Agency Relationships in Marketing: A Review of the ... · PDF file10.10.2016 · Mark Bergen, Shantanu Dutta, & Orville C. Walker, Jr. Agency Relationships in Marketing: A...

Mark Bergen, Shantanu Dutta, & Orville C. Walker, Jr.

Agency Relationships inMarketing: A Review of the

Implications and Applications ofAgency and Related Theories

Agency and related theories have proven useful as theoretical frameworks for examining relationshipsbetween principals and their agents in many disciplines. However, though marketing involves a widearray of agency relationships, agency-based research has only recently begun to be reported in the mar-keting literature. The authors therefore attempt to clarify the marketing implications of agency theory bydescribing its major constructs, the two different types of models embedded within the theory, and someclosely related theoretical structures. They also critically review past and potential agency-based researchin marketing and suggest ways in which marketers might contribute to improving its validity and use-fulness.

AGENCY theory has spawned a large amount ofrecent research in economics, finance, account-

ing, organizational behavior, political science, andsociology. Its proponents prophesy that a revolutionis at hand, that agency and related theories can greatlyimprove our understanding of why organizations existand how they work (Hesterly, Liebeskind, and Zenger1990). Some scholars, however, are troubled by thesetheories' underlying assumptions about human behav-ior and organizational processes (Donaldson 1990). Afew detractors brand them minimalist, tautological,trivial, and even dangerous (Hirsch, Michaels, andFriedman 1987; Perrow 1986). The true significanceof agency theory probably is somewhere between the

Mark Bergen and Shantanu Dutta are Assistant Professors of Marketing,Graduate School of Business, University of Chicago, Orville C. Walker,Jr, is the James D, Watkins Professor of Marketing, Carlson School ofManagennent, University of Minnesota, The research was funded in partby the Graduate School of Business, University of Chicago, and by theBozell, Jacobs, Kenyon and Eckhardt Endowment Fund, The authorsgreatly appreciate comments from Shomu Banerjee, Jan Heide, GeorgeJohn, Herbert Mohring, and Ashkay Rao, as well as three anonymousJM reviewers. The authors' names are listed alphabetically.

key to universal truth promised by its fans and thetotal lack of usefulness suggested by its critics. Whatis clear, though, is that the theory addresses issues ofparticular relevance to marketing management. For thatreason it warrants a critical inspection.

Agency Relationships Pervade MarketingAn agency relationship is present whenever one party(the principal) depends on another party (the agent) toundertake some action on the principal's behalf. Hence,any employment relationship is an agency relation-ship. The hiring firm, or a manager representing theowners' interests, is the principal and the employee isthe agent. Given that most marketing departmentsconsist of multiple employees at various organiza-tional levels (product managers, salespeople, etc.),developing and implementing marketing strategies andprograms necessarily involves managing agency re-lationships.

In addition, marketers rely heavily on independent"facilitating agencies" to perform some of the workinvolved in implementing marketing programs. Facil-itating agencies are specialists—such as advertisingagencies, research suppliers, or public warehouses—

Journal of MarketingVol. 56 (July 1992), 1-24 Agency Relationships in Marketing / 1

whose services are contracted on a fee-for-service orcommission basis. Each of these contractual arrange-ments involves an agency relationship.

At a more fundamental level, agency relationshipsare pervasive in marketing because the essence ofmarketing is exchange, and "the agency relationshipis . . a significant component of almost all [ex-change] transactions" (Arrow 1985). This is particu-larly true because most goods and services are dis-tributed through intermediaries—such as wholesalers,retailers, or franchisees—who act as agents (in anagency theory sense) of the manufacturer or franchi-sor. In addition, the ultimate customer also can beviewed as engaging in an agency relationship as he orshe attempts to gain accurate product information anddesired product benefits from a supplier who may beviewed as his or her agent' (Coughlan 1988; Devinney1988).

But Agency-Based Research in Marketing isLimitedIn spite of the fact that marketing is rife with agencyrelationships, agency-based research has only recentlybegun to be reported in the marketing literature. Whyhave marketing scholars and researchers been rela-tively slow to embrace this new theory? Perhaps thelocation and nature of its early development in theeconomics and finance literatures made it somewhatinaccessible for many of us. The technical style andformal modeling structures of many of the early ar-ticles tend to obscure the rather straightforward con-structs that are the essence of agency theory. To makematters worse, the formal agency literature incorpo-rates two different but related types of models, one ofwhich focuses on precontractual and the other on post-contractual issues. There is also a "positive branch,"which is primarily concerned with the design of ap-propriate intraorganizational governance and controlmechanisms. These multiple foci can make it difficultfor the uninitiated to determine exactly what the the-ory is all about.

PurposeWhen stripped down to its essentials, however, agencytheory provides a useful framework for examining andgaining new insights into a wide range of marketing

'A given individual can shift from the role of principal to that ofan agent across different contexts. For instance, a purchasing agentacts primarily as an agent whose job is to represent the needs andinterests of others in the firm when buying goods and services fromoutside suppliers. However, when seeking information from a givensupplier on which to base a purchase decision, the purchasing agentacts like a principal and the supplier's salesperson is the agent. Thus,agency relationships are context specific, and agency models shouldbe used to examine the relationship that is present within a singlecontext.

issues. The primary purpose of our review is to cutthrough the methodological fog, describe the basiccontent of agency theory, and critically review its pastand potential future applications in marketing.

We first describe the conceptual framework ofagency theory, examining the theory's basic terms andconstructs, the two types of models embedded withinit, and its differences from the closely related frame-work of transaction cost analysis. Next, past appli-cations of agency theory with relevance for marketingare discussed. Though space limitations preclude anexhaustive review of agency research in other func-tional areas, we do incorporate findings from otherdisciplines—particularly economics—that are ger-mane to marketing issues. Finally, we discuss somepotential extensions and new applications of agencytheory in marketing, examine some of its limitations,and suggest ways in which marketing researchers mighthelp to improve its validity and usefulness.

The Conceptual FrameworkAgency theory uses the metaphor of a contract to de-scribe relationships in which one party delegates workto another (Jensen and Meckling 1976). The focus ofthe theory is on determining the most efficient con-tract to govern a particular relationship given the char-acteristics of the parties involved and the fact that en-vironmental uncertainty and the costs of obtaininginformation make it impossible for the principal tomonitor the agent completely.

Though most of the agency theory literature ad-dresses explicit, formal contracts—such as a com-pensation plan for a company's salesforce—the the-ory can also be applied to any "mutual agreementgoverning some aspect of subsequent behavior"(Mitnick 1987, p. 17). Hence, the theory can be usedto evaluate implicit "social contracts"—such as socialnorms, peer pressure, and peer acceptance—as wellas explicit legal contracts (White 1985).

It is important to note that most agency modelsdefine efficiency from the principal's point of view.The assumption is that the principal is the dominantparty in the relationship. Thus, an efficient contractis one that brings about the best possible outcome/orthe principal given the constraints imposed by the sit-uation, rather than one that maximizes the joint utilityof both principal and agent.

Two Types of Agency ProbiemsA principal faces two distinct kinds of problems whenentering a relationship with an agent. First, precon-tractual problems arise before the principal decides tooffer an agent a contract. The major issues here arewhether a particular agent has the characteristics theprincipal is seeking and what strategy the principal

2 / Journal of Marketing, July 1992

should employ to find out. These are the kinds ofproblems marketers face when recruiting new sales-people or marketing personnel, selecting dealers fortheir distribution channels, and choosing advertisingagencies or research suppliers.

Second, postcontractual problems emerge after theprincipal and agent engage in a relationship. The ma-jor issues here are how the principal should evaluateand reward the agent's performance so he or she willbe motivated to behave in a manner consistent withthe principal's goals and what information strategyshould be employed to make such evaluations. Mar-keters commonly face such problems when designingcompensation and incentive programs for marketingexecutives, salespeople, middlemen, and facilitatingagencies.

We refer to precontractual problems as problemsof "hidden information" and to postcontractual issuesas problems of "hidden action."^ Regardless of whatlabels are used, the important point is that these twosets of problems are addressed by two different typesof agency models. The variables incorporated in bothmodels are diagrammed in Figure 1 and discussed inthe following sections. Because the hidden action modeladdressing postcontractual problems has received thegreatest attention in marketing, we examine it first.

The Hidden Action ModePExogenous Variables

Hidden action models involve several assumptionsabout the principal and agent. First, both are usuallyassumed to be motivated by self-interest. Because mostmodels center on economic relationships, such self-interest typically translates into goals of maximizingprofits (or utility). This assumption provides a usefulstarting point and probably refiects the goals under-lying many agency relationships in marketing. How-ever, the theory can also accommodate relationshipsin which one or both parties pursue broader social goals,as when a principal seeks support for a new art mu-seum in order to improve the cultural climate of acommunity (Mitnick 1987).

Much of the agency literature refers to precontractual problems asproblems of "adverse selection" and to postcontractual issues as prob-iems of "moral hazard." However, we use the terms "hidden infor-mation" and "hidden action" as suggested by Arrow (1985), becausethey are more descriptive of the issues involved. An overview of thedifferences in perspective and content of the models addressing thetwo types of problems is given by Hart and Holstrom (1987). How-ever, the two types of problems need not be considered in isolation.Some recent models attempt to address hidden information and hiddenaction problems simultaneously (Davis 1991; Picard 1987).

'For an excellent elementary discussion of the theory underlyinghidden action models and a detailed bibliography, see Kreps (1990,ch. 16). Surveys of applications of hidden action models include thoseof MacDonald (1984), Arrow (1985), Pratt and Zeckhauser (1985),and Hart and Holstrom (1987).

FIGURE 1Agency Theory Variables

AGENCY THEORY VARIABLES

f ENVIRONMENT'!

"""Lincertaintyl'f Moderating •> factors ! -

PRINCIPAL

Personal Characteristics:' Goals' Risk Preferences' Information atwut agent's

charactenstics and actions

- - ^ -,-, ' ; ,

Choices:' Information gathering strategy' Selection of an agent" Design of contract offer

^ :

' ,!EFFICIENT"cdNtRACtl ''": ''

Personal Characteristics:' Goals' Risk Preferences° Reservation utility

•^

Choices:' Accept or reject contract' Choose actions

, . . > J ^ •

Actions:' Functional activities' Signals

REALiZEOOUTCOMES

< •

PAYOFFS1° Principal1° Agent

Another underlying assumption of hidden actionmodels is that principals labor under conditions of in-complete information. Though the principal is as-sumed to know something about an agent's charac-teristics and abilities, his or her knowledge about thatagent's actions on the job is neither perfect nor com-plete. In contrast, the agent has information that theprincipal would like to obtain. For example, thougha manager may know from call reports how many callsa salesperson made last week, that salesperson has ad-ditional information about how much effort he or sheexpended preparing for each call and the quality ofthe presentations made. Such information asymme-tries—when one party has information the other de-sires but does not have—characterize most agency re-lationships. The problem is that self-interest often makesthe agent reluctant to share the information with theprincipal, or may even motivate the agent to send theprincipal false information.

A third assumption in hidden action models is thatrealized outcomes are partly determined by environ-mental factors—such as economic conditions in the

Agency Relationships in Marketing / 3

market, competitors' actions, and tecbnologicalchanges—in addition to the agent's behavior. Unfor-tunately, environmental uncertainty is present becausesuch factors change over time, are difficult to predict,and are beyond the control of either principal or agent.This uncertainty not only makes it impossible to writea contract that foresees and provides for all possiblecontingencies, but also forces principals and agents tokeep in mind the resulting risk—together with theirown risk preferences—when making contractual

choices.'*Problems arise in agency relationships when the

principal and agent have different risk preferences andconflicting goals, which lead them to prefer differentcourses of action. Risk preference can be thought ofas the degree of an individual's or firm's preferencefor adventure rather than security (Arrow 1974; Pratt1964). A risk-averse individual prefers security andtherefore seeks some guarantee of the attainment ofdesirable outcomes or insurance against the occur-rence of undesirable outcomes. A risk-neutral indi-vidual is indifferent to adventure or security.

In most hidden action models the principal is as-sumed to be risk neutral—or at least more neutral thanthe agent—whereas agents are typically assumed tobe risk averse.^ The rationale is that agents, who areless able to diversify their employment, are likely tobe more risk averse than principals, who can diversifytheir investments.

Contractual VariablesIn the basic hidden action model, the agent attemptsto maximize his or her own utility by choosing thebest action available. Given the incompatible goals andrisk preferences of the two parties, though, actions theprincipal would like the agent to perform are oftenrelatively costly for the agent to undertake (i.e., theyrequire more time, effort, or other resources from theagent). Hence, the agent may try to "shirk" on suchactions.

To motivate the agent to engage in desired actionsand to reduce the likelihood that he or she will shirk,the principal must decide between two courses of ac-tion. First, the principal might collect more infor-mation about the agent's behavior by investing inmonitoring activities and systems, and could then writea contract that bases the agent's rewards on infor-

"Most hidden action models use Von Neumann-Morgenstem ex-pected utility maximization as the framework for modeling the impactof uncertainty on the participants' decisions, but other approaches forcoping with decision making under uncertainty might also be used.See Kreps (1990, ch. 3) for a discussion of this issue.

