g., Anderson 1985; Anderson and Coughlan function or change reps appears to be psychologically moti- 1987; Klein, Frazier, and Roth 1990). Manufacturing firms vated, with "economic" rationales assembled as postdeci- often create their sales organization by outsourcing the per- sion justifications. sonal selling function to an independent organization (i.e., a Moreover, observations of such decisions in practice manufacturers' "rep"). Over time, manufacturers may de- suggest the existence of specific psychological motiva- cide to change their sales organization structure by vertical- tions—those grounded in the decision's impact on the firm's ly Integrating the selling function or maintain the structure reputation. Research in management, sociology, and psy- but switch to a different rep (Corey, Cespedes, and Rangan chology is consistent with the idea that managerial decisions 1989; Weiss and Kurland 1997). Currently, research ad- are affected by consideration of tbe firm's reputation dressing the motivations underlying these decisions is limit- (Bromely 1993; Fombrun and Shanley 1990). However, the ed in at least two ways. effect of reputation on sales organization decisions has been A first limitation is that, though several theories have neither theoretically proposed nor empirically tested. been proposed to explain manufacturers" motivations for us- A second limitation of current research on a firm's deci- ing either a rep or a vertically integrated sales organization sion to outsource or integrate its selling function is the as- structure, most are tied to economic-based efficiency or ef- sumption that there was no structure in place and that firms fectiveness motivations (Heide 1994) with an emphasis on select the most appropriate structure for going forward. the role of transaction-specific assets (e.g., Anderson 1985, However, the presence of a preexisting system appears to af- 1988; Heide and John 1988). Although economic motiva- fect the system going forward (Anderson 1985; Anderson tions undoubtedly affect sales structure decisions (Powers and Couglan 1987). Furthermore, changing from one system 1987), findings by Weiss and Anderson (1992) suggest otb- to another is not costless. As Weiss and Anderson (1992) er motivations. Their results indicate that manufacturers ma- demonstrate, perceived switching costs indeed do deter nipulate the perceived costs of changing their current rep so manufacturers from switching from a manufacturers' rep to an employee sales force. Thus, it is useful to consider how firms make sales organization structure decisions, assuming Allen M. Weiss is Associate Professor of Marketing, University of Southern a structure is already in place. California. Enn Anderson is Professor of Marketing, INSEAD, Deborah J. The current research builds on these ideas by examining Maclnnis is Associate Professor of Marketing, University of Southern Cal- ifornia.This research was supported by the Marketing Science Institute. whetber reputation concerns influence how manufacturers Tfie authors thank Bruce Hardie and Vijay Mehrotra for their capable re- intend to structure their sales organization. Our study focus- search assistance. The first author is a David and Jeanne Tappen Fellow es on manufacturers that currently use an independent sales of Marketing. The second author was a visiting professor at the European force. We ask whether their intentions to integrate vertically Institute for Advanced Studies in Management (Brussels) and Ihe Catholic or switch to a different rep are driven by reputational con- University of Mons (Mons, Belgium) while working on this project and is cerns. Reputation theory (Bromley 1993; Dranove and currently the John Loudon Chaired Professor of International Manage- Shanley 1995; Emier 1990; Fombrun and Shanley 1990) ment. guides our empirical predictions. In the following sections

Journal of Marketing 74 / Journal of Marketing, October 1999 Vol. 63 (October 1999), 74-89 we review literature relating to reputation constructs and de- nis 1986), Thus, United Airlines cultivates an image of velop explicit hypotheses. We present the results of the em- friendliness, Apple an image of ease of use, Rolex an image pirical study and end with a discussion of its limitations, im- of luxury, and Campbell's an image of bome cooking. Rep- plications, and directions for further research. utation, in contrast, reflects an overall judgment regarding Our study makes several contributions. First, we offer the extent to which a firm is held in high esteem or regard, theoretical grounding and empirical support for the idea that not the specific identity it has. Thus, whereas image reflects sales organization structure decisions are influenced by rep- what a firm stands for, reputation reflects how well it has utation concerns. Second, we demonstrate tbe unique in- done in the eyes of the marketplace, image and reputation sights gained by examining a manufacturer's perceptions of are distinct concepts as each can vary independent of the its own reputation, its rep's reputation, and the reputation other. A firm can change its image through repositioning, gap between the two. Third, we identify managerial impli- though its reputation remains intact. Thus, the "softer side of cations that stem from our findings. Sears" campaign has altered customers' perceptions of what Sears' stands for (its Image), though it has not affected the esteem or regard with which customers view Sears. (Sears Conceptual Background has been and still is regarded as a highly reputable firm.) Conversely, a firm's reputation may be affected negatively, What Is Reputation? even though its image remains intact. For example, a restau- Reputation is defmed as "an impression of public esteem or rant with an image of "great fast food" may be regarded as high regard judged by others" {Merriam Webster's Colle- nonreputable if it is discovered tbat its owner is engaged in giate Dictionary 1996, p. lOOI). Several aspects of the def- illicit activities. Notably, tbougb, this negative reputation inition warrant attention. First, it is Important to clarify the need not alter customers' image of the restaurant as a place meaning of "public esteem" and "high regard." Prior work that serves great fast food. suggests an organization is held in public esteem or high re- Second, the extent to which customers view a specific gard when it is viewed as botb visible and credible (e.g., es- image as desirable is likely to be specific to a market seg- tablisbed, professional, and a stable player in tbe market- ment. As such, a firm's image may not necessarily be desir- place) (Fombrun 1996; Fombrum and Sbanley 1990). able to customers even tbough they may agree that its visi- Altbough individual persons may have perceptions of an or- bility and credibility render its reputation highly positive. ganization's reputation for specific things (e.g., its reputa- Thus, Rolex may have an image of luxury and a highly fa- tion for fairness, quality, and good hiring practices), re- vorable reputation as a company, even though the image of search suggests that persons tend to make global evaluations luxury may not be desirable to all customers. In contrast, of an organization's reputation. Hence, reputation tends to customers are likely to regard a favorable reputation as de- be regarded as a unidimensional, not multidimensional, con- sirable no matter wbat segment they belong to. struct (Thcus 1993; Yoon, Guffey, and Kijewski 1993). In accord with these ideas, we view reputation as a global per- Basic Elements of Reputation Theory and Its Link ception of the extent to which an organization is held in high to Marketing esteem or regard. Reputation theory (Bromley 1993; Emler 1990) suggests Second, it is useful to consider the entities whose repu- that a social entity (e.g., a person, an organization) engages tations may be evaluated (Bromley 1993). In a sales organi- in several reputation-related processes. It actively monitors zational structure context, a manufacturer may consider the reputations—both its own and otbers'. Evidence suggests reputation of at least two entities; itself and its current rep. firms indeed are interested in monitoring their own reputa- In the present study, we examine the reputation of both. We tion (Bromely 1993; Leuthesser 1988). A well-recognized assess these constructs separately and examine the impact of measure by Owen (1993), for example, examines tbe esteem any gap between them. witb which customers hold various brands and companies. Third, it is important to specify the "other people" who Manufacturers also may be concerned about monitoring the judge reputations. Multiple external publics may judge an reputation of their rep. This idea is consistent with agency organization's reputation (e.g., industry analysts, stockhold- theory (Bergen, Dutta, and Walker 1992). That literature ers). However, in a sales organization context the manufac- views monitoring of an agent (i.e., a rep) as important be- turer is likely to be concerned particularly about how cus- cause it assumes that tbe extent to wbich a manufacturer de- tomers perceive it and Its rep (whether veridical or not). signs an efficient contract should be linked, in part, with its Thus, we focus on the manufacturer's perceptions of the ex- ability to monitor its rep's activities. tent to which customers hold the manufacturer and its rep in Monitoring gives rise to perceptions of the extent to high regard. which the entity's reputation is good or bad. Monitoring al- Fourth, it is important to differentiate conceptually "rep- so facilitates the development of reputation-related beliefs. utation" from the more specific term "image." Reputation Such beliefs may be described as either causal or descriptive and image are conceptually similar, because both reflect (Fishbein and Ajzen 1975; Nisbett and Ross 1980). Tbe for- perceptions of an entity. However, they are conceptually dis- mer reflect the extent to which a specific action is viewed as tinct in several important ways. instrumental in affecting a firm's reputation. In a causal be- First, an image reflects a set of associations linked to a lief statement, the observer believes that X drives Y. For ex- brand or company name that summarizes a brand or firm's ample, a manufacturer may have strong beliefs tbat its own identity (i.e., what it stands for; Park, Jaworski, and Macln- poor reputation is caused by its affiliation witb a certain rep.