'Some models have relaxed the assumption that the agent is morerisk averse than the principal (e.g., Harris and Raviv 1979; Picard1987) This useful extension may increase the external validity of hid-den action models because research sugggests that people vary widelyin their risk preferences (MacCrimmon and Wehrung 1986).

mation about his or her behavior. Alternatively, theprincipal migbt write a contract that evaluates and re-wards the agent on the basis of realized outcomes, butone that includes incentives appropriate to motivatethe agent to pursue outcomes compatible with theprincipal's goals.

Monitoring systems and behavior-based con-tracts.^ Systems for monitoring agents' behavior arecommon in marketing. In a sales context, for in-stance, monitoring systems include requiring call re-ports, field observations by a sales manager, narrowspans of control, and the use of behaviorally anchoredrating scales for periodic review of tbe salesperson.

When such systems are in place, the principal mightwrite a contract specifying that the agent will be eval-uated and rewarded on the basis of information abouthis or her actual behavior. For example, raises in sal-ary and future promotions might be based on sales-peoples' ratings. Unfortunately, there are problems withthis behavior-based approach. Collecting detailed in-formation about the agent's behavior can be very costly.Also, even an extensive monitoring system cannotcompletely overcome the information asymmetriesbetween principal and agent. A sales manager mightmonitor the number of presentations a salespersonmakes by analyzing his or her reports, for instance,but would still not know how much effort and abilitythat salesperson devoted to preparing and making thosepresentations.

Despite such problems, in some conditions behav-ior-based contracts are more appropriate than the kindof outcome-based contracts we discuss next. Eisenhardt(1985, 1989) and others, using an approach consistentwith the positive branch of agency theory, have de-veloped some propositions about such conditions. Thesepropositions are summarized in Table 1.

Outcome-based contracts.'' The second optionavailable to the principal is to evaluate the agent's ac-tions on the basis of realized outcomes. The principalcould write a contract whereby the agent's rewards aredetermined by performance outcomes, such as salesvolume. In a selling context, the extreme form of anoutcome-based contract is a straight commission com-pensation plan. Under the right circumstances, such

<The taxonomy of behavior-based versus outcome-based contractsthat we use to organize the following discussion was originally de-veloped by Eisenhardt (1985, 1989). Note, too, that most contractsdo not cluster into only two dichotomous types. Instead, they are alonga continuum ranging from behavior-based to outcome-based. Thus,though a contract that compensates a salesperson by commissions andbonuses is largely an outcome-based contract, some of the salesper-son's bonus might be tied to specific behaviors, such as the numberof new account presentations made during the period. Also, for anexample of the formal agency literature's treatment of monitoring, seeAmershi and Hughes (1989).

'Classic works related to the topics in this section include those otRoss (1973), Mirlees (1976), Holstrom (1979), and Grossman andHart (1983).

4/Journal of Marketing, July 1992

TABLE 1 ^Propositions About Variabies Affecting the Relative Efficiency of Behavior-Based Versus

Outcome-Based Control Systems"

Proposition Rationale1. As the cost of monitoring an agent's actions de-

creases, the greater is the efficiency of behavior-basedcontracts in relation to outcome-based contracts

2. The greater the difficulty of measuring the outcomesof an agent's task, the greater is the efficiency of be-havior-based contracts in relation to outcome-basedcontracts

3. As the level of goal conflict between a principal andan agent decreases, the greater is the efficiency ofbehavior-based contracts in relation to outcome-basedcontracts

4. As environmental uncertainty increases, behavior-based contracts become more efficient in relation tooutcome-based contracts

5. As an agent's aversion to risk decreases, the greateris the efficiency of outcome-based contracts in rela-tion to behavior-based contracts

As the costs of monitoring an agent's behavior fall be-low the costs associated with shifting risk to the agentand his or her potential shirking, behavior-based con-tracts become more efficient from the principal's per-spective than outcome-based contracts.

When the agent's task requires a long time to complete,involves a team effort, or produces outcomes that aredifficult to evaluate objectively (e.g., creative adver-tising), the difficulties and costs involved in admin-istering outcome-based contracts increase.

As goal conflict decreases, the agent is more likely tobehave in accordance with the principal's wishes re-gardless of the incentives built into the contract. Thus,there is less motivational imperative for outcome-basedcontracting. The issue reduces to the difference in riskpreferences between the parties, and—assuming a risk-averse agent—behavior-based contracts become rel-atively more efficient.

Outcome-based contracts shift risk from the principal tothe agent. When uncertainty and risk are high, the coststo the principal of transferring risk to the agent canalso be high.

As an agent becomes less risk averse, the cost to theprincipal of transferring risk to that agent via an out-come-based contract declines.

'Adapted from material reported by Eisenhardt (1985, 1989).

contracts can motivate the agent to engage in the kindsof actions desired by the principal via a coalignmentof the two parties' objectives. The development of suchefficient contracts is the primary concern of most for-mal hidden action models.

For an outcome-based contract to be effective atinducing an agent to pursue actions consistent with theprincipal's objectives, however, it must be designedwith that agent's personal goals and decision pro-cesses in mind. It must meet the requirement of in-dividual rationality; that is, the contract must be in-teresting enough to the agent that he or she will wantto undertake the project. Each agent is assumed to havesome minimum reservation utility—the value to theagent of the best alternative opportunity outside a re-lationship with a given principal. Consequently, theagent may refuse to work for the principal if the po-tential payoff from the contract is not at least as at-tractive as what he or she can obtain elsewhere.

Obviously, too, the contract should be designedso that the actions with the highest payoff to the agentare also the actions that are most appropriate from theprincipal's point of view. This requirement is knownas incentive compatibility (Hurwicz 1972). When acontract is not incentive compatible, the agent mayshirk or undertake actions of less value to the prin-

cipal. For instance, if a firm's objective is to increasesales volume, offering a large commission would beone way to make increased volume a more incentive-compatible outcome for the salesperson.

The principal's goal, then, is to design a contractthat will obtain the "constrained best" outcome thatis incentive compatible for the agent. The principal'sproblem is that such outcome-based contracts shift riskfrom the principal to the agent, and—assuming a risk-averse agent—such a transfer of risk can be costly forthe principal.

Suppose, for example, that deteriorating economiccircumstances make it impossible for a firm to realizeits expected sales volume in the current year. A sales-person on straight salary would be unaffected, at leastin the short term, but a salesperson on straight com-mission would bear some of the consequences in theform of lower income. Thus, a risk-averse agent wouldfind the potential for eaming $100 in commissions lessvaluable than $100 in salary. A risk-neutral firm, then,might be tempted to compensate the salesperson en-tirely with salary and absorb the risk. After all, thefirm could pay less than $100 in salary, leaving thesalesperson perceiving himself to be as well off as hewas with a higher commission, and achieve higher ex-pected profit. Unfortunately, the salesperson would


then have less incentive to put forth substantial saleseffort. In other words, a commission structure may benecessary to make it incentive compatible for thesalesperson to expend the level of effort desired bythe firm.

For the principal, then, the heart of the problemis to design a contract that provides an efficient trade-off between the costs associated with shifting risk tothe agent and the agent's potential shirking. The mostappropriate solution depends on the level of environ-mental uncertainty, the kinds of tasks the agent is ex-pected to perform, and the characteristics of the twoparties—particularly their goals and risk preferences.

The Hidden Information Model̂The basic hidden information model addresses theproblems that arise because of information asymme-tries occurring before the principal enters into a con-tract with a given agent. The principal is assumed toknow the nature of the tasks the agent must performand the personal characteristics needed to perform thosetasks successfully. The problem is that the principaldoes not know whether a particular agent whom he orshe might hire has those desired traits. A principal canattempt to overcome this problem in at least three ways:screening, examining signals from potential agents, orproviding opportunities for self-selection.

ScreeningA principal must choose an information-gatheringstrategy to determine accurately a potential agent's truecharacteristics. He or she might collect additional in-formation—over and above the signals sent by theagent—by various screening activities. A sales man-ager, for example, might observe potential salespeo-ple through personal interviews, track down personalreferences, or administer a battery of aptitude tests.The problem, of course, is that such activities add coststo the hiring process.

Altematively, the principal might forgo screening,hire agents on the basis of the imperfect informationprovided by their signals, and then leam from sub-sequent experience the true characteristics of the agentshired. Here the problem is that the principal may hireagents of the wrong type, and those hiring mistakesmay lead to unsatisfactory performance outcomes. Fromthe principal's perspective, then, the basic tradeoff isbetween the costs of acquiring better informationthrough screening before offering a contract to an agentand the loss resulting from poor performance as a re-sult of hiring an inappropriate agent.

'Early work in the development of hidden information models in-cludes that of Akerloff (1970), Spence (1973), and Rothschild andStiglitz (1976). For a more detailed introduction to this topic and adetailed bibliography, see Kreps (1990, ch. 17).

Screening is most likely to be an efficient solutionto the hidden information problem when it is rela-tively easy for the principal to obtain information aboutimportant personal attributes of potential agents. Hence,screening is most likely to be extensive when mea-sures of agents' personal attributes have proven to bevalid predictors of their future performance and whenthose measures can be obtained by the principal withrelatively little effort or expense. These factors, forinstance, may help to explain the increasing use ofpaper-and-pencil aptitude tests in hiring salespeople(Kem 1988).

Even when screening is very costly, however, itmay still be the most efficient solution to the hiddeninformation problem if the consequences of a hiringmistake would be even more costly to the principal.Such is most likely to be the case when differencesin personal characteristics can result in wide varia-tions in performance across agents and when varia-tions in agent performance have a substantial impacton the principal's ultimate profit.

When screening costs and the costs of leaming byexperience are both high, the principal may have nochoice but to assume that all available agents are un-desirable types and to adjust the contract offer ac-cordingly. This breakdown, known as the "market forlemons," suggests that hidden information problemscan sometimes result in losses to both the principaland the agent (Akerloff 1970). Two possible solutionsto this problem are signaling and self-selection.

The Agent's Actions—SignalingGiven that an agent knows he or she has certain de-sirable capabilities, and that it is to his or her advan-tage to be offered a contract, the agent may engagein actions aimed at signaling to the principal that heor she is the type of agent the principal is seeking.For example, a salesperson might enroll in an eveningMBA program in order to attract an offer of a higherlevel, better paying job. The individual would takesuch an action in the belief that attaining an MBAwould improve his or her knowledge and skills andthat it would also signal potential employers that theindividual has desirable attributes such as a high levelof mental ability and intrinsic motivation. For such anaction to yield a worthwhile signal from the agent'spoint of view, however, it must meet the criteria ofindividual rationality and incentive compatibility. Theaction must make the agent better off than he or shewould be by doing nothing and the benefits to be de-rived from the action must exceed its cost.

Unfortunately, an agent who realizes he or she doesnot have the characteristics a principal is looking formay nevertheless find it in his or her self-interest tosend the principal false signals indicating that he orshe is the type of agent the principal desires. This would


be the case if the benefits the agent can derive fromduping the principal exceed the costs of engaging inthe action necessary to send a false signal. Hence, foran action to provide a signal that is useful from theprincipal's perspective, it must be more costly to pur-sue for agents with relatively little of the characteristicdesired by the principal than it is for agents with moreof the characteristic. In other words, it must not onlybe incentive compatible for agents of the desired typeto undertake the action, but also incentive compatiblefor inappropriate types of agents to avoid the action.For example, attaining an MBA probably meets bothrequirements as a useful signal to potential employersbecause the personal costs (e.g., effort, opportunitycosts) involved in attaining the degree are lower forindividuals with relatively high mental abilities andmotivational levels than for less intelligent and dili-gent individuals.

When a given signal enables the principal to dis-tinguish the characteristics of individual agents, it de-scribes a "separating equilibrium." Some signals,however, only separate a group of agents with a par-ticular characteristic from another group without thecharacteristic. This kind of "pooling equilibrium" en-ables the principal to distinguish the two groups fromeach other, but differences among members within eachgroup cannot be discovered (Spence 1973).

It is interesting to note the similarities between theproblems a principal faces when attempting to classifyand predict the behavior of "pooled" potential agentson the basis of their signals and those marketers facein defining and forecasting the behavior of marketsegments. This comparison suggests there may besubstantial synergies to be gained from melding agencytheory's hidden information models with the segmen-tation techniques developed in marketing.

The Principal's Actions—ProvidingOpportunities for Self-Selection

The principal need not be merely a passive receiverof signals from potential agents. He or she might alsoproactively construct choices that enable potential agentsto signal their abilities and/or willingness to expendeffort through self-selection. For example, a firm thatdesires a technically competent and knowledgeablesalesforce might require a lengthy and rigorous train-ing program for all new recruits. Potential employeeswho lack the necessary technical competencies or themotivation to acquire the desired knowledge (i.e., thosefor whom extensive technical training would not beindividually rational or incentive compatible) couldthen self-select themselves out of a relationship withthe firm by not applying for the job or by droppingout before completing the training.