Reputation Management / 75 Descriptive beliefs, in contrast, reflect a set of associations with the current structure but switching reps. If so, such ac- tied to a social category. For example, a manufacturer may tions may be direct efforts on the part of the manufacturer to attach to its schema of "highly reputable firms" beliefs manage its reputation. In the following sections, we use rep- about the characteristics associated with these firms (e.g., utation theory to offer insight into these issues. highly reputable firms use their own sales force). Descrip- tive beliefs simply represent observations without reference to causality. Hypotheses Writings by practitioners indicate that practitioners in- The Perceived Gap in Reputation and Beliefs as deed do hold reputation-related beliefs. Lebell (1971, 1977), Bases for Actions a practitioner-writer, writes that reputable manufacturers support the rep better and improve the rep's credibility. Sim- The gap in perceived reputation. A fundamental princi- ilarly, Lavin (1991) of Lavin Associates, a consulting firm ple of reputation theory is that a social entity is an assiduous that specializes in consulting to reps and manufacturers, monitor of its own reputation and that it actively attempts to writes that the lines the rep carries affect its reputation. That uncover the reputations of others (Bromley 1993; Emier these beliefs exist is also evident in our own discussions 1990). In a sales organization structure context, this sug- with manufacturers and reps. gests that a manufacturer may monitor customers' percep- tions of both its own reputation and the reputation of its rep. In addition to monitoring reputations, reputation theory Moreover, because a manufacturer may monitor the reputa- proposes that a social entity also acts to manage its reputa- tion of both entities, the two reputations may be compared. tion (Bromely 1993). Reputation management is a general Social comparison theory (Festinger 1954) purports that so- phenomenon that need not occur in response to a reputation cial entities are motivated to compare their abilities and problem. In other words, whereas an entity with a poor rep- characteristics with referent others. Such referent others utation may engage in actions that enhance its reputation, may provide a frame of reference against which one's own even an entity with a favorable reputation may engage in ac- characteristics are judged (Tversky and Kahneman 1981). tions designed to sustain or enhance its reputational stand- ing. Although ideas regarding reputation management have Prior research indicates that social comparisons are en- not received much attention in the marketing literature, the demic in an organizational context. They occur among per- notion that a manufacturer is concerned about managing its sons within a firm (e.g., Sweeney, McFarlin, and Interri- reputation seems plausible, because reputation perceptions dieden 1990), between organizational members and the firm are linked with outcomes deemed important to the firm (for as a whole (Dutton, Dukerich, and Harquail 1994), and reviews, see Bromley 1993; Yoon, Guffey, and Kijewski among firms (e.g., Elsbach and Kramer 1996). Such com- 1993). For example, favorable reputation perceptions have parisons are particularly likely when the dimension of com- been linked with a firm's ability to survive crises (Shrivas- parison is viewed as important or highly relevant to the so- tavas and Siomkos 1989), positive customer attitudes to- cial entity. Because reputations are regarded as important ward the company's products and salespeople (Brown and relevant to firms, we anticipate that a social comparison 1995), enhanced buying intentions (Yoon, Guffey, and Ki- process regarding reputation indeed does operate among jewski 1993), and choice (Traynor 1983). firms. The reputation of the manufacturer and the rep are Ideas regarding reputation monitoring and management comparable because, though they may be based on different have not been applied directly to marketing activities, in- things (e.g., the ability to produce high-quality products, the cluding the sales organization structure decision.' Yet, a ability to manage customer relationships through selling), firm's sales force is the face of the company to what is ar- each is a perception of how well the manufacturer or selling guably its most important group of constituents: its cus- organization has done in the eyes of the marketplace. In a tomers. The company's survival and prosperity depends on sense, the reputation is the grade the customer assigns the customers' reactions to its product offering, a reaction heav- organization or the rep. Because they do different things, the ily dependent on the competence of the firm's sales force. grades are computed differently, but an "A" grade still Thus, il seems important to examine whether reputation re- means high regard. lated perceptions and beliefs (derived from monitoring ac- If a firm does compare its reputation with referent oth- tivities) affect a manufacturer's intentions to change its sales ers', such monitoring may give rise to a gap between the organization structure by vertically integrating or by staying perceived reputation of one entity (e.g., the manufacturer) and another (e.g., its rep). Reputation gap is defined as the distance between the perceived reputation of two entities. •In using reputation theory to predict the intentions of a manu- facturer, we often a.ssume an analogy between an individual person The distance between the perceived reputation of the and an organization. In his review of organizational theory on this two entities has two critical components: direction and mag- issue, Rousseau (1985) suggests such analogies be made with cau- nitude. Direction refers to whether the manufacturer's repu- tion. The analogy is most likely to apply in settings where a few tation is more favorable (a positive gap) or less favorable (a people are key players and where a corresponding economic ratio- negative gap) than that of its rep. Magnitude reflects the ex- nale can be developed for social-psychological processes. In our tent to which such a gap (whether positive or negative) is hypothesis development in the next section, we note such ratio- large or small. Thus, a large positive gap exists when the nales. Our setting, as noted in the "Methodology" section, is one in which the key {though not the only) players are sales managers and manufacturer perceives its own reputation as far more posi- sales agency heads. tive than that of its rep. No gap exists when the manufactur- er perceives its reputation as similar to that of its rep. A large

76 / Journal of Marketing, October 1999 negative gap exists wben the manufacturer perceives its rep- lief variable provides a fairly strong lest of reputation tbeo- utation as far less positive than that of its rep. Notably, al- ry. Although firms may intend to act in a manner consistent though reputation theory assumes social entities are assidu- with what they believe causally affects their own reputation, ous reputation monitors, the concept of a reputation gap is it is not obvious that descriptive beliefs about what highly an extension of the theory. Hence, the subsequent "gap" hy- reputable firms do are sufficiently salient or important to af- potheses are regarded as exploratory. Moreover, although fect such intentions. we have found no publisbed literature describing the extent to which manufacturers and sales reps develop perceptions Perceptual gap-belief consistency. If reputation does af- of a reputation gap, that perceptions of a reputation gap are fect sales organization structure decisions in a manner con- formed seems to be natural because manufacturers and reps sistent with reputation management theory, we should find are assumed to be active monitors of their own reputation that reputation perceptions are more likely to affect those and that of other entities. To the extent that evidence impli- reputation management actions that are consistent with rep- cating effects for a reputation gap are observed, tbis re- utation-related beliefs. We therefore expect that the reputa- searcb will be tbe first to establish tbe importance of the gap tion gap and beliefs about the efficacy of reputation-related concept. actions jointly affect tbe reputation-management actions in- tended by the manufacturer. Thus, the more a manufacturer The motivational properties of a positive perceived gap. perceives that its own firm's reputation is better tban the If the manufacturer does compare its reputation with that of reputation of its rep (i.e., there is a positive gap), the more its rep, the existence of a positive gap may affect the manu- the manufacturer should manage this reputation gap in a facturer's desire to take action to manage this gap. There are manner consistent with its reputation-related beliefs. The two reasons to expect that a large positive gap heightens the more strongly it believes that highly reputable firms use manufacturer's need to act. First, because tbe manufacturer tbeir own sales forces, the more it should manage this repu- is concerned about its reputation, a perception of its rep as tation gap by doing what it believes highly reputable firms having a reputation significantly inferior to its own may lead do—using a vertically integrated sales organization. Specif- a manufacturer to believe that the rep's reputation is costing ically, we expect the following: the manufacturer sales and perhaps margin. Framing theory H|a:The manufacturer's intentions to vertically integrate in- predicts that social entities are generally sensitive to mone- crease wilh combined increases in (I) the positive gap be- tary losses and that the effects of such losses loom larger tween the manufacturer's perception of its own reputation than gains (Kahneman and Tversky 1979; Tversky and Kah- and that of its rep and (2) its belief thai highly reputable neman 1991). Because manufacturers generally are averse firms use their own sales force. to loss, they may be motivated to act to reduce these eco- Whereas H[a focuses on belief-consistent reputation nomic costs. Second, a reputation gap also may motivate the management actions, a direct implication of the theory is manufacturer with a significantly favorable reputation as that a manufacturer should be less likely to perform belief- compared with its rep to act because it may believe that it inconsistent reputation management actions. If a manufac- has outgrown its rep and is capable of either attracting a rep turer perceives a large positive reputation gap and strongly with a better reputation ("moving up") or performing tbe believes that highly reputable firms use tbeir own sales selling function better itself force, one action inconsistent with this belief is to change The role of reputation-related beliefs. Desire to manage reps. It should therefore not manage this gap by doing some- a reputation, however, does not provide any indication of the thing different from what it believes reputable firms do (e.g., reputation-management action taken. Reputation theory pre- cbange reps). Thus, to the extent that switching reps is a vi- dicts that another factor is necessary to understand the spe- able action, we should observe tbe following: cific actions taken to enhance or sustain a reputation. More Hit,: The manufacturer's intentions to switch to a new rep de- specifically, reputation perceptions affect reputation man- crease with combined increases in (I) the positive gap be- agement actions when they are consistent with reputation- tween the manufacturer's perception of its own reputation related beliefs. Because our previous discussion of tbe gap and that of its rep and (2) its belief that highly reputable concept explicitly focuses on reputation perceptions (of tbe firms use their own sales force. manufacturer as compared with its rep) and not beliefs, we clarify here the concept of reputation-related beliefs. Decoupling Reputation Perceptions Previously, we identified several classes of reputation- The following sections extend our analysis of reputation related beliefs. We focus here on descriptive beliefs— perceptions and beliefs in two ways. First, although the gap specifically, the extent to which the manufacturer strongly concept provides an interesting extension of reputation the- believes that higbly reputable firms use their own sales ory, it focuses only on the relative difference in direction force. We focus on tbis belief variable for two reasons. First, and magnitude in reputation perceptions between the manu- beliefs about tbe practices of highly reputable firms may facturer and its rep. It therefore ignores the absolute level of serve as a frame of reference, prototype, or example of "best reputation perceptions of tbe manufacturer and/or its rep. practice." Thus, wbereas tbe rep may provide one frame of Two different gaps could be equal in magnitude and direc- reference against which the firm's reputation is judged, the tion yet represent vastly different reputation levels for both practices of highly reputable firms may provide another the manufacturer and its rep. Second, the gap concept ob- frame of reference against which reputation-related sales or- fuscates the distinction between the effects of the manufac- ganization structure issues are considered. Second, this be- turer's reputation and those of the reputation of its rep. As