The Positive Branch of AgencyTheory

A separate stream of inquiry within agency theory,known as the positive branch, has much in commonwith the basic hidden action model. Both are primar-ily concemed with the design of appropriate gover-nance and control mechanisms, though the positivebranch has tended to be more narrowly focused onintraorganizational relationships, particularly those in-volving corporate managers. One difference in thepositive branch, however, is that it has tended to adoptthe assumption that agents are risk neutral rather thanrisk averse, an assumption that is also common intransaction cost economics (Williamson 1988). More-over, the positive branch has differed from the rest ofagency research in that it has been less mathematicallyformal in its approach.

The positive branch of agency theory affords in-sights relevant to marketing management issues, in-cluding the design of contracts and control structuresthat are most successful at inducing managers to be-have in a manner consistent with the interests of thefirm's shareholders, and how the effects of those con-trol mechanisms are mediated by environmental fac-tors such as the labor market (e.g., Fama 1980; Famaand Jensen 1983a,b; Jensen and Meckling 1976). Thus,among other things, it provides some guidance aboutthe design of effective compensation and incentive plansfor marketing executives. For example, researchershave found empirical support for the notion that con-tingent compensation schemes—whereby a portion ofthe manager's income comes from stock or option plansor from bonuses tied to corporate performance—mo-tivate executives to behave in compliance with thefirm's strategic objectives and are related positivelyto shareholder wealth (Coughlan and Schmidt 1985;Murphy 1985).

The Relationship Between AgencyTheory and Transaction Cost

EconomicsThe potential usefulness of agency theory for exam-ining marketing issues is strengthened by the fact thattransaction cost analysis (TCA) has already been use-fully applied in the marketing literature (e.g., Anderson1985; Day and Klein 1987; Heide and John 1988; Johnand Weitz 1989) because the two offer closely relatedconceptual approaches. As Williamson (1988, p. 568)points out, " . . . these two perspectives are mainlycomplementary. Both have helped and will continueto inform our understanding of economic organiza-tion." Both examine efficiency aspects of how firmsorganize functional relationships. Both assume that theparties involved in a relationship are motivated by


economic self-interest and will engage in opportun-istic behavior. Moreover, though the two frameworksexplicitly incorporate some different exogenous vari-ables (e.g., asset specificity in TCA and risk prefer-ences in agency theory), many of those variables haveunderlying commonalities. For example, the notion ofinformation impactedness in TCA is similar to theproblem of environmental uncertainty in the agencyliterature.

Despite their commonalities, the two approachesdiffer along several dimensions.' One such dimensionis the unit of analysis. Whereas the transaction is thebasic unit of analysis in TCA, the individual agent isthe elementary focus of agency theory. Both are mi-croanalytic units and both imply a study of contract-ing. However, TCA's focus on the transaction has lednaturally to an examination of the basic dimensionson which transactions differ (e.g., asset specificity)and the implications of those differences for designingappropriate govemance structures—concems that havereceived little attention in agency research. In con-trast, agency theory's focus on the agent has led togreater attention to the impact of individual differ-ences across agents. Note, for instance, the emphasisof the hidden information model on selecting agentswith characteristics appropriate for the principal'spurpose, and the concern of the hidden action modelwith tailoring incentives for agents whose goals andrisk preferences differ from the principal's.

Another difference arises from the fact that TCAadopts an incomplete-contracting view of transactionsbetween principal and agent. Thus, it emphasizes expost transaction costs, which include (1) the maladap-tion costs incurred when transactions drift out ofalignment, (2) the haggling costs incurred if bilateralefforts are made to correct ex post misalignments, (3)the setup and running costs associated with the gov-emance structures to which disputes are referred, and(4) the bonding costs of effecting secure commitments(Williamson 1988, p. 572). Reducing these ex posttransaction costs by aligning transactions (which dif-fer in their attributes) with appropriate govemancemechanisms (the costs and competencies of which alsodiffer) is the distinctive TCA orientation.

In contrast, agency theory adopts an ex ante viewof relations between principal and agent. Its emphasisis on precluding or minimizing ex post costs throughan ex ante aligrunent of incentives. Historically, agencytheory has paid little attention to the possible prob-lems of subsequent maladjustment other than to allowimplicitly for the realignment of incentives to pricethem out.

'For a more detailed discussion of both the commonalities and thedifferences between agency theory and transaction cost analysis, seeWilliamson (1988).

Even some of these basic differences betweenagency theory and TCA have begun to blur, however,in recent years. For instance, as TCA has examineda wider range of "hybrid" govemance mechanismsbetween markets and hierarchies, it has begun to paymore attention to the characteristics of individual agentsand to incentives appropriate for different govemancestmctures, such as characteristics and incentives thathelp build "trust" or "conunitment" (e.g., Heide andJohn 1992). Similarly, recent work in agency theoryhas focused more attention on the incomplete-con-tracting problems involved in achieving an ex antealignment of incentives between principal and agent(e.g., Grossman and Hart 1986).

Given that TCA and agency theory are concemedwith similar issues and appear to be moving towardeven more common conceptual ground, blending con-structs and propositions from the two theories mayfurther improve our understanding of marketing phe-nomena. For example, Anderson and Oliver (1987)compare and combine propositions from agency the-ory and TCA, as well as other perspectives, to gaininsights about the appropriate conditions for the useof altemative compensation systems for controlling asalesforce. Their contributions add further weight tothe argument that agency theory and TCA should betreated as complementary perspectives.

Agency Applications in MarketingIn recent years, researchers in marketing and relateddisciplines have employed agency theory to examinea variety of marketing issues, including (1) salesforcemanagement, (2) channel coordination and control,and (3) promotion and other market signaling deci-sions. We discuss a representative, though not ex-haustive, sample of the work in each of these areas.The applications are summarize in Table 2.

Though the surface has been scratched, many un-answered questions remain in each of the three topicareas. Consequently, the following discussion in-cludes suggestions for extending and improving uponprevious research. A subsequent section also exam-ines other issues that, though amenable to the appli-cation of agency theory constmcts, remain largelyunexplored.

Salesforce ManagementThe relationship between sales manager and salesper-son is an agency relationship, and many of the issueswith which managers stmggle when hiring, motivat-ing, and controlling sales personnel can be examinedwith agency-based analyses. To date, however, mostof the literature has centered on a single issue—de-termining the most appropriate type of sales compen-sation plan to offer under various conditions.


TABLE 2Summary of Representative Agency Applications in Marketing

Authors

Salesforce CompensationBasu et al. (1985); Laiand Srinivasan (1989)

Lai, Outland, andStaelin (1990)

John and Weitz (1988)

John, Weiss, andWeitz (1987)

Lai and Staelin (1986)

Oliver and Weitz(1991)

Eisenhardt (1985)

ExogenousVariables

Risk aversion;environmentaluncertainty

Risk preferences;environmentaluncertainty

Environmentaluncertainty

Proportion of nonsellingactivities

Types of salespeople;environmentaluncertainty

Risk preferences;uncertainty in effort-performancerelationship

Environmentaluncertainty; taskprogrammability

FocalContractualVariables

Salary vs. commissioncompensation




Menu of compensationcontracts

Salesperson'smotivation; preferencefor salary vs.commissioncompensation

Behavior-based controlsystems (i.e., salarycompensation)

Context

Hidden action model;single period anddynamic

Three salesforces of aFortune 500 computercompany

Survey of 161 largemanufacturers

Survey of 161 largemanufacturers

Hidden informationmodel

Survey of 367salespeople frommore than 100companies

54 specialty retail stores

Conclusions

The proportion of salaryto commissioncompensation shouldincrease as theuncertainty of theselling environment orthe salesperson's riskaversion increases

Support for many ofBasu et al.'s (1985)propositions

Limited support forsome of Basu et al.'s(1985) propositions

As the proportion ofnonselling activitiesrequired increases.the proportion ofsalary compensationalso increases

Offering a variety of payplans to the salesforceallows self-selectionby salespeople

Risk-averse salespeopleprefer low levels ofincentivecompensation, butperceived uncertaintyhas only marginalimpact oncompensationpreferences

Use of behavior-basedcontrol increases asenvironmentaluncertainty and/ortask programmabilityincreases

Coughlan andNarasimhan(forthcoming)

Opportunity cost of asalesperson's time

Channel Coordination and ControlMoorthy (1987) Ouantity discounts; two-

part tariffs

Salary as a proportionof total compensation

Degree of channelcoordination; channelcosts

286 firms in 39industries

Hidden action model;no uncertainty

Mathewson andWinter (1984, 1986)

Promotional spillovers;retailer differentiation

Manufacturer's profits;resale pricemaintenance;exclusive territories

Hidden action model;many retailers;promotionalspillovers; nouncertainty

Existence of promotionprograms is positivelyrelated to percentageof salary in total pay

Ouantity discounts or atwo-part tariff canhelp increase channelefficiency by betteraligning retailers'incentives with themanufacturer'sobjectives

Resale pricemaintenance—andunder someconditions, exclusiveterritories—i ncreasethe manufacturer'sprofits when there arepromotional spilloversacross retailers


TABLE 2 (continued)

AuthorsExogenousVariables

FocalContractualVariables Context Conclusions

Ray and Tirole (1986a)

Dutta, Bergen, andJohn (1991)

Bergen and John(1991)

Brickley and Dark(1987)

Norton (1988)

Banerji and Simon(1991)

Mathewson andWinter (1985); Lai(1990)

Risk preferences;environmentaluncertainty

Manufacturer's profits;resale pricemaintenance;exclusive territories

Hidden action modelwith two retailers

Risk preferences;potential foropportunism

Manufacturer's profits;exclusive territories

Hidden action modelwith two retailers;environmentaluncertainty

Promotional spillovers;retailer differentiation

Environmentaluncertainty;observability ofretailer behavior

Manufacturer's profits;two-part tariff;cooperativeadvertising allowances

Efficiency of franchisedvs. company-ownedoutlets

Industry labor-output Proportion of franchisedratios and percentage to company-ownedof population living in outletsrural areas used asproxies for monitoringcosts

Value of a franchisor'strademark, asmeasured by Torbin'sO-ratio

Environmentaluncertainty;observability offranchisee behavior

Proportion of franchisedto company-ownedoutlets in a firm'sdistribution channel

Efficiency of franchiseddistribution systems;two-part tariffs;royalty payments

More than 1800 coopadvertising contractsacross many differentindustries

10,524 retail units from36 franchiseorganizations

Aggregate industry datadrawn from 1977Censuses of RetailTrade and ServiceIndustries

Survey of 34 franchisorsin the fast-foodindustry

Hidden action model;environmentaluncertainty;promotional spillovers

When retailers are veryrisk averse, themanufacturer's profitsare greater when itallows unrestrictedcompetition amongretailers than when itimposes either resaleprice maintenance orexclusive territories

When retailers are riskneutral and there ishigh potential forretailer opportunism(e.g., bootleggingacross territories), themanufacturer's profitsare higher whenretailers are allowedto compete than whenthey are assignedexclusive territories

Cooperativeparticipation rates arehigher for products inindustries withrelatively highspillovers

Franchising is preferredto vertical integrationwhen the environmentis uncertain and it isdifficult for themanufacturer tomonitor the behaviorof individual outlets

Both surrogates formonitoring costs wererelated negatively tothe' incidence offranchising acrossindustries

Proportion of franchisedoutlets is relatedpositively to the valueof the franchisor'strademark

Royalty paymentsenhance franchisesystem efficiency onlywhen both thefranchisor and thefranchisee must investin factors that affectretail sales

Consumer PromotionsNelson (1974) Volume of

manufacturer'sadvertising

Product type;manufacturer's profit

Hidden informationmodel withasymmetricinformation

For "experience" goods(i.e., those that aredifficult for customersto evaluate prior touse), advertising canact as an efficientsignal of productquality


TABLE 2 (continued)

AuthorsExogenousVariables

FocalContractualVariables Context Conclusions

Milgrom and Roberts(1982)

Klein and Leffler(1981)

Rao and Monroe(1991); Rao andBergen (1991)

Manufacturer's costs;asymmetricinformation

Repeat purchase rate;asymmetricinformation

Quality consciousnessof purchaser;perceived ability toevaluate qualityobjectively prior topurchase

Manufacturer's profit;relative price; volumeof manufacturer'sadvertising

Manufacturer's short-and long-term profits;product quality

Price premium paidabove full-information,profit-maximizingprice

Hidden informationmodel withasymmetricinformation

Repeated hidden actionmodel

Survey of 149 seniorpurchasing executivesin a variety ofindustries

Relatively high priceand heavy advertisingcan jointly provide anefficient signal ofproduct quality

Premium received forhigh quality productsgives manufacturer anincentive to maintainquality over time

Higher price premiumsare likely whenpurchaser is relativelyprice conscious andwhen product qualityis relatively difficult toevaluate objectively

Sales compensation plans. Agency models devel-oped by Basu et al. (1985) and Lai and Srinivasan(1989) derive several propositions about the optimalmix of salary and commission components within asales compensation plan under a variety of conditions.For example, these models suggest that the proportionof salary to commission compensation should increaseas the amount of environmental uncertainty increasesand as the salesperson's aversion to risk increases.