Reputation Management / 77 such, it is difficult to tell whether perceptions of the manu- their reputations, we might expect that a manufacturer's mo- facturer's reputation and the rep's reputation each interact tivation to change its current rep relationship might be with beliefs to affect sales organization structure intentions strong wben it perceives its own reputation as poor. Howev- or whether the reputation of one entity (the manufacturer or er, we do not believe that a manufacturer with a poor repu- the rep) plays a role and the other does not. We explore these tation attempts to enhance its reputation by changing its issues in the following sections. sales organization structure. Several factors motivate our reasoning. The rep's reputation and the manufacturer's beliefs as Note that a manufacturer may attribute a poor reputation bases for actions. A manufacturer's motivation for taking to either external factors or its own (poor) efforts. Let us reputation management actions is likely to surface when its lake each situation in turn. First, it is highly likely that a rep's reputatioti is perceived to be poor or low. The notion manufacturer will attribute a poor reputation to external fac- that a rep with a poor reputation would motivate a manufac- tors. Tbis is predicted on the basis of social entities being bi- turer to manage this situation actively is cotisistent wilh the ased in the extent to which they attribute success or failure "fundamental attribution error" (e.g., Ross 1977). Specifi- to themselves or others (Myers and Bach 1976), If a poor cally, social entities tend to attribute negative outcomes that reputation is attributed to external factors, it is questionable accrue to another entity (e.g., sales below expectations whether this poor reputation is managed through changes in achieved by the rep) to that entity's internal characteristics the selling function. A poor reputation perceived to be (e.g., the rep's motivation or ability), as opposed to external caused by external factors might be due to external circum- sources (e.g., unfortunate extraneous circumstances). Tbus, stances or the actions of any one of several agents with if a manufacturer perceives tbe rep's reputation as poor, it is whom it has direct contact (e.g., suppliers, financial institu- likely to view tbe rep as soleiy and causally responsible for tions, distributors, its selling partners). Thus, altbough a disappointing results. The manufacturer may fail to consid- firm with a poor reputation indeed may engage in reputation er whether there might be mitigating circumstances (sucb as management activities, it is most likely to engage in actions an economic downtown) or wbetber the manufacturer itself directly related to those entities viewed as causally respon- is at fault (by missing delivery deadlines, for example). In sible. Only in some cases may the rep be viewed as the short, the manufacturer may blame reps witb poor reputa- causal agent. In tbose other situations, the most effective ac- tions for negative outcomes. Second, a rep with a poor rep- tion is not to change the selling organization structure, but utation also may motivate a manufacturer to act because the rather to engage in reputation management activities that are manufacturer may fear that this poor reputation may "in- most directly relevant to the offending party. fect" its own reputation. Second, even if the manufacturer does attribute its poor If the rep's poor reputation does motivate the manufac- reputation to its own efforts, it may be motivated not to act turer's desire to take reputation management actions, consis- and to stay with its existing rep, because staying with the tent with reputation management theory, we anticipate that status quo seems like a safe option. Taking on the selling the manufacturer's beliefs about the extent to which highly function internally (e.g., by vertically integrating) would reputable firms use their own sales force will affect how it seem to be a relatively ineffective method of enhancing a acts. If the manufacturer perceives the rep as poorly reputed manufacturer's reputation, because its reputation is already and it strongly believes tbat highly reputable firms use their poor. Furthermore, a manufacturer might not believe that own sales force, it is likely to vertically integrate. It sbould switching reps has the potential to enhance its reputation be- not do something different than what it believes reputable cause its own poor reputation might make it difficult for the firms do (i.e., change reps). Thus, we expect the following: manufacturer to attract a more reputable rep. H2a: The manufacturer's intentions to vertically integrate in- In contrast, we believe that a manufacturer is motivated crea.se with combined decreases in the manufacturer's to manage its reputation through its sales organization struc- perception of its rep's reputation and increases in the ture when its current reputation is highly favorable. First, be- manufacturer's belief that highly reputable firms use their cause successes tend lo be attributed to oneself, a manufac- own sales force. turer with a favorable reputation is likely to perceive that its V\2h- The manufacturer's intentions to switch to a new rep de- own actions are responsible for its reputational success. Be- crease with combined decreases in the manufacturer's perception of its rep's reputation and increases in the cause it has been responsible for garnering a positive reputa- manufacturer's belief that highly reputable firms use their tion, it sbould be equally capable of sustaining it in the future. own sales force. Second, firms that believe they are esteemed highly by cus- tomers may see themselves as actual or potential members of The manufacturer's reputation and beliefs as bases for tbe broad category of "bighly reputable" firms. Researcb in actions. We also might question whether the manufacturer's both individual (Tajfel et al. 1971; Taylor, Crocker, and perception of its own reputation and its beliefs affects its D'Agostino 1978) and organizational contexts (Dutton, Duk- reputation-management actions. The general applicability of erich, and Harquail 1994; Dutton and Penner 1993) suggests reputation theory to sales organization structure decisions social entities wish to perceive themselves as belonging to a would be quite powerful if manufacturers' actions also were group. Moreover, if a social entity views its actual or ideal affected by these reputation-related variables. Again, extant self-identity as consistent with a social category, it is moti- theory provides some interesting perspectives on this issue. vated to act to maintain or preserve its association with such Because firms are assumed to be active reputation mon- a group so as to differentiate itself from outgroup members itors and because they have an implicit desire to manage (i.e., nonreputable firms) (Kelley 1988; Taylor etal. 1978).

78 / Journal of Marketing, October 1999 If increases in the manufacturer's perception of its repu- sponse rate was estimated conservatively as 24%. Compari- tation do increase its desire to manage its reputation active- son of early and late respondents, in which later respondents ly, we anticipate that the manufacturer's beliefs about the are taken as representative of non respondents (Armstrong extent to which highly reputable firms use their own sales and Overton 1977), revealed no statistically significant dif- force will affect how it acts. Consistent with reputation man- ferences in the descriptive characteristics of these groups. agement theory, the manufacturer should act in a manner Reps are local entities, and manufacturers typically cov- consistent with its reputation-related beliefs. Thus, the more er a marketplace by patching together a network of reps, all the manufacturer perceives itself as highly reputed and the operating in different geographies or with different customer more strongly it believes that highly reputable firms use bases. The questionnaire asked respondents to choose one their own sales force, the more it is likely to vertically inte- current rep organization in the sales manager's district and grate. It should not do something different than what it be- respond with this rep and its customers in mind. Three ver- lieves reputable firms do (i.e., change reps). Thus, we expect sions of the survey were distributed in equal proportions. the following: Each asked managers to respond regarding one of their bet- ter, midrange, or poorer reps, respectively. This procedure H3a: The manufacturer's intentions to vertically integrate in- crease with combined increases in the manufacturer's was designed to create variation in the dependent variables. perception of its own reputation and its belief Ihat highly Returns came back in approximately equal proportions to reputable firms use their own sales force. the three distributed questionnaires; of the 258 returned Hii,: The manufacturer's intentions to switch to a new rep de- questionnaires, 90 f(K:used on a poorer-performing rep, 84 crease with combined increases in the manufacturer's on a midrange rep, and 84 on a best-performing rep. Over- perception of its own reputation and its belief that highly all, the threat of nonresponse bias appears minimal. reputable firms use their own sales force. The questionnaire was developed using extensive pretesting. The survey was administered personally to a dis- trict sales manager from each of five different manufactur- Methodology ers. The survey minimized halo effects by allowing indica- tors of the constructs to be separated by several other Context and Sampling Strategy questions. For example, seven pages of questions separated We gathered data from managers in the electronic compo- the scales measuring managers' perceived reputation of the nents industry. Several factors motivated the selection of rep and their intention to use a direct sales force. The pretest this industry. First, sales organization structure is regarded revealed that respondents could respond easily to the mea- as an important factor that determines marketplace success sures and the survey instrument. in this industry. The industry context therefore provides an opportunity to observe the potential impact of reputation- Measure Development based motives on sales organization structure decisions. Measures of all constructs were developed using guidelines Second, this industry is characterized by intense competi- recommended by Nunnally (1978). The domain of the rele- tion, a wide range of product offerings, and diverse suben- vant construct first was specified. Items subsequently were vironmcnts (creating variation on crucial variables). More- drafted on the basis of their mapping with the construct's over, researchers (e.g., Anderson 1985) have noted conceptual detinition. Most of the items are recorded on a considerable variation in the usage of rep and direct sales seven-point agree-disagree format. Items were pretested for forces in this industry. Third, because work on sales organi- clarity and appropriateness using a pilot sample of five dis- zation structure typically has focused on economic motiva- trict sales managers and were rewritten if necessary. Com- tions, using an industry for which economic motivations for bined, the three hypotheses reflect the two dependent vari- sales structure decisions are strong creates a relatively se- ables we describe next. vere test of the reputational motive. With the cooperation of two trade associations and a ma- Dependent Measures jor industry publication, a random sample of 1209 compa- nies known to use manufacturers' reps was drawn from two Intention to integrate vertically. The manufacturer's in- trade association mailing lists. Pretested surveys were sent tention to integrate the selling function flNTEGRATEJ was to the district sales manager in each firm who was in a posi- indicated by a six-item scale (Cronbach's alpha = .90). tion to indicate the firm's intention in relation to the rep. These items index the manufacturer's seriousness about dis- These managers serve as the manufacturer's liaison to the pensing with its rep and setting up a direct sales force in the rep and strongly influence the organization's view of the near future (two years or less). An example of an item from rep. They are responsible for determining the future of the this scale is, "We will convert this territory to a direct sales rep's relationship with the manufacturer (Gibbons 1988), force in the near future." and they are the critical decision maker regarding sales or- Intention to change reps. The manufacturer's intention ganization structure decisions. Because our respondents are to change to a different manufacturers' rep (CHANGEREP) knowledgeable and influential regarding the rep organiza- was assessed by a separate four-item scale (Cronbach's al- tion, they satisfy Campbell's (1955) criteria for key infor- pha = .90) tapping the manufacturer's seriousness about mants. Callbacks and a follow-up mailing resulted in 258 re- switching reps over the near term (two years or less). Ex- sponses. One hundred forty-three of the 1209 firms were tensive pretesting indicated this could be operationaiized as subsequently identified as "out of business." Hence, the re- two years or less. An example of an item for this scale is, "In

Reputation Management / 79 this district, we intend to do our sales through a different rep they believe firms that use their own sales force are regard- within the next two years." ed with confidence and are taken seriously (e.g., are rep- Managers' responses to the CHANGEREP and INTE- utable). An example of this item is, "Customers in this busi- GRATE scales were orthogonal (r = .002, p < .48). This em- ness have more confidence in firms which have their own pirical result is consistent with pretest respondents' com- sales force" (! = disagree; 7 = agree). The reliability of this ments that they did not necessarily view these as either-or four-item scale is adequate (Cronbach's alpha = .75). Be- decisions. The statistical independence of these two mea- cause there is no consensus in this industry about whether sures is also consisteni with the idea that consideration of using a direct sales force signals a firm's reputation, there is both options can proceed in parallel for some time. The sta- considerable variation in our data set on this belief variable. tistical independence of these two measures is important, because if these were either-or decisions, empirical support Control Variables for the second part of each of the three hypotheses (e.g., We included several variables known to affect organization- Hih) could be considered redundant with the first part (e.g.. al structure decisions in all statistical tests as control vari- Hi a). That the constructs are orthogonal suggests that statis- ables. We also included other variables thought potentially tical support or rejection for either part of the hypothesis is to affect intentions to vertically integrate or switch to a dif- possible. ferent rep as controls. We describe the specific control vari- ables and the rationale for their inclusion next. Independent Variables Transaction-specific assets. Transaction-specific invest- Perceptions of the manufacturer's reputation. The man- ments stabilize relationships by putting both parties at risk ufacturer's impression of how customers perceive its own for losses from potential opportunistic hehavior (Heide and reputation (OWNREPUTAT) was assessed by a seven-point John 1988; Weiss and Kurland 1997; Williamson 1985). semantic differential scale consisting of five pretested items This contrasts with de novo contexts, in which the anticipa- (e.g., "How do buyers in this sales territory view your com- tion of investing in specific assets can resuU in vertical inte- pany at this time?" [unstable; stable]). Operationally, firms gration (Anderson 1985, 1988; Williamson 1985). The exis- with more favorable reputations believe their customers per- tence of transaction-specific assets should reduce the ceive them as highly regarded, professional, successful, manufacturer's intention to vertically integrate or switch well-established, and stable. These items are not designed to reps. We examined two types of transaction-specific invest- assess whether the manufacturer has a reputation for any- ments: (I) the extent of the rep's transaction-specific invest- thing in particular. Nor is it evident what actions, attributes, ments in the manufacturer and (2) the extent of the rep's or signals the manufacturer may possess or employ to create transact ion-specific investments in the manufacturer's cus- the impression of having a favorable reputation. Thus, con- tomers. The former (MANTSA) was indicated by a seven- sistent with the unidimensional perspective on reputation item scale (e.g., "This rep has invested a lot of time and ef- (e.g., Yoon, Gulfey, and Kijewski 1993), this measure cap- fort to learn all the ins and outs of my organization"; tures generalized reputation perceptions. Reliability of this Cronbach's alpha = .88), whereas the latter (CUSTSA) was measure is reasonably high (Cronbach's alpha = .84). There- indexed with a four-item scale (e.g., "This rep builds a lot of fore, all items were summed to form a composite index. value into our product line by providing special services to buyers"; Cronbach's alpha = .82). All items were measured Perceptions of the rep's reputation. The manufacturer's using seven-point agreement scales. impression of how customers perceive its rep's reputation (REPsREPUTAT) was assessed using the same five items in Rep performance. Another reason for maintaining the response to a question about how customers view the man- current rep relationship is the rep's performance. Therefore, ufacturers' rep. Reliability of this scale is also high (Cron- we included the performance of the rep as a control variable bach's alpha = .86), and items were summed to form a com- that was based on the three categories of rep performance posite index. The extent to which a manufacturer perceives (poor, midrange, best) employed in the questionnaires. Us- its rep to have a poor reputation is measured with a reverse ing poor rep performance as a basis, we constructed two scale of REPsREPUTAT. We call this variable REPs- dummy variables. MIDPERF is I for a midrange-perform- DEFICIT. It is used in our statistical test of H2. ing rep and 0 otherwise, and BESTPERF is 1 for a best-per- forming rep and 0 otherwise. Perceptions of the reputation gap. The measure of the gap between the perception of the manufacturer's reputation Manufacturer size. Large manufacturers may have the and that of its rep (GAP) was assessed by subtracting REP- resources both to exit current rep relationships and to set up sREPUTAT from OWNREPUTAT. The coefficient alpha for a direct sales force. We employed a two-item index of the this difference score is .80, as calculated using the method manufacturer's size (SIZE) to account for this effect (r = of Peter, Churchill, and Brown (1993). .67). A sample item is, "Please circle the number which best describes how buyers in this sales territory view your orga- Beliefs about what reputable ftrms do. The manufactur- nization at this time" (seven-point Likert scale, anchored er's belief that highly reputable firms use an integrated sales small/large). force (BELIEF) was indicated by a set of four questions that asked managers to rate on a 1-7 disagree/agree scale the ex- £.r//co^/. A major consideration in a firm's ability to ver- tent to which they believe the most established and success- tically integrate is the cost of exiting the rep relationship and ful firms use their own sales force and the extent to which the cost of setting up a direct sales force. A six-item, seven-