A growing body of evidence provides some sup-port for these propositions. For instance, using busi-ness-level data on compensation plans collected froma survey of 161 large manufacturing firms, John andWeitz (1988) found support for many of Basu's prop-ositions and, though others were not confirmed at sta-tistically significant levels, the evidence was gener-ally in the predicted direction.

In a related article, John, Weiss, and Weitz (1987)used data from the same survey to find support forseveral additional compensation hypotheses. For ex-ample, they found that as the proportion of nonsellingactivities required of the salesforce increases, higherproportions of salary tend to be offered.

Similarly, a recent study by Oliver and Weitz (1991)also provides mixed support for several agency-basedpropositions. In a cross-sectional survey of 367 sales-people from more than 100 companies, Oliver andWeitz examined respondents' perceptions about thecharacteristics of their current compensation plan, riskpreferences, the amount of uncertainty in the rela-tionship between their selling effort and sales perfor-mance, and their motivation to expend effort on thejob. As agency theory predicts, they found that risk-averse salespeople prefer (i.e., choose to work for firmsthat offer) compensation plans that have low levels ofincentive compensation. In contrast, their results sug-

gest that perceived uncertainty in the effort-sales re-lationship has only a marginal impact on preferencefor salary versus incentive compensation.

Lai, Outland, and Staelin (1990) explored a sim-ilar set of propositions, but at a more disaggregate level.By examining differences in sales compensation planswithin three separate salesforces of a single Fortune500 firm, they found significant support for thosepropositions.

Finally, Eisenhardt (1985) also tested severalagency-based propositions about the relative effi-ciency of altemative control systems with data from54 retail stores. She found that behavior-based controlsystems (i.e., a high proportion of salary) are morelikely under conditions of high environmental uncer-tainty and when salesperson behavior is relatively easyto monitor.

Potential extensions. The agency studies are in-teresting first steps toward understanding the appro-priate use of different sales compensation and controlsystems, but many questions remain unanswered. Forinstance, future models might examine the effects ofadditional exogenous variables such as:

• The differences in goals and risk preferences that occuracross salespeople at different stages in their career cycles(Cron and Slocum 1986). For example, as the risk aver-sion of salespeople increases in the later stages of theircareer cycles, the theory suggests that the efficiency ofsalary-based compensation should increase. Indeed, arecent study by Coughlan and Narasimhan (forthcoming)based on survey results from 286 firms in 39 industriesfound that salesforce compensation is more heavilyweighted toward salary in firms whose salespeople's av-erage years of service is relatively high. However, yearsof service in a particular firm is not a perfect indicatorof a salesperson's career stage. Also, the career stagesmodel suggests that other psychological changes occur

Agency Relationships in Marketing / 1 1

as people move through their careers (e.g., shifts in goalsor valences) that may affect the efficiency of differentcompensation and reward systems.

• The demands imposed by different types of selling jobs(Moncrief 1986). Different types of selling jobs requiredifferent amounts of effort on different sets of activities(q.g., servicing current accounts vs. prospecting for newcustomers). Because those activities may involve differ-ent amounts of risk and produce different types of out-comes, a given compensation scheme may be more ef-ficient for some types of sales jobs than others.

• Differences in social norms and performance attributionsacross salespeople (Teas and McElroy 1986). The strengthof norms such as "trust" or "organizational commit-ment" may moderate the use of commission schemes ina firm's total compensation package. Hence, measuresof such norms, similar to those recently developed in theTCA literature (e.g., Heide and John 1992), might beincorporated into future agency studies.

Also, additional contractual variables reflecting thewide variety of incentives that finns offer—from bo-nuses to sliding commissions to job promotions—mightbe incorporated in future work. For instance:

• One important question concems the conditions underwhich pay plans based on relative performance (as is thecase in many sales contests) are superior to pay plansbased on absolute performance, such as commissions orbonuses (Coughlan and Narasimhan, forthcoming).

• Job promotions have been examined as tournaments thatprovide incentives for good performance (Ferrall 1991;Green and Stokey 1983; Lazear and Rosen 1981), butthe effects of promotion opportunities on performancehave received little attention in a sales context.

• Questions about interactions among multiple rewards,and the optimal mix of rewards, also should be explored.As a first step in this direction, Coughlan and Narasim-han (forthcoming) found that the presence of promotionopportunities is related positively to the proportion ofsalary in salesforces' compensation plans.

Additional insights might also be gained by com-bining more traditional sales management variableswith those of agency theory. For instance, Oliver andWeitz (1991) combined agency constructs with vari-ables from an expectancy model of motivation. Oneinteresting result is the finding that risk-averse sales-people who receive a high proportion of incentive payactually experience an increase in motivation when theyare uncertain about the relationship between effort andsales performance. The authors speculate that sales-people working under incentive plans may feel the needto achieve some minimum level of income, and thatthey are therefore motivated to work harder when con-fronted with uncertainty so that they can achieve thattarget level of income.

The previous work is also largely restricted toanalyses of compensation plans for individual agents.Because of the growing use of team selling and theproblems firms encounter in motivating and compen-sating such teams (Murray 1991), future extensionsmight examine how different compensation and con-

trol systems affect the performance of sales teamsconsisting of multiple agents with heterogeneouscharacteristics.

Finally, many other sales management issues be-sides the design of compensation plans might be ex-amined within the agency framework. For instance,several organizational design issues might be ad-dressed, such as:

• Whether to rely on extemal sales agents (e.g., manu-facturers' reps) or to employ company salespeople. Us-ing agency-based arguments, one might hypothesize thatmanufacturers' reps are preferable when uncertainty ishigh or when monitoring costs are high.

• The appropriate span of control for sales managers. Con-ditions favoring behavior-based contracts may also ne-cessitate narrower spans of control.

• The appropriate basis for salesforce specialization (e.g.,organization by customer type vs. product line). Suchdecisions are likely infiuenced by differences in envi-ronmental uncertainty and the ease of monitoring sales-person behavior.

Similarly, the hidden information model might pro-vide insights on personnel issues, such as (1) the mostefficient amount of screening for a firm to use whenhiring new sales recruits, (2) the kinds of signals thatbest predict future selling success,.and (3) appropriatecriteria for selecting salespeople for promotion intomanagement.

Operationalization and measurement. Much needsto be done to improve the operationalization and mea-surement of crucial agency constructs such as envi-ronmental uncertainty, reservation utility, and the goalsand risk preferences of principals and salespeople. Forexample, a variety of different operational definitionsof the "uncertainty" construct can be found in the lit-erature. John and Weitz (1988) and others measurethe construct in terms of the uncertainty of productsales. However, Oliver and Weitz (1991) suggest thata more appropriate definition in a sales context is theperceived uncertainty in the relationship between asalesperson's effort and the sales produced, and theyemploy multi-item scales to measure the construct.

A related problem is the measurement of risk pref-erences. Such preferences can be difficult to pin downin a survey setting. Researchers have tended to use alottery technique in which subjects are asked to choosebetween risky alternatives and their certainty equiv-alents (e.g., Oliver and Weitz 1991), but such mea-sures can have validity and reliability problems (Hauserand Urban 1979). One altemative for research pur-poses might be to manipulate such preferences ex-perimentally. For example. Berg et al. (1986) induceddifferent risk preferences (i.e., risk aversion and riskloving) in a laboratory setting by employing different


transformation functions.'° In a field setting, some an-alysts (e.g., Coughlan and Sen 1989) suggest that afirm might offer salespeople a menu of contracts withthe same expected value and infer risk aversion fromtheir choice of contract, thereby offering an oppor-tunity for self-selection by salespeople. Indeed, Laiand Staelin (1986, p. 180) report that IBM and St.Regis Paper have offered such a menu of contracts totheir salespeople and allowed them to select differentcompensation plans by prespecifying their expectedsales volume for the next period. Finally, a more per-plexing measurement issue involves finding ways toseparate the effects of risk from other factors in asalesperson's decision process (Dyer and Sarin 1982).

Measuring reservation utility, which correspondsto the "most desirable altemate employment" (Basuet al. 1985, p. 274) available to the salesperson, isalso problematic. It often is estimated by examiningwage rates for alternative jobs inside the firm or inother, similar firms (e.g., John and Weitz 1988) orfor salespeople with similar seniority and educationlevels within the same industry (Coughlan andNarasimhan, forthcoming). However, monetary wagesare conceptually very different from the utility an agentwith specific characteristics and preferences might gainfrom an alternative job. One possible solution may beto develop multi-item measures of the perceived avail-ability and relative attractiveness of alternative em-ployment opportunities.

Another research problem is inconsistent levels ofanalysis between hypothesized relationships and thedata used to test them. Though agency theory is ba-sically concerned with relationships between individ-uals, much of the empirical work focuses on firm- orbusiness-level variables, such as differences in theproportion of salary across firms' compensation plans.

Some researchers have examined agency propo-sitions at an individual level of analysis, but they havetended to rely on cross-sectional surveys (e.g., Oliverand Weitz 1991), some conducted within a single firm(Lai, Outland, and Staelin 1990). Those researchersadmit that a more appropriate dataset for conductingsuch research " . . . would be longitudinal and includemeasures of the personal, environmental, and orga-nizational factors postulated to affect the form of thecompensation for each person" (Lai, Outland, andStaelin 1990, p. 9). In contrast, some researchers ar-gue that the business level is a more realistic, andtherefore appropriate, level of analysis to employ whenstudying variations in sales compensation plans eventhough it is inconsistent with the individual-level fo-cus of agency theory. Their rationale is that person-specific compensation plans tend to be found only at

'"For a more detailed discussion of alternative measures of risk andrisk preference, see MacCrimmon and Wehrung (1985).

top management levels and that most firms currentlyemploy a single, homogeneous compensation plan foreveryone in the salesforce (John and Weitz 1988). Inany case, it is clear that researchers need to devotemore thought to the problems involved in matchingthe conceptual focus of agency theory with the con-straints of "mundane realism" (Berkowitz andDonnerstein 1982). Such problems may help explainwhy some agency theoretical propositions have re-ceived less than perfect empirical support.

Channel Coordination and Control^^A distribution channel constitutes a set of agency re-lationships. The manufacturer depends on resellers toperform a variety of functions on its behalf, includingthe provision of shelf space, local advertising, point-of-purchase promotion, and implementation of an ef-fective pricing strategy.'^ However, differences in goalsand preferences between a manufacturer and a resellercan lead to confiicts over how channel revenues andcosts should be allocated among the participants andcan produce incentive problems within the channel. Ifa powerful manufacturer attempts to increase profitsby offering relatively low margins to a reseller, forinstance, that reseller may not have adequate incen-tives to promote or price the product very aggres-sively. In tum, the reseller's tendency to shirk by un-derpromoting or overpricing the product under suchcircumstances may lower sales volume and lead to aless than optimal amount of profit for the manufac-turer.

A manufacturer might attempt to control the ac-tions of the members of its distribution channel in sev-eral ways by making it incentive compatible for themto engage in actions consistent with the firm's objec-tives. Such controls include pricing mechanisms, con-straints on channel members' freedom of action (e.g.,promotional allowances), and franchising arrange-ments.

Pricing mechanisms. Current models suggest thatincentive problems can sometimes be overcome throughvarious types of pricing mechanisms. One is for themanufacturer to offer the reseller quantity discountsin order to increase marginal profit from selling ad-ditional volume and thus give the reseller more in-centive to expend the effort needed to do so. A secondapproach for aligning the reseller's incentives with themanufacturer's objectives is for the manufacturer toset a "two-part tariff" consisting of both a wholesaleprice and a franchise fee. The wholesale price should

"For a more extensive survey of the economic research on channelrelationships, see Rey and Tirole (1986b), Tirole (1989, ch. 4), andKatz (1989).

'^Likewise, retailers and wholesalers also depend on their suppliersto perform certain functions, but most of the literature to date has notexamined issues from their perspective.


reflect only the true marginal cost of the product,whereas the franchise fee should constitute the man-ufacturer's share of the resulting profits. Because thefranchise fee is fixed, the retailer is better off sellinga larger volume of the product and is thus motivatedto do so (Moorthy 1987).

Vertical restraints. It is often in a manufacturer'sinterest for its retailers to engage in local advertisingand promotion efforts aimed at stimulating demandfor its brand. Retailers may be reluctant to undertakesuch efforts, however, particularly when a brand isdistributed through several competing retailers andwhere promotional "spillovers" occur. It may not beincentive compatible for a retailer to promote a prod-uct aggressively in such circumstances because he orshe is unable to capture the full benefits of such ef-forts. Some of the resulting sales will accrue to com-peting retailers. Worse, such situations provide a re-tailer with a motivation to "free ride," to forgopromotional expenses while reaping the benefits of otherretailers' efforts.