BO / Journal of Marketing, October 1999 point agreement scale (EXITCOST) measures these costs with that of the original model. The various chi-square dif- (Cronbach's alpha = .84). A sample item is, "We face very ference tests are all significant and provide evidence of dis- high harriers lo switching over to a direct organization." criminant validity (Bagozzi, Yi, and Phillips 1991). Ability to find qualified reps. The manufacturer is more likely to switch to a different rep if the manufacturer per- Tests of the Hypotheses ceives that other qualified reps are available and are willing The results of the hypothesis tests are shown in Tables 2 and to represent the manufacturer. To account for this possihili- 3. Before testing each hypothesis, we estimated two baseline ty, a two-item index (OURPICK) of the ability of the man- models, each of which included only the relevant dependent ufacturer to find other qualified reps was employed (r= .61). variable (e.g., INTEGRATE, CHANGEREP) and the con- For example, one of (he items asked subjects to indicate trol variables. Subsequently, we added the reputation vari- their agreement with the statement "Our line is so desirable ables to each baseline model and reestimated the model. By we could have our pick of manufacturers' reps in this terri- comparing the baseline and full models using an F statistic, tory." Both items were measured using seven-point agree- we can examine the explanatory power of the reputation ment scales. variables. As shown at the bottom of Tables 2 and 3, the Satisfaction with the current rep. Manufacturers that are models with the reputation variables added significantly to satisfied with their current rep should be likely to remain the explanatory power of the baseline models in all cases. with that rep. Consequently, we included a five-item, seven- Therefore, we present only the estimation results of the point agreement scale (SATISFACTION) in all statistical models that include the reputation variables. models. A sample item from this scale is, "We are very dis- Hi predicts that the manufacturer's intention to vertical- satisfied with this manufacturers' rep" {reversed item). ly integrate (H|a) or switch reps (H|b) is determined jointly Cronbach's alpha for this scale is .92. by the perceived reputation gap between the manufacturer In summary, we included several control variables in our and its current rep and its belief that reputable firms use a di- statistical tests. These include variables identified in prior rect sales force. To test the two parts of this hypothesis, we literature as relevant to organizational structure decisions, estimated two multivariate models using intention to verti- several of which are economic in nature (i.e., transaction- cally integrate and intention to switch reps as the dependent specific assets, rep performance, size, exit costs, and ability variables. The independent variables included ihe interac- to find qualified reps). The correlation matrix for the scales tion between the gap in reputation perceptions and the belief is presented in Table I. variable (GAP x BELIEF) and the control variables. In ad- dition, we included two other sets of variables. First, be- cause the gap measure is a difference score, spurious corre- Results lations may result from the correlation of the component measures (REPsREPUTAT and OWNREPUTAT) with the Measure Validation Procedures gap measure. We add the two component measures in this Prior to testing the hypotheses, the multi-item measures statistical test to control for this possibility. Second, al- were subjected to a series of validity checks. First, we con- though there is no theoretical rationale for proposing direct ducted item-to-iteni total correlations and exploratory factor effects of these reputation-related variables, we included the analysis. Each item appeared to belong to the appropriate variables composing the interaction term to avoid possible domain, and none exhibited significant cross-loadings. specification error. Therefore, no items were deleted from further analysis. We The test of Hla corresponds to the interaction term (GAP then subjected the entire set of items to a confirmatory fac- X BELIEF) having a significant and positive effect on the tor analysis. Following Gerbing and Anderson (1988), we manufacturer's intention to integrate the selling function. estimated a measurement model in which every item was re- Noting the results in column 1 of Table 2, we see that the co- stricted to load on its a priori specified factor, and the fac- efficient for the interaction between GAP and BELIEF is tors themselves were permitted to correlate. positive and significant (.222, t = 4.96, p < .01), thereby sup- We obtained maximum likelihood estimates of the mea- porting Hia. Specifically, as both the gap in reputation per- surement mode! using EQS. The overall chi-square statistic ceptions and the belief that highly reputable firms use their for the model was significant (X^(I2I9) = 2226.54, p < own in-house sales forces increase, manufacturers' inten- .001), as might be expected given the size of our sample tions to engage in vertical integration also increase. Of the (Bagozzi and Yi 1988). However, the comparative fit index control variables, only the exit cost variable is significantly (CFI - .89; Bentler 1990) and the root mean square resid- related to the intention to vertically integrate (-.261, t = ual (RMSR = .066) represent evidence of good model fit. -5.21,/7<. 01). Each factor loading is significant at the .01 level. The glob- The test of H|b corresponds to the interaction terms al measurement model thus provides satisfactory evidence (GAPx BELIEF) having a significant and negative effect on of the unldimensionality of the measures. The coettlcient the manufacturer's intentions to change reps. Column 2 of alpha for each item set aiso provides satisfactory evidence Table 2 shows that the coefficients of the interaction term of reliability. (GAP X BELIEF) is significant and negative (-.183, t = Next, we estimated a set of additional models in which -3.87, p < .01), thus supporting Hj^,. Thus, manufacturer's the individual factor correlations were restricted to unity one intentions to switch to a new rep decrease with combined in- at a time. We compared the fit of these restricted models creases in the positive gap between the manufacturer's per-

Reputation Management / 81 O) CO

CO eg CO o o CM O o r ai

O3 CO CM €0 O CM O ^ CM O O

CO CO eg -^ ay q CVJ q LO CO q q O) •-- I l' r -* T^ T^ r^

CO eg q o .1 0 .1 3 7. 0 00 -.6 4 4.4 5

CO CO CD CO en CO in r- O en co o CM

CM o CO O '- CO eM q m q eg o •-- r r r r cvi CD

o cy CD in 1- CD a> CO r^ CM Cvl eg 19 9 CO 1- CD

en 00 CO •.- eg co q CO •>- CM q q Hi -^ m (0 c o CNJ 1- cy o •>- (D .0 O •<- CO CO CO q in q in o o iri '-^ T-^ i--^ 0)

o eg CD CM CM to r- CO o q o CM eg o to in o o [ q q r r I r

CO Ol CO in in o o CO q in q 1- CO O O r r r

O) CJ) CD O 2 c c c Is '^ 2 "3= o 0 | 0 « U) TJ TJ (0 n in 3 0 C TJ a C 0 o a; "5 0 5 i^ CT cn o t 5 (0 (fl (0 ••- *' 5 o a. (0 w O 0) i8r„~ 0 t- 0 C7LU c "5 ~ ^-^ o •SuJ .2 Q- - 0 UJ O Q_ 52- CO .^ -9 ro < 0*- -0 )' CO ii: -s a. Q) C ffl LU •-O 5 fflO « yi« LUUJ ."_' LU -^ p -ll 2 0 -> 0 Ol ]£ (O w ^'^-- 1-5 *- 0 -C. .Si _ ?^ 3 .fcj *-' c 0 0 Q. 0 A M il 3 ^ 3 Qu-- _33 3^ o c 0 S" — to CO ^^ ^ 3 D ^ m O CD D- 0 — ^' -^ XJ 3. CO Ii •o CO iS ro CO Q- CO < CO Q.»= "I "Jo .Cfl o B 2? .•^^ c o «!=

82 / Journai of Marketing, October 1999 TABLE 2 Un standardized Estimation Resuits for (t-statistjcs in parentheses) Hypothesis

H1b Dependent Variables INTEGRATE CHANGEREP Manufacturer's Manufacturer's intention to intention to integrate the change to a selling function different rep Independent Variables

CONSTANT 1.48 (1.84) -1.84 (-.40)

GAP Difference between manufacturer's perception of its own reputation and rts rep's reputation .492 (.773) .172 (.26)

BELIEF Manufacturer's beiief that highly reputable firms use an integrated sales force .441 (7.80)" -.147 (-2.52)*

GAP X BELIEF .222 (4.96)" -.183 (-3.87)* MANTSA Extent of the manufacturer's transaction-specific investments in the rep -.017 (-.22) -.021 (-.26) CUSTSA Rep's specific investments in the manufacturer's customers .011 (.17) -.041 (-.57) EXITCOST Costs of exiting the rep relationship and setting up a direct sales force costs -.261 (-5.21)" OURPICK Ability of the manufacturer to find and attract other qualified reps .034 (.59)

SIZE Manufacturer's size .057 (1.02) -.056 (-.95)

SATISFACTION Manufacturer's satisfaction with its rep -.133 (1.71) -.714 (8.66)"

OWNREPUTAT Manufacturer's perception of its own reputation -.114 (-1.85) -.441 (-.58)

REPsREPUTAT Manufacturer's perception ot its manufacturers' rep's reputation 1.14 (1.83) .427 (.659)

MIDPERF -.089 (-.51) -.151 (-.80)

BESTPERF -.133 (.66) -.038 (-.18)

Adjusted R2 .42 .53

F 16.60" 24.94"

F statistic for baseline model comparison 35.4" 8.6" •p < .05 (2-tailed test). "p< ,01.

Reputation Management / 83 ception of its own reputation and its rep and its belief that The process of testing tbe remaining hypotheses is iden- highly reputable firms use their own sales force. Of the con- tical to tbe method described previously. In particular, to test trol variables, only the manufacturer's satisfaction with its H2 and H3, we estimated another set of multivariate models. current rep is significant and negatively related to the likeli- The results are shown in Table 3. The dependent variables hood of changing reps (-.714, t = -8.66,/j< .01). include tbe intentions to vertically integrate and to switcb

TABLE 3 Unstandardized Estimation Resuits for H2 and H3 (t-statisties in parentheses) Hypothesis

H,,2a'3/ a '3b Dependent Variables INTEGRATE CHANGEREP Manufacturer's Manufacturer's intention to intention to integrate the change to a selling function different rep

Independent Variables

CONSTANT 5.93 (4.36)" -3.14 (-2.27)'

OWNREPUTAT Manufacturer's perception of its ovi/n rsputafion -.647 (-3.49)** .633 (3.25)"

REPsDEFICIT Manufacturer's perception of the extent that its rep has a poor reputation -.663 (-3.82)" .584 (3.21 )**

BELIEF Manufacturer's belief that highly reputable firms use an integrated sales force -1.09 (-3.02)** 1.16 (3.09)**

OWNREPUTAT X BELIEF 217 (3.75)" -.190 (-3.15)"

REPsDEFICIT X BELIEF .227 (4.29)* -.178 (-3.20)** MANTSA Extent of the manufacturer's transaction-specific investments in the rep -.019 (-.25) -.022 (.27) CUSTSA .020 (.28) Rep's specific investments in the manufacturer's customers -.042 (-.57) EXITCOST Costs of exiting the rep relationship and setting up a direct sales force costs -.249 (-4.97)"

OURPICK Ability of the manufacturer to find and attract other qualified reps .035 (.60)

SIZE Manufacturer's size .054 (.96) -.053 (-.91)

SATISFACTION Manufacturer's satisfaction with its rep -13 (164) -.716 (8.67)*

MIDPERF -.122 (-.69) -.139 (-.75)

BESTPERF -.148 (-.74) -.032 (-.15)

Adjusted R2 .41 .53

16.08" 24.88**

F statistic for baseline model comparison 20.2" 4.96** *p < .05 (2-tailed test), **p<.01.