One way a manufacturer might prevent its retailersfrom engaging in opportunistic actions in such situa-tions is by imposing constraints on their behavior. Suchconstraints might (conceptually, at least) take the formof exclusive territory assignments or resale pricemaintenance. Such restrictions reduce the ability of afree-riding retailer to entice customers away from otherretailers. Also, by limiting the competitive erosion ofretail margins, they help ensure appropriate incentivesto overcome spillover effects.

Work by Mathewson and Winter (1984, 1986) in-dicates that resale price maintenance—and sometimesexclusive territories—can work in conjunction with atwo-part tariff to improve the efficiency of channelsinvolving competing retailers and promotional spill-overs, but no environmental uncertainty. However,other work indicates that the usefulness of vertical re-straints is moderated by the level of environmental un-certainty, retailers' risk preferences, and their gov-ernance properties. For instance, Rey and Tirole (1986)explore the impact of differing risk preferences onchannel outcomes when there are no promotionalspillovers among retailers, but when the environmentis uncertain and there is information asymmetry (i.e.,retailers have more infonnation about local demandand retail costs than the manufacturer). Their findingsindicate that when retailers are very risk averse, amanufacturer's profits are greater when it allows un-restricted competition among those retailers than whenit imposes either resale price maintenance or exclusiveterritories. Further, Dutta, Bergen, and John (1991)show that, from a manufacturer's perspective, thegovernance properties (i.e., enforceability) of exclu-sive territories have a substantial impact on a chan-

nel's profitability. They suggest that exclusive terri-tories are more desirable when the dealers perceive a"longer shadow of the future" with respect to themanufacturer.

Functional incentives. Instead of trying to imposerestrictions on its resellers, a manufacturer might re-solve incentive compatibility problems in a channelby offering positive inducements in the form of func-tional incentives. Such incentives involve additionalpayments to the agent if he or she engages in a spe-cific activity in the manufacturer's interest. Examplesinclude co-op advertising and promotional allow-ances. For instance, Bergen and John (1991) show thatwhen there are promotional spillovers between retail-ers but no environmental uncertainty, promotional al-lowances can work in conjunction with a two-part tar-iff to improve channel efficiency.

Franchising agreements. Agency theory can alsohelp identify conditions in which franchising leads togreater channel efficiency than do other alternatives,such as vertical integration. For instance, Brickley andDark (1987) suggest that franchising is more efficientthan using company-owned outlets when there is en-vironmental uncertainty and when it is difficult for themanufacturer to monitor the behavior of individualoutlets. Though company-owned outlets do not ne-cessitate sharing profits with an intermediary, suchprofit sharing provides strong incentives for franchi-sees to operate efficiently. Hence, unless the companyis able to monitor the behavior of local managers ef-fectively, the increase in margin gained by directownership is not sufficient to offset the greater effi-ciency of franchisees. Brickley and Dark found em-pirical support for their propositions from a survey of36 franchise organizations with 10,524 retail unitswhere the geographic proximity of retail outlets to themanufacturer's headquarters was used as a proxy forthe ease of monitoring an outlet's behavior.

In a supporting study, Norton (1988) used industrylabor-output ratios and the proportion of a state's pop-ulation living in rural areas as proxies for a franchi-sor's monitoring costs. His rationale was that moni-toring costs should increase as a greater amount oflabor is involved in an outlet's operations and as greatergeographic dispersion is necessary to cover a sub-stantially rural population. As hypothesized, both sur-rogates for monitoring cost were related negatively tothe incidence of franchising when examined across in-dustries with aggregate data drawn from the U.S.Censuses of Retail Trade and Service Industries.

Agency research has also demonstrated that the in-cidence of franchising in a distribution channel is re-lated positively to the expected value of the franchiseto potential franchisees. Banerji and Simon (1991), ina study of 34 franchisors in the fast food industry.


found the proportion of franchised outlets in theirchannels to he linked positively to the value of thefranchisor's trademark, as measured by Tohin's Q-ratio.They argue that the incentives for franchisors andfranchisees are more closely aligned in channels wherethe value of the franchisor's trademark is high becauseof the size of the future income stream the franchiseewould lose should his or her contract be terminatedfor cheating.

Another franchising question that has been ad-dressed involves the most appropriate terms of thefranchise contract under various conditions. For in-stance, Mathewson and Winter (1985) and Lai (1990)demonstrate that incorporating royalty payments in afranchise contract is most desirahle when actions takenby hoth the manufacturer and the franchisee have sig-nificant positive effects on retail demand. When amanufacturer's actions have little effect on demand,it is better off offering a contract that incorporates afixed initial franchise fee, quoting a wholesale price,and then monitoring franchisees' actions.

Potential extensions. Though constructs fromagency and related theories have heen applied to a widevariety of channel issues, many research opportunitiesremain because most of the propositions that have beendeveloped have gone largely untested. In addition, fu-ture research should attempt to incorporate a more de-tailed and comprehensive set of the variables that af-fect channel relationships. For example:

• Because most previous research is static, examinationsof how solutions to agency problems in channels mayvary over time are needed. Some analysts have begunto develop propositions about shifts in channel members'behavior, and the appropriate incentives to direct thatbehavior, over time (e.g., Lambert 1983; Radner 1981,1985), but those propositions have yet to be examinedwith longitudinal data. One approach for incorporatingtime into channel analysis is to examine whether pricepremiums offered by a manufacturer create incentivesfor its retailers to provide appropriate levels of serviceover time (Klein and Leffler 1981). Indeed, assigningexclusive territories may be one way to ensure such pricepremiums (Klein and Murphy 1988).

• Though many real-world channels involve resellers whohandle products from multiple manufacturers, most agencymodels assume a single principal. Some analysts havebegun to tackle the competitive issues that emerge whenmultiple principals deal with the same agent (e.g.,Bemheim and Whinston 1985). For instance, there issome evidence that a manufacturer is more likely to dis-tribute its product through retailers who carry competingbrands than to integrate vertically when substitutabilityacross brands is high because the retailer helps deflectcompetition among the manufacturers (Coughlan 1985;McGuire and Staelin 1983). Some analysts have also ex-amined problems that emerge when multiple principalsdeal with multiple agents (Bergen 1990; Rey and Stiglitz1988), but many such issues have yet to be examined.

• In most agency-based analyses, resellers are assumed tohave little or no power in their dealings with the man-

ufacturer. In these days of scanner data, mergers, andincreased concentration among resellers, many channelsare characterized by relative power distributions that areeither balanced or skewed in favor of large resellers suchas Walmart, Jewel, and Sears. Yet, most published workhas not yet grappled with the complications that emergewhen the agent is assumed to have power in relation tothe principal. One possible step in this direction is sug-gested by the work of Jeuland and Shugan (1983, 1988),who treat channel members as symmetric partners, eachmaking independent decisions and sharing the ensuingprofits. Their fmdings suggest that quantity discountschedules can be an effective coordination device in suchbalanced power relationships. Future work might ad-dress reseller power by incorporating bargaining pro-cesses into the framework of the hidden action model(Dwyer and Walker 1981; Zusman and Etgar 1981) andby allowing for offers by agents as well as principals.

• Differences in social norms and performance attributionsacross channel members may also affect the efficiencyof various control mechanisms and incentives. Research-ers have only recently begun to examine the effects ofsocial norms such as "trust" and "commitment" withina channels context (Heide and John 1992). Such normsare likely to moderate the need for restrictive contractualarrangements, such as exclusive territories, and for com-plex incentive schemes like two-part tariffs.

Operationalization and measurement. As with theagency-based research on salesforce compensation, theresearch on channel relationships has shortcomings inthe operationalization and measurement of commonagency constructs, such as environmental uncertainty,reservation utility, and the goals and risk preferencesof principals and agents. In addition, the proxy vari-ables used to measure channel-specific constructs, suchas monitoring costs, the value of the principal's trade-mark, and promotional spillovers, have also been sus-pect.

For example, Brickley and Dark's (1987) relianceon geographic proximity and Norton's (1988) use oflabor intensity and population dispersion as indicatorsof a manufacturer's ability to observe the behavior ofretail outlets all seem questionable in the presence ofscanners and computerized inventory control systems.Similarly, Banerji and Simon's (1991) use of Tobin'sQ-ratio as an indicator of the value of a manufactur-er's trademark can be questioned because it is infiu-enced by several other factors not controlled in theirstudy. Yet another measurement problem is illustratedby the work on promotional spillovers. For instance,Bergen and John (1991) assume that industrial prod-ucts have fewer spillovers than consumer products be-cause intermediaries are less likely to promote suchproducts through mass media. Though their assump-tion may be valid, it yields a rather gross and uncal-ibrated measure of the construct.

Agency research on channel relationships also hassome of the same level-of-analysis problems encoun-tered in other subject areas. Agency theory is con-cemed with relationships between individual firms.


but at least some of the empirical research has at-tempted to test agency propositions with aggregate dataon differences in channel structures across entire in-dustries (e.g., Norton 1988).

Many of the operationalization and measurementproblems arise from the fact that much of the researchin the channels area has relied on secondary data totest agency-based propositions. Hence, more fieldsurveys and the use of more specific measures devel-oped by sound psychometric techniques should helpresolve some of these problems. Laboratory studieswould allow direct manipulation of agency constructsthat are difficult to measure or control in field set-tings, such as risk preferences, reservation utilities,and relative power. Because marketing has a historyof laboratory studies of channel issues (e.g., Dwyerand Walker 1981; Eliashberg et al. 1986), establishedtechniques can be melded with related work in ex-perimental economics (e.g.. Berg et al. 1986) to de-velop more useful methods for studying agency-basedpropositions in a channels context.

Consumer Promotion and Competitive SignalsWhen customers make purchase decisions, they searchfor infomiation about altemative brands. This searchmay include personal and noncommercial sources ofinformation, but it is also likely to incorporate infor-mation from the manufacturers, distributors, or. re-tailers who are competing for the customer's business.After all, the producer of a high quality brand hasinfomiation about the product that customers mightfind useful and that would be in the producer's self-interest to share with potential buyers. Unfortunately,producers of inferior brands might be motivated tomislead customers into believing that their productsare better than they actually are by heavily promotingtheir brands or engaging in puffery. The crucial ques-tion for a manufacturer, then, is whether the costs andrevenues associated with a particular promotional ac-tivity make it a valid and efficient signaling mecha-nism for helping potential customers differentiate be-tween high quality brands and their inferior competitors.This question can be analyzed as a hidden informationproblem in which the customer is viewed as a prin-cipal who must choose among several altemative agents(manufacturers).

In general, agency models suggest that the size ofpromotional expenditures can serve as an efficient sig-nal of product quality. Nelson (1974) was one of thefirst to suggest that the volume of advertising couldprovide such a signal. More recent models show thatpromotional investments can work in conjunction withhigh prices (Milgrom and Roberts 1982) and invest-ments in expanded production capacity (Devitiney 1988)to efficiently signal high product quality. Indeed, pricealone can serve as a signal of product quality, as Davis

(1991) shows in a world with both hidden informationand hidden action problems.

Though a high price can signal high quality, it canalso provide an incentive for the manufacturer tomaintain the quality of its product over time. Kleinand Leffler (1981) used a repeated hidden action modelto examine the price premium that high quality man-ufacturers must receive to make it incentive compat-ible for them to continue producing a high qualityproduct rather than maximizing profits in the currentperiod by skimping on quality while charging a highprice. Consumers may stop buying a manufacturer'sgoods if they discover that it has begun to lower qual-ity. Thus, if the quality premium is high enough, theexpected future revenue loss due to skimping on qual-ity will exceed the additional profits to be gained inthe current period.

Recently, a few tests of agency-based propositionsabout signals of product quality have begun to appear.For instance, Rao and Monroe (1991) recently devel-oped several propositions about the conditions underwhich suppliers are most likely to use price premiumsto signal product quality. For example, they suggestthat such premiums are likely to be used when

• the product's quality is relatively difficult to evaluate ob-jectively prior to purchase,

• prospective purchasers are relatively quality conscious,and

• the seller does not have a well-established reputation forquality.

In a subsequent attempt to test those hypotheses, Raoand Bergen (1991) collected perceptual data from 149senior purchasing executives in a variety of industries.Their evidence provides support for most of the prop-ositions.

Potential extensions. Though several different po-tential signals of quality have been examined, the workto date has only scratched the surface of the wide rangeof promotional tools—and combinations of tools—available to marketers. Moreover, those altemative toolsmay vary both in their value as signals and in the agencycosts to the manufacturers who use them. For exam-ple:

• Some evidence suggests that an extensive warranty canserve as a signal of a product's quality (Wiener 1985).From the manufacturer's perspective, however, suchwarranties can pose a potential hidden action problembecause they make it more vulnerable to possible abusesby buyers. Customers may be less likely to treat theproduct with proper care and may rely on the extensivewarranty to compensate for any resulting product failure,thus increasing the manufacturer's warranty costs. Cooperand Ross (1985), among others, employed a combina-tion hidden infonnation and hidden action model to ex-plore efficient ways to balance the signaling value ofwarranties against the potential costs of customers' hid-den actions.