84 / Journal of Marketing, October 1999 reps. The independent variables are the interaction among FIGURE 1 the reputation perception variables and beliefs (OWNREP- Manufacturers' Intentions Across the Range of UTAT X BELIEF and REPsDEFICIT x BELIEF), the indi- Beliefs vidual components of the interaction terms (OWNREPU- TAT REPsDFFICIT, BELIEF), and the control variables. A: The Effects of Beliefs (BELIEF) on the Relationship The test of H2a corresponds to the interaction term Between the Rep's Reputation and Intentions to (REPsDEFICIT x BELIEF) having a significant and posi- integrate (INTEGRATE) (from Table 3) tive effect on the manufacturer's intentions to integrate the 6 INTEGRATE/ selling function. The results shown in column I of Table 3 5 REPsDEFICIT support this hypothesis. Specifically, the interaction be- tween REPsDEFICIT and BELIEF is significant and in the hypothesized positive direction (.227, t = 4.29, p< .01). The test of Hib corresponds to the interaction term (REPs- DEFICIT X BELIEF) having a significant and negative ef- •• BELIEF fect on the manufacturer's intentions to change reps. As shown in column 2 of Table 3, tbe interaction is indeed sig- = -.663 + -227 (BELIEF! nificant and negative (-. 178, t =: -3.20, p<.0\). Thus, as the 6 REPsDEFICIT manufacturer's perception of its rep's reputation decreases and its belief tbat bighly reputable firms use an integrated sales force increases, the likelihood of vertically integrating the selling function increases and the likelihood of changing reps decreases. B:The Effects of Beliefs (BELIEF) on the Relationship Between the Rep s Poor Reputation (REPsDEFICIT) The test of H^^ corresponds to the interaction term and Intentions to Change Reps (CHANGEREP) (OWNREPUTAT x BELIEF) having a significant and posi- (from Table 3) tive effect on the manufacturer's intentions to vertically in- tegrate, whereas the test of H31, corresponds to the interac- 8CHANGEREP/ tion term having a significant and negative effect on the 5 REPsDEFICIT

manufacturer's intentions to switch reps. Support for H^^ is 6CHANGEREP = .5K4 -.I7tt (BELIEF) found in column 1 of Table 3. The coefficient for the inter- SREPsDEFICIT action between OWNREPUTAT and BELIEF is positive and significantly related to the manufacturer's intention to vertically integrate (.217, t = 3.75, p < .01). Column 2 of Table 3 shows that the estimate for the same interaction is significant and negatively related to the intention to change reps(-.l90, t = -3.l5,/j< .05). Of the control variables, only the exit cost variable is significantly related to the intention to vertically integrate (-.249, t = ^.97, p < .01). Only the manufacturer's satis- faction with its rep is related significantly to its intention to high levels. Therefore, although these direct effects are in- change reps (-.716. t = -8.67, p< .0\). teresting, they can be understood only within the context of Although the data support each of the hypotheses, the re- the interaction. sults in Table 3 suggest that the reputation-related variables Using the same method, a similar set of results can be exert a direct effect on tbe dependent variables. Because of obtained to demonstrate that the direct effect of OWNREP- the presence of tbe interactions, bowever, tbese direct ef- UTAT on both intentions to vertically integrate and to fects can be interpreted only within the context of the inter- change reps is not monotonic over the entire range of BE- action (Friedrich 1982). As an example, the direct effect of LIEF For low levels of BELIEF «.647/.217 = 2.98), an in- REPsDEFICIT on the manufacturer's intentions to vertical- crease in OWNREPUTAT actually has a negative effect on ly integrate is not monotonic over the entire range of BE- the likelihood of going direct, whereas this effect is positive LIEF. This can be seen by taking tbe partial derivative of the for relatively high levels. However, for low levels of BE- regression equation with respect to REPsDEFICIT and plot- LIEF {<.633/.19O = 3.33), an increase in OWNREPUTAT ting the resulting equation (Schoonhoven 1981). In Figure I, actually has a positive effect on the likelihood of switching Part A, we show that for low levels of BELIEF reps, whereas this effect is negative for relatively high lev- (<.663/.227 = 2.92), an increase in REPsDEFICIT actually els. We discuss these direct effects more fully in the "Dis- has a negative effect on the likelihood of vertical integra- cussion" section. tion, whereas this effect is positive for relatively high levels. In Figure I, Part B, we show the direct effect of REPs- Discussion DEFICIT on intentions to switch reps. Here, for low levels of BELIEF (<.584/.I78 = 3.26), an increa.se in REPs- Summary DEEiCIT actually has a positive effect on the likelihood of The purpose of this research was to determine whether rep- switching reps, whereas this effect is negative for relatively utation concerns influence how manufacturers that current-

Reputation Management / 85 ly use an independent sales force intend to structure their eral, as we observed support for the proposed interaction for sales organization. Relying on reputation theory, we suggest tbe manufacturer's reputation perception variable, the rep's that manufacturers both monitor and attempt to manage reputation perception, and the gap between the two. Fur- their own reputation and that decisions regarding the struc- thermore, support for the GAP x BELIEF interactions also ture of the sales organization are driven by these reputation- appears quite strong, as it was observed even when we con- related processes. Specifically, we hypothesize that manu- trolled for the individual variables tbat constituted tbe gap facturers that currently use an independent sales force measure. develop perceptions of their own reputation, the reputation Second, it is notable tbat the proposed interactions proved of their rep, and the reputation gap between the two and that significant even when we included a host of additional vari- these perceptions work jointly with their beliefs about what ables that might explain intentions to engage in vertical inte- reputable firms do to influence their sales organization gration or switch reps. Some of these variables are economic structure intentions. in nature and bave been posited previously as motivators of We hypothesize and find that as the gap becomes more sales organization structure decisions (e.g., perceived exit positive, such that the manufacturer's reputation exceeds costs, transaction-specific investments). Others are more spe- that of the rep, and the manufacturer's beliefs that highly cific to the performance of the rep (e.g., satisfaction with the reputable fttTns use their own sales force increase, the man- rep, rep performance), and still others reflect other external ufacturer's intention to change its sales organization struc- constraints affecting these decisions (e.g., ability to find qual- ture by vertically integrating the sales force also increases. ified reps). It is also notable that the models that included the This action reflects a consistency between manufacturers' reputation variables represented a significantly better flt to reputation perceptions and their beliefs. As predicted by the data than the models that excluded them. reputation management theory, manufacturers are unlikely Third, although the results indicate how perceptions and to switch reps, because that would be inconsistent with their beliefs jointly affect reputation management actions, the di- beliefs about what highly reputable firms do. rect effects provide additional insight into how perceptions Decomposing tbe gap into its constituent elements, we and beliefs individually affect such decisions. For example, propose and observe that perceptions of the reputation of the as shown in Figure I, the effect of increases in the manu- rep and perceptions of the reputation of the manufacturer al- facturer's perceptions of the rep's reputation on going direct so motivate reputation management actions. Interestingly, the depends on beliefs about what reputable firms do. Wben effects of these perceptions, along with their beliefs, are manufacturers strongly believe that highly reputable firms asymmetric, because manufacturers consider tbeir own repu- use their own sales force, intentions to vertically integrate tations differently tban that of their rep. As the perception of become higber because the rep's reputation is perceived to the rep's reputation becomes poorer, the manufacturer is like- be poorer. This is consistent with H2 (Figure I, Part A). ly to be increasingly motivated to manage this reputation However, when manufacturers do not strongly believe that problem, because the rep's poor reputation may affect the highly reputable firms use their own sales force, intentions reputation of the manufacturer negatively. The way in which to vertically integrate become lower as the rep's reputation it manages the rep's reputation problem depends on its beliefs is perceived to be poorer. What would the manufacturer do about wbat bighly reputable firms do. As the manufacturer's in tbis case? Figure I, Part B, resolves the issue, showing beliefs that reputable firms use their own sales force increase, that when beliefs are weak, the more the rep's reputation is it is likely to solve this reputation problem by vertically inte- perceived to be poor, the more manufacturers intend to grating tbe selling function. In contrast, it is unlikely to switch reps. Thus, although manufacturers are motivated to cboose an alternative course of action such as switching reps. act when their rep's reputation is poor, they do so in a man- When it comes to the manufacturer's perception of its ner that is not inconsistent with their beliefs. This same re- own reputation, the effects are reversed. Those who count sult pertains to the situation in which the manufacturer's themselves highly reputable appear ready to imitate what own reputation is the focus. Thus, the direct effect results re- they believe is the choice of" highly reputable firms. Thus, as inforce the argument that consistency between reputation the manufacturer's perception of its own reputation be- perceptions and beliefs guide action. comes more positive, it appears to be increasingly motivat- Fourth, it is notable tbat tbe reputation variables we ex- ed to engage in reputation management activities, and it ap- amined bere appear to affect manufacturers' intentions to en- pears to cboose tbose actions tbat are consistent witb its gage in a fairiy drastic sales organization structure change— beliefs. Accordingly, manufacturers with more favorable specifically the decision to vertically integrate the sales force. reputations bave greater intentions to vertically integrate the Although it might appear that manufacturers should solve a selling function and lower intentions to switch reps as they reputation problem with a rep by simply switching reps, that increase their belief that highly reputable firms use their they choose a more costly and intensive course of action own sales forces. This phenomenon, in which firms come to when they believe that highly reputable firms use their own resemble one anotber by a process of imitation, has been la- sales force is suggestive of the power of reputation-related beled "mimetic isomorphism" in the organization theory lit- concerns on sales organization structure decisions. erature (Galaskiewicz and Wasserman 1989). Several additional aspects of the results also deserve Limitations highlighting. First, the notion that reputation perceptions af- This study, like most, is subject to limitations. The principal fect specific reputation management actions when they are shortcoming is the use of intentions as opposed to bebavior consistent with reputation-related beliefs appears quite gen- as the object of study. Unfortunately, studying actual bebav-