• Similar cost-benefit tradeoffs might arise with a varietyof other potential signals of product quality, such as var-ious types of consumer promotions and joint branding(i.e., the promotion of brand-name component parts oringredients, as in Diet Coke with NutraSweet).

Though tnost signalitig tnodels are static, the itn-pact of a giveti signal of product quality tnay changeover titne as potential buyers gain access to additionalkinds and atnounts of infomiation. Hence, an obviousextension is to exatnine changes in the efficiency ofa given signal over time. As a first step in this direc-tion, Bagwell and Riordan (1991, p. 224) used a va-riety of hidden information models to examine the factthat "the rank-order correlation between price andquality is declining over time . . . [that] firms signalhigh-quality new products with prices that are abovefull-information, profit-maximizing prices . . . [andthat] . . . over time, as infonnation about the productdiffuses, this price distortion lessens or vanishes en-tirely." Thus, future research might examine how price-quality relationships change as the amount of varioustypes of infonnation available to customers changesover time (e.g., the publication of a Consumer Re-ports evaluation, the cumulative effects of advertis-ing, etc.).

Though most of the work on signaling concemsmanufacturers' efforts to influence the current behav-ior of potential customers toward current product of-ferings, signals also can be aimed at shaping futurebehavior of potential customers as well as other au-diences. Agency theory provides a framework for im-proving our understanding of the circumstances inwhich such signals are most likely to be employed andtheir impact on the behavior of various groups. Forinstance:

• A manufacturer might preannounce the future introduc-tion of a new product in an attempt to persuade potentialbuyers to postpone purchase decisions until it is avail-able or to increase their predisposition to buy it. How-ever, such action risks retaliation by competitors. A re-cent study by Eliashberg and Robertson (1988) foundmanagers are most likely to preannounce new productswhen they perceive (1) their firms to be small and with-out market dominance in the product category, (2) thecompetitive situation to be relatively noncombative, and(3) the new product to involve substantial switching costs,and therefore an extensive decision process, for potentialcustomers. To date, however, little is known about theimpact of such preannouncements on customers' sub-sequent purchase decisions (Farquhar and Pratkanis 1986),firms' sales performance, or the reactions of their com-petitors.

• Similar questions might be examined for preannounce-ment signals of other marketing actions, such as forth-coming price or distribution changes.

• Signals of a firm's planned future actions may be aimedat infiuencing the behavior of other audiences in additionto customers and competitors, such as members of thesalesforce or the distribution channel. Therefore, future

research might focus on the audience-specific rationaleand impact of such signals.

Agency theory provides a useful framework forexamining the efficiency of various market signals,but it is somewhat constrained by its economic foun-dations and its implicit assumptions about the "eco-nomic rationality" of customers and other audiences.For instance, the theory assumes that the essential valueof advertising is in providing functional informationto the customer and largely ignores other psycholog-ical outcomes, such as perceptions of a brand's imageand positioning. One contribution marketers mightmake, then, is to incorporate social and psychologicalfactors that infiuence the interpretation of and behav-ioral responses to different signals by customers,competitors, and other audiences. For instance:

• Heil (1988) offers some evidence that a receiver's eval-uation and interpretation of a given signal is infiuencedby such factors as (1) the signal's clarity, (2) its com-patibility with related signals, (3) the credibility of thesender, and (4) the perceived commitment behind it (e.g.,building a new plant in support of a new product prean-nouncement). However, other factors affecting the en-coding of different types of signals, the attention paid tothem, and the responses they provoke among differentaudiences have received little attention.

• More specifically, though agency theory suggests thatinvestments for large-scale production are a signal of highproduct quality, such investments might also reduce con-sumers' perceptions of the product's uniqueness and so-cial prestige, particularly if the product is an expensivedurable good (Jacobson and Aaker 1987).

Operationalization and measurement. Researchersworking on signals in a marketing context will haveto wrestle with many of the same problems in oper-ationalizing and measuring underlying constructs thatresearchers face in studying salesforces and channelrelationships. An additional problem arises in this area,however, because of the fact that "objective" mea-sures of such factors as product quality and price pre-miums are needed as benchmarks for evaluating thevalidity and efficiency of various signals. For exam-ple, though Wiener (1985) and others have used rat-ings from Consumer Reports as an objective measureof quality, considerable debate continues about thedefinition and measurement of the construct (e.g.,Holbrook and Corfman 1985).

In view of such measurement difficulties inherentin studying signaling phenomena with secondary dataor field surveys, laboratory experimentation may pro-vide a useful altemative. Such research could buildon the methods and measures developed within themarketing literature on the price-quality relationship(e.g., Rao and Monroe 1988, 1989) and in relevantresearch in experimental economics (e.g.. Lynch etal. 1986; Palfrey and Romer 1986).


Unexplored Agency Applicationsin Marketing

In addition to the many opportunities to extend andempirically test agency applications in topic areas thathave already been addressed, entire topics within mar-keting are amenable to agency analysis but have re-ceived little or no attention. Promising areas for futureapplications include aspects of intemational market-ing, industrial buying behavior, and issues involvingadvertising agencies.'^

International MarketingBecause cross-cultural differences magnify the prob-lems of uncertainty, asymmetric information, andmonitoring, efficient agency relationships can be evenmore difficult to achieve in multinational markets thanin domestic markets. To date, however, agency-basedresearch has examined the relative efficiency of onlytwo of the many organizational mechanisms firms mightemploy for entering foreign markets—licensing anddirect investment (Horstmann and Markusen 1987;Senbet and Taylor 1986).

Licensing enables the manufacturer to share therisk inherent in an unfamiliar foreign environment andto take advantage of the licensee's greater informationabout local conditions. However, the informationasymmetries inherent in licensing arrangements anddifficulties in monitoring a licensee's behavior mayenable a licensee to free ride on the licensor's repu-tation by letting product quality slip and thereby low-ering costs and increasing profits. Under such con-ditions, a manufacturer may need to offer additionalincentives to the licensee, or engage in extensivemonitoring activities, to maintain the quality of thelicensed brand.

In contrast, though direct investment involvesgreater upfront costs and exposes the manufacturer togreater risk, it also provides greater control over thequality of output. In addition, it can help increase lo-cal customers' information about the firm's productand provide them assurances of product quality (Sen-bet and Taylor 1986).

Given that only two foreign market entry modeshave been examined within an agency framework, anobvious direction for future development is to inves-tigate a broader range of entry options—such as theuse of independent export or import agents—and theconditions and contractual variables that affect their

"Agency theory might be applied to many other more narrowlyfoeused marketing issues. For instance, the hidden information modelcould help answer questions about how industrial purchasers choosesuppliers, how not-for-profit organizations can efficiently signal po-tential donors about the quality of their programs during fund-raisingefforts, and how donors might efficiently screen the many organiza-tions seeking support. Unfortunately, space limitations preclude a dis-cussion of such issues.

relative efficiency. Analyses of other forms of "stra-tegic alliances" between multinational firms, such asjoint R&D and product development agreements, arealso needed.

Industrial Buying Behavior—ReciprocityIndustrial purchasing often involves obtaining largequantities of technically complex products from a smallnumber of altemative suppliers. These task charac-teristics increase the uncertainty and risk for the buyerand enhance the potential for hidden information andhidden action problems in relationships with sup-pliers. Hence, industrial buyer behavior is a fertile fieldfor analysis from an agency theory perspective.

One example of a behavior pattem seen in indus-trial purchasing that might be examined with agencytheory is reciprocity, whereby a firm gives preferenceto a supplier who is also a customer for the firm'sproducts. Marketers offer several explanations for thisphenomenon (e.g., Moriarty 1983), but reciprocitymight be seen simply as an attempt to attenuate po-tential agency problems. It may be difficiilt for a buyerto determine whether a supplier is expending a properamount of effort to control and maintain the qualityof the products and services. The establishment of traderelations with a supplier helps align the interests ofthe two parties because the supplier then becomes acustomer for the purchaser's output. This approach maybe an efficient organizational response if it reduces thepurchaser's screening and monitoring costs.

Advertising Agency-Client RelationshipsRelationships between firms and their advertisingagencies face a variety of agency problems. Concemscited in the literature include excessive agency tum-over (Hagopian 1985), disagreements over objectives,the development of confiicting accounts by agencies(Wackman, Salmon, and Salmon 1987), and ineffi-cient incentive and control systems (Calantone andDrury 1979).

The agency tumover problem is one good exampleof an issue that might be usefully addressed withagency concepts. At least a part of this problem canbe traced to inadequate screening and selection pro-cedures, which lead clients to select agencies that latertum out to be unsatisfactory. It also raises a closelyrelated question about the efficacy of various signalsof agency quality.

The agency compensation problem is anotherissue that is ripe for agency analysis. Difficulty inobjectively measuring the outcomes of advertisingcampaigns has forced clients to rely heavily onbehavior-based methods for controlling their agencies.The traditional advertising commission stmcture, forinstance, ties agency compensation largely to mediapurchases. However, this system does little to control


the creative quality of the ads placed in the media andit exacerbates goal conflicts by making it in the agen-cy's self-interest to argue for larger advertising bud-gets, even when such budgets might not be strategi-cally justified.

Consequently, firms have recently adopted a va-riety of monitoring schemes for better evaluation oftheir agencies' actions and they have sought new com-pensation methods to align agency incentives betterwith their own objectives (Calantone and Dmry 1979).For example, DDB Needham has announced an out-come-based commission plan that "asks clients to re-ward the agency with payments equalling as much asa 20 percent commission when sales exceed presetgoals, and to penalize it with sharply reduced com-pensation—or none at all—if goals aren't met" (Levinand Lafayette 1990). Such a compensation plan wouldshift some risk from the client to the agency, but thefact that sales volume is affected by many other vari-ables besides advertising makes it difficult to measureobjectively the outcomes of an agency's performance.Thus, the proposed plan might be an inappropriate typeof control mechanism—a concem expressed by sev-eral of Needham's competitors.

In any case, the efficiency of Needham's pro-posal, and of most other attempts to resolve the manyongoing problems involved in advertising agency-clientrelationships, has yet to be examined. The issues inthis area afford many opportunities to apply agencytheory.

Conclusions and ImplicationsOur review of agency applications suggests that agencytheory can contribute much to our understanding of awide range of marketing issues. Throughout our ar-ticle, we suggest specific areas in which the theorymight profitably be applied or extended. However, onegeneral principle seems to underlie all of those sug-gestions: agency theory is likely to prove most usefulfor examining situations characterized by factors uniqueto the theory—factors that make contracting with andcontrolling the performance of agents especially dif-ficult. Hence, the theory might be used most produc-tively to examine situations involving (1) substantialgoal confiict between a principal and its agents, (2)sufficient envirotimental uncertainty to trigger the risk-sharing implications of the theory, (3) substantial in-formation asymmetries, and/or (4) difficulty in eval-uating performance (e.g., creative, team-oriented, orlong-term marketing activities).

The Theory's LimitationsOur review also suggests that agency theory has somelimitations as a vehicle for improving our understand-ing of marketing phenomena. Some of those limita-

tions are inherent in the scope and stmcture of thetheory itself, whereas others may be more a reflectionof the early stage of its application within the mar-keting discipline.

Agency theory's origin in economics is a sourceof conceptual strength as well as a cause of some ofits inherent limitations. As Hirsch, Michaels, andFriedman (1987) argue, economic theories have theadvantage of carefully developed assumptions andlogical, intemally consistent propositions. In addition,they incorporate variables and concepts that have beengiven relatively little attention in other behavioral sci-ences, such as the notions of risk preferences, the be-havioral consequences of risk sharing, and informa-tion asymmetries found in agency theory.

On the other side, however, Hirsch and his col-leagues argue that economics is dominated by a singleparadigm, price theory, and a single view of humannature, self-interest. This narrowness of focus aids theintemal consistency of economic theories, but it alsolimits their usefulness for fully understanding manymarketing phenomena, phenomena that are often af-fected by a variety of human motives and complexwebs of contingency variables. Because agency the-ory's narrow focus and rigorous stmcture make it dif-ficult to analyze such complexity, "pure" applicationsof the theory in marketing have had to center on onlythe simplest cases or to assume away many of thevariables that complicate the decisions faced by mar-keting managers. Consequently, it is not surprising thatmany potential extensions and new applications ofagency theory in marketing call for examinations ofmore complex situations and the incorporation of abroader range of variables.

Another shortcoming of past agency theory appli-cations is the paucity of rigorous testing. Though theevidence is encouraging, it is insufficient to resolvequestions about the theory's external validity or thegeneralizability of its propositions. Also, some resultsmust be viewed with caution because of problems inthe operationalization and measurement of key con-stmcts. Those problems are exacerbated in some stud-ies because they examine agency propositions at in-appropriate levels, testing individual-level models withcorporate- or industry-level variables. Some of the morecommon measurement and analysis problems identi-fied throughout our review are summarized in Table3, together with some suggestions for altemative ap-proaches that might be used in future agency research.

Marketing ImplicationsThough agency theory's shortcomings present prob-lems, they can also be viewed as attractive opportu-nities for marketing researchers. Not only are theremany ways in which marketers might contribute to thefurther development of the theory, but the actions that


TABLE 3Selected Measurement and Analysis Problems in Agency Research

Problem Possible Alternative Approaches

Level of AnalysisUse of corporate- or SBU-level data to test individual-

level models of salesperson behavior (e.g.,Eisenhardt 1985; John and Weitz 1988)

Use of industry-level data to test company-level hy-potheses (e.g., Norton 1988)

Time FrameReliance on static models and cross-sectional studies

to examine agency relationships that likely changeover time because of changing situational factorsand incomplete contracting

Operationalizationand Measurementof Key Constructs

Aggregate economic measures of environmental un-certainty, such as historical variance in a firm's fi-nancial performance, may not adequately capturethe psychological dimensions of the construct froman individual agent's perspective

Lottery techniques for measuring agents' risk prefer-ences, where respondents are asked to choose be-tween risky alternatives and their certainty equiva-lents (e.g., Oliver and Weitz 1991), can have lowreliability and validity

Some argue that business-level data are most "realis-tic" because most businesses employ a single com-pensation program for all salespeople (e.g., John andWeitz 1988).

Examination of individual differences in prefererices forand responses to different compensation plans is moreconsistent with conceptual focus of agency theory:• Evaluate alternative plans offered with a single

company (e.g., Lai, Outland, and Staelin 1990).• Use choice of employer and employer's current plan

as proxy for employee's preferences.• Explicitly measure individual's preferences for alter-

native plans, satisfaction with current plan, currentmotivation level, etc. (e.g., Oliver and Weitz 1991).

Relate differences in channel structures and/or controlmechanisms to differences across firms in a single in-dustry (e.g., Banerji and Simon 1991).

Examine the impact of a broader range of environmen-tal contingency variables by examining differences inchannels or contracts employed by individual firmsoperating in a variety of different industries and/orproduct categories (e.g., Bergen and John 1991).

Develop multi-item or rating scales of perceived prod-uct quality, supplier reputation, price premium, etc.(e.g., Rao and Bergen 1991).

Develop hypotheses based on dynamic agency relation-ships (e.g., Klein and Leffler 1981; Klein and Murphy1988; Rao and Monroe 1991).

Evaluate changes in channel structures or relationshipsover two or more discrete time periods (e.g., Banerjiand Simon 1991).

Longitudinal examinations of interactions between in-centives offered by principals and behavioral re-sponses of agents would be the ideal, though per-haps not always practical, design for field studies ofagency problems.

Laboratory experiments with multiple trials can be usedto stimulate dynamic agency relationships withincontrolled settings.

Subjective measures of the uncertainty perceived by in-dividual agents, such as the perceived uncertainty inthe linkage between individual effort and job perfor-mance, might be used in conjunction with more ob-jective financial measures (e.g., Oliver and WeiU 1991).

Risk preferences might be inferred by allowing agentsto self-select from a menu of alternative contractswithin a single environmental setting (i.e., level of en-vironmental uncertainty; e.g., Coughlan and Sen 1989).

Variations in risk preferences might be manipulated inlaboratory settings by employing different transfor-mation functions (e.g.. Berg et al. 1986).


TABLE 3 (continued)

Problem Possible Alternative Approaches

Measuring an agent's reservation utility by examiningwage rates for alternative jobs inside a firm or incompeting firms may fail to capture the full utilitya given agent may attain from an alternative job (e.g.,John and Weitz 1988)

Many of the proxy variables used to estimate the costsof monitoring the actions of agents assume a directrelationship between physical proximity and easeof monitoring (e.g., Brickley and Dark 1987; Norton1988); electronic data collection and communica-tion technologies make this assumption question-able

More subjective measures of "intention to quit" and"attractiveness of changing jobs" might be adaptedfrom the industrial psychology literature on job turn-over as a means of estimating the perceived relativeutility an agent ascribes to a given relationship. Also,multiple measures might be used (e.g., Coughlan andNarasimhan, forthcoming).

More direct, though subjective, scales might be devel-oped for measuring the difficulty and costs a principalperceives to be associated with monitoring agents'actions in a given situation (e.g., Eisenhardt 1985).

most need to be taken also happen to fit well withmarketers' distinctive competencies. Thus, a marriagebetween agency theory and marketing offers somepromise for wedded bliss—or at least for a mutuallybeneficial union.

The marketing discipline has a long and produc-tive tradition of creatively borrowing, adapting, andsynthesizing constructs from a variety of social sci-ences to obtain multiple perspectives and richerframeworks for understanding complex marketingphenomena. Given that agency theory provides onlya partial view of the world, there are many opportu-nities to meld agency constructs with those adaptedfrom other theories—and from our own empiricallyacquired wisdom—to achieve a more thorough un-derstanding of complicated marketing relationships.

Such synthesis is likely to necessitate sacrificing someof agency theory's intemal consistency and rigor, butimprovements in extemal validity and explanatorycompleteness should make such sacrifices worth-while.

Similarly, marketers have much experience incoping with the complexities of doing research withreal-world consumers and organizations. Given thenature of many of the shortcomings of current agency-based research and the strong traditions within themarketing discipline for careful measurement, surveyresearch, and experimentation, our greatest contri-bution to the further development of agency theorymay involve finding more appropriate ways to testcurrent propositions within meaningful marketingsettings.

REFERENCESAkerloff, George (1970), "The Market for 'Lemons': Quality

and the Market Mechanism," Quarterly Journal of Eco-nomics, 84 (August), 488-500.

Amershi, Amin H. and John Hughes (1989), "Multiple Sig-nals, Statistical Sufficiency, and Pareto Orderings of BestAgency Contracts," Rand Journal of Economics, 20 (Spring),102-12.

Anderson, Erin (1985), "The Salesperson as Outside Agent orEmployee: A Transaction Cost Analysis," Marketing Sci-ence, 4 (Summer), 234—54.

and Richard Oliver (1987), "Perspective on Be-havior-Based Versus Outcome-Based Salesforce ControlSystems," Journal of Marketing, 51 (October), 76-88.

Arrow, Kenneth J. (1974), Essays on the Theory of Risk Bear-ing. Amsterdam: North Holland Publishing Company.

(1985), "The Economics of Agency," in Principalsand Agents: The Structure of Business, J. Pratt and R.Zeckhauser, eds. Boston: Harvard Business School Press,37-51.

Bagwell, Kyle and Michael H. Riordan (1991), "High and De-clining Prices Signal Product Quality," American Eco-nomic Review, 81 (March), 224-39.

Banerji, Shumeet and Carol Simon (1991), "Franchising vs.Ownership: A Contracting Explanation," mimeo. GraduateSchool of Business, University of Chicago.

Basu, Amiya K., Lai Rajiv, V. Srinivasan, and RichardStaelin (1985), "Salesforce Compensation Plans: An AgencyTheoretic Framework," Marketing Science, 4 (Fall), 267-91.

Berg, Joyce, Lane Daley, John Dickhaut, and John O'Brien(1986), "Controlling Preferences for Units of ExperimentalExchange," Quarterly Journal of Economics, 101 (May),281-306.

Bergen, Mark (1990), "Two Essays on Vertical ArrangementsBetween Manufacturers and Retailers: Vertical Restraintsand Cooperative Advertising," unpublished PhD disserta-tion. University of Minnesota.


and George John (1991), "Understanding the Par-ticipation Rate Decision in Cooperative Advertising Plans,"Working Paper #87, Marketing Technical Report and Re-print Series, Graduate School of Business, University ofChicago.

Berkowitz, Leonard and Edward Donnerstein (1982), "Exter-nal Validity Is More Than Skin Deep," American Psy-chologist, 37 (3), 245-57.

Bemheim, B. Douglas and M. Whinston (1985), "CommonMarketing Agency as a Device for Facilitating Collusion,"Rand Journal of Economics, 16 (Summer), 269-81.

Brickley, James A. and Frederick Dark (1987), "The Choiceof Organizational Form: The Case of Franchising," Journalof Financial Economics, 18 (June), 401-20.

Calantone. Roger J. and Donald H. Drury (1979), "Advertis-ing Agency Compensation: A Model for Incentive andControl," Management Science, 25 (July), 632-42.

Cooper. Russell and Thomas W. Ross (1985), "Product War-ranties and Double Moral Hazard," Rand Journal of Eco-nomics, 16 (Spring), 103-113.

Coughlan, Anne T. (1985), "Competition and Cooperation inMarketing Channel Choice: Theory and Application," Mar-keting Science, 4 (2), 110-29.

(1988), "Pricing and the Role of Information inMarkets." in Issues in Pricing, T. M. Devinney, ed.Lexington, MA: Lexington Books, 59-62.

and Chakravarthi Narasimhan (forthcoming), "AnEmpirical Analysis of Salesforce Compensation Plans.Journal of Business.

and Ronald M. Schmidt (1985), "Executive Com-pensation. Management Tumover, and Firm Performance:An Empirical Investigation," Journal of Accounting andEconomics, 7 (April), 43-66.

and Subrata K. Sen (1989) "Salesforce Compen-sation: Theory and Managerial Implications." MarketingScience, 8 (4), 324-42.

Cron. William L. and John W. Slocum, Jr. (1986), "The In-fluence of Career Stages on Salespeople's Job Attitudes,Work Perceptions, and Performance." Journal of Market-ing Research, 23 (May). 119-29.

Davis, Scott (1991). "The Role of Price as an Indicatorof Product Quality in Monopolistic Markets," mimeo,Washington University at St. Louis.

Day, George S. and Saul Klein (1987). "Cooperative Behaviorin Vertical Markets: The Influence of Transaction Costs andCompetitive Strategies." in Review of Marketing, M.Houston, ed. Chicago: American Marketing Association.

Devinney, Timothy M. (1988). "Price, Advertising, andScale as Information-Relevation Mechanisms in ProductMarkets," in Issues in Pricing. T. M. Devinney. ed.Lexington, MA: Lexington Books, 63-82.

Donaldson. Lex (1990) "The Ethereal Hand: OrganizationalEconomics and Management Theory," Academy of Man-agement Review, 15 (July), 369-81.

Dutta. Shantanu, Mark Bergen, and George John (1991), "TheGovemance of Exclusive Territories When Dealers CanBootleg," Working Paper #85, Marketing Technical Re-print Series, Graduate School of Business, University ofChicago.

Dwyer, F. Robert and Orville C. Walker, Jr. (1981), "Bar-gaining in an Asymmetrical Power Structure," Journal ofMarketing, 45 (Winter), 104-15.

Dyer, James S. and Rakesh K. Sarin (1982), "Relative RiskAversion," Management Science, 28 (August), 875-86.

Eisenhardt, Kathleen M. (1985), "Control: Organizationaland Economic Approaches," Management Science, 31(February), 134-49.

(1989), "Agency Theory: An Assessment andReview," Academy of Management Review, 14 (January),57-74.

Eliashberg, Jehoshua, Stephen A. LaTour, Arvind Rangaswamy.and Louis W. Stem (1986), "Assessing the Predictive Ac-curacy of Two Utility-Based Theories in a Marketing Chan-nel Negotiation Context," Journal of Marketing Research,23 (May), 101-10.

and Thomas S. Robertson (1988), "New ProductPreannouncing Behavior: A Market Signaling Study,"Journal of Marketing Research, 25 (August). 282-92.

Fama, Eugene F. (1980). "Agency Problems and the Theoryof the Firm," Journal of Political Economy, 88 (April),288-307.

and Michael C. Jensen (1983a), "Separation otOwnership and Control," Journal of Law and Economics,26 (June), 301-25.

and (1983b), "Agency Problems andResidual Claims," Joumal of Law and Economics, 26 (June),327-49.

Farquhar, Peter H., and Anthony R. Pratkanis (1986), "Mar-ket Phantoms and Consumer Choice," presented at Mar-keting Science Conference. Dallas, TX (March 15).

Ferrall, Christopher (1991). "Promotion and Incentives inPartnerships: Theory and Evidence," Institute for Eco-nomic Research Working Paper #808, Queens University.

Green, J. and Nancy Stokey (1983). "A Comparison of Tour-naments and Contracts." Journal of Political Economy, 91(June), 349-64.

Grossman, Sanford and Oliver Hart (1983), "An Analysis ofthe Principal-Agent Problem," Econometrica, 51 (Janu-ary), 7—45.

L_ and (1986), "The Costs and Benefits ofOwnership: A Theory of Vertical and Lateral Integration,"Journal of Political Economy, 94 (August), 691-719.

Hagopian, J. (1985), "AAAA Chairman Hagopian TargetsAgency-Client Relationship." AAAA Newsletter (July-Au-gust).