66 / Journal of Marketing, October 1999 ior poses practical constraints. These include the fast rate of related variables affect intentions to terminate the relation- personnel turnover in industries and the difficulty of discov- ship with the current rep appears to be important. This ering the correct point in time at which to observe actual be- would require identifying a belief variable that directly havior. Intentions do not correspond perfectly to eventual maps onto beliefs about termination. One belief of interest is behavior; however, they tend to become self- fulfilling. For the manufacturers' belief that they can "climb the ladder of example, a manufacturer that secretly intends to terminate reps" in the sales district. This terminology was suggested its rep relationship may fail to support the rep, thereby dam- by a manufacturers' rep during preliminary interviews. Al- aging the rep's performance and increasing the manufactur- though this idea is speculative, manufacturers that strongly er's incentive to terminate the relationship. A second limita- believe that it is possible to climb the ladder of reps may in- tion is our use of cross-sectional data, which limits our tend to terminate a relationship with a rep when they per- ability to determine empirically the causal relations implied ceive that their own reputation exceeds that of their rep. in our hypotheses. Third, our empirical study is conducted They, in turn, may choose a rep whose reputation better within a single industry (for the reasons discussed previous- matches their own. Further, enhancing their own reputation ly), which limits the generalizability of our results. There- may give rise to a reputation gap between the manufacturer fore, additional study of these reputation-related variables in and the new rep, which subsequently motivates the search the context of other industries is warranted. Another limita- for a rep with an even more favorable reputation. Thus, tion involves our use of key informant data. Although care more reputable sales agencies may be paired with even more was taken to ensure that our key informants were chosen in highly reputable, and hence highly demanding, manufactur- accord with Campbell's (1955) criteria, a multiple informant ers, which may terminate them if they are perceived to have method might be employed in the future to reduce any in- not kept pace with the manufacturer's reputation. formant biases that may arise in this research context. If a better understanding of reputation factors affecting termination decisions is derived, manufacturers' reps may Further Research benefit by monitoring their manufacturer's reputation per- The findings here, if replicated when the previously men- ceptions, their own reputation perceptions, and relevant tioned methodological problems are resolved, are sugges- manufacturer's beliefs. A manufacturer that perceives its tive of a host of additional research to extend our findings. own reputation as good (or the rep's reputation as relatively First, the concept of a reputation gap appears to be a signif- poor) and believes reputable firms use a vertically integrat- icant factor that affects intentions to engage in vertical inte- ed sales force may be expected to be recruiting personnel gration and therefore may be a basic factor affecting in- (often in a clandestine fashion) and looking for office space terorganizational relationship continuation. Both the gap in the sales district. When the manufacturer does not believe concept and its implications for relationship continuation, reputable firms engage in vertical integration, it may be ex- though exploratory in the present research, are extensions to pected to hold discrete conversations with other reps in the reputation theory. Research that further examines the repu- same district. (The latter activity is often easier to observe, tation gap concept seems warranted. because it is a ready basis of gossip at trade association Second, because reputation perceptions are assumed to meetings.) influence belief-consistent reputation management actions, Fifth, given the important role of reputation perceptions it is useful to examine the interaction of reputation percep- in our study, additional research that examines how the man- tions with beliefs other than the one examined here. For ex- ufacturer forms reputation perceptions appears warranted. ample, a natural implication of the theory is that when a What data do they use and what biases enter the collection manufacturer perceives the rep as having a poor reputation, and processing of information relevant to reputation assess- it likely may switch reps when it believes that highly rep- ment? How veridical are assessments? How malleable are utable firms use reps. Although this research did not exam- they to, for example, political processes within the manu- ine explicitly this belief variable, the results observed here facturer? And how can the rep influence—or indeed man- are not inconsistent with this idea. As Figure I indicates, for age—the process by which manufacturers assess reputations example, when the manufacturer does not believe that high- as viewed by customers? Because the rep is the primary link ly reputable firms use their own sales force and when it per- between customers and manufacturers, reps should be able ceives its rep's reputation as poor, it is likely to manage its to manage the manufacturer's impression to some extent. affiliation with a reputation-deficient rep by switching reps. These findings suggest it is to the rep's advantage when the Research in the future should extend the present study to in- manufacturer believes customers think highly of the rep— vestigate the interaction of reputation perceptions with be- and less highly of the manufacturer. liefs about reps. This research is a first and exploratory step in the study Third, because beliefs seem to be important factors that of the effect of reputation on sales force organization. The guide the nature of reputation management actions, addi- results suggest that reputation may have an important im- tional research that catalogs the types of reputation-related pact beyond (rather than in place of) more traditional expla- beliefs would be useful. We focused on descriptive beliefs, nations. If so, reputation can be expected to influence other so additional study that incorporates causal beliefs would be strategic marketing decisions. Traditionally, customer es- useful. teem and regard has been studied in terms of its impact on Fourth, whereas our study focused on the manufactur- marketing outcomes, such as sales and margins. These re- er's intentions to vertically integrate or switch reps, a more sults suggest that reputation significantly affects not only basic set of issues regarding the extent to which reputation- marketing outcomes but also marketing strategy. That repu-

Reputation Management / 87 tation plays a role in marketing strategy and outcomes rais- a daily basis, further research should explore the various di- es an even broader question that deserves research attention. mensions of reputation management and identify the behav- Namely, in what ways do companies manage their reputa- iors of firms that are useful in maintaining or increasing tions? Because companies must manage their reputation on their reputation.