Harris, Milton and Arthur Raviv (1979), "Optimal IncentiveContracts With Imperfect Information," Journal of Eco-nomic Theory, 20 (April). 231-59.

Hart, Oliver and Bengt Holstrom (1987), "The Theory of Con-tracts," in Advances in Economic Theory, Fifth World Con-gress, Truman Bewley, ed. London: Cambridge UniversityPress.

Hauser, John R. and Glen L. Urban (1979), "An Assessmentof Attribute Importances and Consumer Utility Functions,"Journal of Consumer Research, 5 (March), 251-62.

Heide. Jan B. and George John (1988). "The Role of Depen-dence Balancing in Safeguarding Transaction-Specific As-sets in Conventional Channels." Journal of Marketing, 52(January), 20-35.

and (1992), "Do Norms Matter in Mar-keting Relationships?" Journal of Marketing, 56 (April),32-44.

Heil, Oliver (1988), "Explaining and Predicting CompetitiveReaction: A Market Signaling Approach," unpublisheddoctoral dissertation. University of Pennsylvania.

Hesterly, William S., Julia Liebeskind, and Todd R. Zenger(1990), "Organizational Economics: An Impending Revo-lution in Organization Theory?" Academy of ManagementReview, 15 (July). 402-20.

Hirsch, Paul, Stuart Michaels, and Ray Friedman (1987), " 'DirtyHands' Versus 'Clean Models': Is Sociology in Danger ofBeing Seduced by Economics?" Theory and Society, 16 (3),317-36.

Holbrook, Morris B. and Kim P. Corfman (1985), "Quality


and Value in the Consumption Experience: Phaedrus RidesAgain," in Perceived Quality, i. Jacoby and J. Olson, eds.Lexington MA: Lexington Books, 31-57.

Holstrom, Bengt (1979), "Moral Hazard and Observability,"Bell Journal of Economics, 10 (Spring), 74-91.

and Jean Tirole (1987), "The Theory of the Firm."in Handbook of Industrial Organization, RichardSchmalensee and Robert Willig, eds. New York: Springer-Verlag, 61-133.

Horstmann, I. and James Markusen (1987), "Licensing VersusDirect Investment: A Model of Intemalization by Multi-national Enterprise." Canadian Journal of Economics, 20(3), 464-81.

Hurwicz. Leonid (1972), "On Informationally DecentralizedSystems," in Decision and Organization: A Volume in Honorof Jacob Marschak, Roy Radner and C. B. MacGuire, eds.Amsterdam: North Holland Publishing Company. 297-336.

Jacobson, Robert and David A. Aaker (1987), "TTie StrategicRole of Product Quality." Journal of Marketing, 51 (Oc-tober), 31-44.

Jensen, Michael and W. Meckling (1976), "Theory of the Firm:Managerial Behavior, Agency Costs and Capital Struc-ture," Journal of Financial Economics, 3 (October). 305—60.

Jeuland, Abel and Steven Shugan (1983), "Managing ChannelProfits," Marketing Science, 2 (Summer). 239-72.

and (1988), "Rejoinder to'ManageringChannel Profits: Comment'," Marketing Science, 7 (Win-ter), 103-6.

John, George, Allen Weiss, and Bart Weitz (1987). "An Or-ganizational Coordination Model of Salesforce Compen-sation Plans: Theoretical Analysis and Empirical Test."Journal of Law, Economics, and Organization, 3 (Fall),373-96.

and Barton Weitz (1988). "Explaining Variation inSales Compensation Plans: Empirical Evidence for the Basuet al. Model," working paper.

and (1989), "An Empirical Investiga-tion of Factors Related to the Use of Salary Versus Incen-tive Compensation," Journal of Marketing Research, 26(February), 1-14.

Katz, Michael L. (1989), "Vertical Contractual Relations."in Handbook of Industrial Organization, RichardSchmalensee and Robert Willig, eds. New York: North-Holland Publishing Company, 655-721.

Kern, Richard (1988), "I.Q. Tests for Salesmen Make aComeback," Sales and Marketing Management, 140(April), 42-6.

Klein, Benjamin and Keith Leffler (1981), "The Role of Mar-ket Forces in Assuring Contractual Performance." Journalof Political Economy, 89 (August), 615-41.

and Kevin M. Murphy (1988), "Vertical Restraintsas Contract Enforcement Mechanisms," Journal of Law andEconomics, 31 (2), 265-97.

Kreps, David M. (1990), A Course in Microeconomic Theory.Princeton, NJ: The Princeton University Press.

Lai, Rajiv (1990), "Improving Channel Coordination ThroughFranchising," Marketing Science 9 (4). 299-318.

, Donald Outland, and Richard Staelin (1990),"Salesforce Compensation Plans: An Empirical Test of theAgency Theory Framework," working paper. GraduateSchool of Business. Stanford University.

and V. Srinivasan (1989), "Salesforce Compen-

sation Plans in Environments With Asymmetric Informa-tion," Marketing Science, 5 (Summer), 179-98.

Lambert, R. (1983). "Long Term Contracting and Moral Haz-ard," Bell Journal of Economics, 14 (2), 441-52.

Lazear, Edward and Sherwin Rosen (1981), "Rank-OrderTournaments: On Optimum Labor Contracts." Journal ofPolitical Economy, 89 (October). 841-64.

Levin. Gary and Jon Lafayette (1990), "Guarantee Panned: D.D. B. Needham's Plan Called Too Risky," Advertising Age(May 7). 64.

Lynch, M., Ross M. Miller. Charles R. Plott, and RussellPorter (1986). "Product Quality, Consumer Examinationand 'Lemons' in Experimental Markets, in Empirical Ap-proaches to Consumer Protection Economics, P. M.Ippolito and D. T. Scheffman, eds. Washington, DC: Fed-eral Trade Commission. 251-306.

MacCrimmon, Kenneth R. and Donald A. Wehrung (1985),"A Portfolio of Risk Measures." Theory and Decision, 19(July), 1-29.

and (1986), Taking Risks: The Man-agement of Uncertainty. New York: The Free Press.

MacDonald, G. (1984). "New Directions in the EconomicTheory of Agency," Canadian Journal of Economics, 17(August), 415-40.

Mathewson. George and Ralph Winter (1984), "An EconomicTheory of Vertical Restraints," Rand Journal of Econom-ics, 15 (Spring), 27-38.

and (1985). "The Economics of Fran-chise Contracts," Journal of Law and Economics, 2 (Oc-tober), 503-26.

and (1986), "The Economics of Ver-

sation Plans: A Dynamic Perspective," working paper.Graduate School of Business, Stanford University.

and Richard Staelin (1986), "Salesforce Compen-

tical Restraints in Distribution." New Developments in theAnalysis of Market Structures, Joseph E. Stiglitz and G.Frank Mathewson, eds. Cambridge, MA: MIT Press.

McGuire, Timothy W. and Richard Staelin (1983). "An In-dustry Equilibrium Analysis of Downstream Vertical Inte-gration," Marketing Science, 2 (2), 161-91.

Milgrom. Paul and J. Roberts (1982), "Prediction, Reputationand Entry Deterrence," Journal of Economic Theory, 27(August). 280-312.

Mirlees, James (1976), "The Optimal Structure of Authorityand Incentives Within an Organization," Bell Journal ofEconomics, 7 (Spring), 105-31.

Mitnick, Barry M. (1987), "The Theory of Agency and Or-ganizational Analysis: Annual Meeting of the American Po-litical Science Association," revised April 1987.

Moncrief, William L.. in (1986). "Selling Activities and SalesPosition Taxonomies for Industrial Salesforces," Journal ofMarketing Research, 23 (August), 261-70.

Moorthy, K. Sridhar (1987), "Managing Channel Profits:Comment," Marketing Science, 6 (Fall), 375-9.

Moriarty, Rowland T. (1983). Industrial Buying Behavior.Boston: Lexington Books.

Murphy. Kevin J. (1985), "Corporate Performance and Man-agerial Remuneration: An Empirical Analysis," Journal ofAccounting and Economics, 7 (April), 11—42.

Murray. Tom (1991), "Team Selling: What's the Incentive?"Sales and Marketing Management, 143 (June), 89-92.

Nelson, Philip (1974), "Advertising as Information," Journalof Political Economy, 82 (July/August), 729-54.

Norton, Seth W. (1988), "An Empirical Look at Franchisingas an Organizational Form." Journal of Business, 61 (2),197-218.

Oliver. Richard L. and Barton A. Weitz (1991), "The Effectsof Risk preference. Uncertainty, and Incentive Compen-sation on Salesperson Motivation," Report Number 91-104.Cambridge. MA: Marketing Science Institute (February).


Palfrey, Thomas R. and Thomas Romer (1986). "An Experi-mental Study of Warranty Coverage and Dispute Resolu-tion in Competitive Markets," in Empirical Approaches toConsumer Protection Economics, P. M. Ippolito and D. T.Scheffman, eds. Washington. DC: Federal Trade Commis-sion, 307-72.

Perrow, Charles (1986), Complex Organizations. New York:Random House, Inc.

Picard, Paul (1987), "On the Design of Incentive Schemes Un-der Moral Hazard and Adverse Selection," yoMrna/ of Pub-lic Economics, 33 (August), 305-32.

Pratt, John W. (1964), "Risk Aversion in the Small and theLarge Firm," Econometrica, 32 (12). 122-36.

and Richard Zeckhauser, eds. (1985), Principalsand Agents: The Structure of Business. Boston: HarvardBusiness School Press.

Radner, Roy (1981), "Monitoring Cooperative Agreements ina Repeated Principal-Agent Relationship." Econometrica,49 (5), 1127-48.

(1985). "Repeated Principal-Agent Games WithDiscounting," Econometrica, 53 (5), 1173-98.

Rao, Akshay and Mark Bergen (1991), "Price Premiums as aQuality Enforcement Mechanism: Empirical Evidence."Working Paper #88. Marketing Technical Report and Re-print Series. Graduate School of Business. University ofChicago.

and Kent B. Monroe (1988), "The Moderating Ef-fect of Prior Knowledge on Cue Utilization in ProductEvaluation," Journal of Consumer Research, 15 (Septem-ber). 253-64.

and (1989), "The Effect of Price. Brand

elusive Territories in Producers' Competition," workingpaper, Princeton University.

and Jean Tirole (1986a), "The Logic of Vertical

Name, and Store Name on Buyers' Perceptions of ProductQuality: An Integrative Review." Journal of Marketing Re-search, 26 (August), 351-2.

and (1991), "The Causes and Conse-quences of Price Premiums: Linking Product Quality andProfitability," working paper, Carlson School of Manage-ment, University of Minnesota.

Rey, Patrick and Joseph E. Stiglitz (1988), "The Role of Ex-

Restraints," American Economic Review, 76 (5), 921-39.and (1986b). "Vertical Restraints From

a Principal-Agent Viewpoint." in Marketing Channels, L.Pellegrini and S. K. Reddy, eds. Lexington. MA: Lexing-ton Books. 1-30.

Ross. Steven (1973), "The Economic Theory of Agency: ThePrincipal's Problem." American Economic Review, 63 (2),134-9.

Rothschild, Michael and Joseph E. Stiglitz (1976), "Equilib-rium in Competitive Insurance Markets: An Essay on theEconomics of Imperfect Infonnation," Quarterly Journalof Economics, 90 (4), 629-49.

Senbet, Lemma W. and William Taylor (1986), "Direct For-eign Investment With Imperfect Information," working pa-per. University of Wisconsin-Madison.

Spence. Michael (1973), "Job Market Signalling," QuarterlyJournal of Economics, 87 (3). 355-74.

Teas, R. Kenneth and James C. McElroy (1986). "Causal At-tributions and Expectancy Estimates: A Framework for Un-derstanding the Dynamics of Salesforce Motivation," Jour-nal of Marketing, 50 (January). 75-86.

Tirole, Jean (1989), The Theory of Industrial Organization.Boston: The MIT Press.

Wackman, Daniel B.. C. T. Salmon, and C. Salmon (1987)^"Developing an Advertising Agency-Client Relationship,"Journal of Advertising Research, 26 (January), 21-8.

White, Harrison (1985), "Agency as Control." in Principalsand Agents: The Structure of Business, J. Pratt and R. J.Zeckhauser, eds. Boston: Harvard Business School Press,187-212.

Wiener, Joshua L. (1985), "Are Warranties Accurate Signalsof Product Reliability?" Journal of Consumer Research, 12(September). 245-50.

Williamson, Oliver (1988). "Corporate Finance and CorporateGovemance," Journal of Finance, 43 (3), 567-91.

Zusman, P. and Michael Etgar (1981), "The Marketing Chan-nel as an Equilibrium Set of Contracts," Management Sci-ence, 27 (March), 284-302.

Reprinl No. JMS63100


Agency Relationships in Marketing: A Review of the ... · PDF file10.10.2016 · Mark Bergen, Shantanu Dutta, & Orville C. Walker, Jr. Agency Relationships in Marketing: A...

Documents