REFERENCES Anderson. Erin (1985), "The Salesperson as Outside Agent or Em- and Mark Shanley (1990), "What's in a Name? Reputation ployee: A Transaction Cost Analysis," Marketing Science, 4 Building and Corporate Strategy." Academy of Management (Summer), 234-54. Journal, 33 (June), 233-59. (1988), "Strategic Implications of Darwinian Economics Friedrich, Robert J. (1982). "In Defense of Multiplicative Terms in for Selling Efficiency and Choice of Integrated or Independent Multiple Regression Equations." American Journal of Political Sales Forces," Management Science. 34 (May), 599-618. Science, 26 (4). 797-833. and Anne T. Coughlan (1987), "International Market Entry Galaskiewicz, Joseph and Stanley Wasserman (1989). "Mimetic and Expansion Via Independent or Integrated Channels of Dis- Processes Within an Organizational Field: An Empirical Test," tribution," yourna/ of Market itig, 51 (January), 71-82. Administrative Science Quarterly, 34 (September), 454-79. Armstrong, J. Scott and Terry S. Overton (1977). "Estimating Non- Gerbing, David W. and James C. Anderson (1988), "An Updated response Bias in Mail Surveys." Journal of Marketing Re- Paradigm for Scale Development Incorporating Unidimension- search, 14 (August). 396-403. ality and Its Assessment." Journal of Marketing Research, 25 Bagozzi. Richard P. and Youjae Yi (1988), "On the Evaluation of (May). 186-92. Structural Equation Mtxtels," Journal of the Academy of Mar- Gibbons. James (1988), How to Start and Build a Successful Man- keting Science, 16(1), 74-94. ufacturers' Agency. Englewood Cliffs. NJ: Prentice Hall. , , and Lynn W. Phillips (1991), "Assessing Con- Heide, J. B. (1994), "Interorganizational Governance in Marketing strucl Validity in Organizational Research," Administrative Sci- Channels," Joumal of Marketing, 58 (January), 71-85. ence Quarterty. 36 (September), 421-58. and George John (1988), "The Role of Dependence Bal- Bentler, Peter M. (1990), "Comparative Fit Indexes in Structural ancing in Safeguarding Transaction-Specific Assets in Conven- Models," Psychological Bulletin. 107 (2), 238-46. tional Channels." Journal of Marketing. 52 (January), 20-35. Bergen, Mark, Shantanu Dutta, and Orville C. Walker Jr. (1992), Kahneman, Daniel and Amos Tversky (1979), "Prospect Theory: "Agency Relationships in Marketing: A Review of the Implica- An Analysis of Decision Under Risk," Econometrica, 47 tions of Agency and Related Theories," Journal of Marketing, (March), 263-91. 56 (July), 1-24. Kelley, C. (1988), "Intergroup Differenliation in a Political Con- Bromley, D. B. (1993), Reputation. Image, and Impression Man- text," British Journal of Social Psychology, 27 (4), 319-32. agement. West Sussex, England: John Wiley & Sons. Klein, Saul, Gary L. Frazier, and Victor J. Roth (1990), "A Trans- Brown. Steven P. (1995), "The Moderating Effects of Insuppli- action Cost Analysis Model of Channel Integration in Interna- er/Outsupplier Status on Organizational Buyer Attitudes." Jour- tional Markets," Journal of Marketing Research, 27 (May), nat of the Academy of Marketing Science., 23 (Summer). 196-208. 170-82. Lavin, Hank (1991), How to Get—and Keep!—a Profitable Rep Campbell. D. T (1955), "The Informant in Quantitative Research," Agencv b\ Effective Marketing. Cheshire, CT: Lavin Associates. American Journal of Sociology. 60 (January), 339^2. Lebell. Frank (1971). The Manufacturers' Representative. Boston: Corey, E. Raymond. Frank V. Cespedes, and V. Kasturi Rangan Herman Publishing. (1989). Going to Market: Distrihution Systems for Industrial (1977) , Independent Marketing/Selling. San Mateo, CA: Products. Boston: Harvard Business School Press. Hills-Bay Press. Dranove, David and Mark Shanley (1995), "Cost Reductions or Leuthesser, Lance (1988), "Defining. Measuring and Managing Reputation Enhancement as Motives lor Mergers: The Logic of Brand Equity." Conference Summary Report No. 88-104. Cam- Mulithospitai Systems," Strategic Management Journal, 16 bridge, MA: Marketing Science Institute. (January), 55-75. Merriam Webster's Collegiate Dictionary (1996). lOt*" ed. Spring- Dutton. Jane E., Janet M, Dukerich, and Celia V. Harquail (1994), field. MA: Merriam Webster, Inc. "Organizational Images and Member Identification,"/Wmm/i- Myers, D.G. and P J. Bach (1976), "Group Discussion Effects on trative Science Quarterly, 39 (June). 239-63. Confiict Behavior and Self-Justification," Psychological Re- and Wendy J. Penner (1993). "The Importance of Organi- port. 38(1), 135^0. zational Identity for Strategic Agenda Building," in Strategic- Nisbett. Richard and Lee Ross (1980). Human Inference: Strate- Thinking: Leadershtp and the Management of CItange, J. gies and Shortcomings of Social Judgment. Englewood Cliffs, Hendry and G. Johnson, eds. New York: Strategic Management NJ: Prentice Hall. Society, John Wiley & Sons, 89-113. Nunnally. Jum C. (1978), Psychometric Theory, 2d ed. New York: Elsbach. Kimberly D. and Roderick M. Kramer (1996), "Mem- McGraw-Hill. bers' Responses to Organizational Identity Threats: Encounter- Owen, Stewart (1993), "The Landor Image Power Survey: A Glob- ing and Countering the Business Week Rankings." Administra- al Assessment of Brand Strength." in Brand Equity and Adver- tive Science Quarterty, 41 (3), 442-77. tising. David A. Aaker and Alexander L. Biel, eds. Hillsdale, Emler, N. (1990), "A Social Psychology of Reputation," in Euro- NJ: Lawrence Eribaum and Associates. 11-30. pean Review of Sociology, Vol. 13, W. Stroebe and M. Hew- Park, C. Whan. Bernard J. Jaworski, and Deborah J. Maclnnis stone. eds. New York: Academic Press, 173-237. (1986). "Strategic Brand Concept Image Management." Jour- Festinger. Leon (1954), "A Theory of Social Comparison Process- nal of Marketing, 50 (October). 135^5. es." Human Relations. 7 (2), 117^0. Peter. J. Paul. Gilbert A. Churchill, and Tom J. Brown (1993), Fishbein, Martin and leek Ajzen (1975), Belief. Attitude, Intention "Caution in the Use of Difference Scores in Consumer Re- and Beliavior. Reading. MA: Addison-Wesley Publishing. search." Journal of Consumer Research, 19 (4). 655-62. Fombrun, Charles (1996), Reputation: Realizing Value from the Powers, Thomas L. (1987), "Switching from Reps to Direct Sales- Corporate Image. Cambridge, MA: Harvard Business School people," Industrial Marketing Managemenl, 16 (August), Press. 169-72.

88 / Journal of Marketing, October 1999 Ross, L. (1977), "The Intuitive Psychologist and His Shortcom- . S. T. Fiske, N. Etcotf, and A. Ruderman (1978). "Catego- ings: Distortions in the Attribution Process." in Advances in Ex- rization Bases of Person Memory and Stereotyping," Journal of perimental Social Psychology, Vol. 10. L. Berkowiiz. ed. New Personality and Social Psychology. 36 (7), 778-93. York: Academic Press, 174-221. Theus. Kalhryn T. (1993) "Academic Reputations: The Process of Rousseau. Denise M. (1985), "Issues of Level in Organization Re- Formation and Decay." Public Relations Review. 19 (3), 277-92. search: Multi-Level and Cross-Level Perspectives." in Re- Traynor, K. (1983), "Accountant Advertising," Journal of Adver- search in Organizational Behavior. Vol. 7, Lawrence L. Cum- tising Research, 23 (December), 35-40. mings and Barry M. Staw, eds. Greenwich, CT: JAI Press. Tversky, Amos and Daniel Kahneman (1981), "The Framing of 1-37. Decisions and the Psychology of Choice," Science, 211 (Janu- Schoonhoven, C. B. (1981). "Problems with Contingency Theory: ary), 453-58. Testing Assumptions Hidden Within the Language of Contin- and (1991), "Loss Aversion in Riskless Choice: A gency Theory." Administrative Science Quarterly, 26 (Septem- Reference-Dependence Model." Qttarterly Journal of Econom- ber). 349-77. ics, 106(4), 1039-62. Shrivastavas, Paul and George Siomkos (1989). "Disaster Con- Weiss. Allen M. and Erin Anderson (1992), "Converting from In- tainment Strategies," Journal of Business Strategy, 10 (Septem- dependent to Employee Sales Forces: The Role of Perceived ber-October). 26-31. Switching Costs," Journal of Marketing Research. 29 (Febru- Sweeney, Paul D., Dean B. McFarlin, and Edward J. lnterri- ary). 101-15. dieden (1990). "Using Relative Deprivation Theory lo Ex- and Nancy Kurland (1997). "Holding Distribution Channel plain Satisfaction with Income and Pay Level: A Multistudy Relationships Together: The Role of Transaction-Specific As- Examination." Academy of Management Journal, 33 (June), sets and Length of Prior Relationship," Organization Science. 8 423-36. (November-December). 612-23. Tajfel, H., M. Billig, R. P. Bundy. and C. Flament (1971), "Social Williamson. Oliver E. (1985), The Economic Institutions of Capi- Categorization and Intergroup Behavior," European Journal of talism. New York: The Free Press. Social Psychology, I (2), 149-77. Yoon, Eusang, Hugh Guffey, and Valerie Kijewski (1993), "The Taylor, Shelly E., Jennifer Crocker, and J. D'Agostino (1978), Effects of Information and Company Reputation on Intentions "Schematic Bases of Social Problem-Solving," Personality and to Buy a Business Service," Journal of Business Research. 27 Social Psychology Bulletin, 4 (3), 447-51. (3), 215-28.

Mailing Lists

Business Lists: Over Two Million Customers Can't Be Wrongl Over 11 Million Businesses For FREE Catalog & Personalized Service Call: Select by Type of Business, Employee Size & Sales Volume for any ZIP Code, County, City or State. (800) 555-5335 Consumer Lists: or Fax Us at: (402)331-1505 Over 195 Million Consumers Plus, FREE Directory Assistance & Select by Age, Income, Household Value & More. National Yellow Pages on Our Web Site. Business Credit Reports www.infoUSA.com only *5.00 each! m/r.USA-,1 Public Company, Nasdai/: IVSAA & tUSAB 5711 S. 86th Circle • Omaha, NE 68127 Also Available, E-Mail: [email protected] Monniy Updates! Media Code: 01-JMK

Reputation Management / 89