Advertising agency compensation: An agency theory explanation
Spake, Deborah F
Advertising agency compensation may change dramatically in coming years as advertisers put pressure on agencies to reduce commissions and tie compensation to performance resulting from advertising campaigns. Drawing on agency theory from the economics discipline, the authors develop and test several hypotheses to address the advertising agency compensation decision. Their study provides the first comprehensive look at the prevalence of outcome-based compensation tied to performance and other compensation systems currently used among U.S. advertisers.
Much has been written about advertising agency compensation and the changes currently occurring in the advertising industry. One industry observer comments: “The way agencies are compensated for their purchases of advertising promises to change dramatically” (Zbar 1994, p. S-24). Calls have been made for agencies to “work harder to customize their compensation structures and find new ways of redefining their role in brand building” (Selinger 1995, p. 3). As a result, advertising agencies have witnessed a steady erosion of client budgets, indicative of client opinion that full-service agencies are too costly for the impact they deliver (LaBahn 1996). In effect, though studies by the Association of National Advertisers (Gleason 1996) show that a majority of advertisers (59% in 1995 vs. 61% in 1992) have been paying their agencies in commissions, more than three quarters of the agencies were paid reduced-rate commissions in 1995 compared with less than half in 1992. Further, in addition to lower agency compensation advertisers are demanding greater agency accountability.
The pressure on advertising agencies to enhance their performance and deliver value is the result of several events: the Total Quality Management revolution of the 1980s to improve quality, now extended to include supplier performance; the increased pressure on marketing budgets resulting in calls to justify expenditures; and a competitive environment wherein market prices are stable or declining, forcing profit growth to come from other sources such as volume generated by advertising. For the advertiser, the message is clear. Advertising, like any other activity, must be managed better to obtain superior results at a lower cost, Agency compensation is critical to that objective, but what guidance can theory offer to the advertising manager?
What characteristics of the client-agency relationship and the advertising environment drive the compensation decision? What theory can inform the discussion? Drawing on the organizational governance literature, Ellis and Johnson (1993) proposed that agency theory would contribute to our understanding of the advertising agency compensation issue. The purpose of our article is to apply an agency theory framework to advertising agency compensation in an initial attempt to answer the preceding questions and test the propositions set forth by Ellis and Johnson (1993). Agency theory is concerned with designing incentives into contracts that make them selfenforcing. In other words, only the self-interest of the contracting parties will enforce the contract, and a fortiori enable them to achieve their objectives. We address the prevalence of outcome-based compensation tied to advertising impact and other compensation systems currently being used by U.S. advertisers. Our study is the first in which agency theory has been used to explore outcome-based advertising agency compensation.
Agency Theory
Agency theory, along with transaction cost economics and the theory of vertical restraints, grew out of a need by economists to explain firms’ departure from profit maximization, and should be seen as complementary to, rather than as an alternative for, the other theories. Agency theory has been applied to several issues in marketing such as salesforce management, distribution channel coordination and control, and consumer promotions (see Bergen, Dutta, and Walker 1992, for a listing of agency theory applications in marketing). Though a call for agency theory to address issues in advertising agency/client relationships has been made (Bergen, Dutta, and Walker 1992) and a framework for applying agency theory in an advertiser/agency context has been developed (Ellis and Johnson 1993), no study to date has addressed the issues empirically. In spite of a “positive branch,” agency theory has evolved in a rather technical and formal modeling style which has tended to obscure the fairly straightforward constructs that form the essential core of agency theory. In this section, we describe the conceptual framework of agency theory and examine its basic terms and constructs.
Agency theory centers around two parties, a principal (e.g., an advertiser) and an agent (e.g., advertising agency) who collaborate to achieve certain outcomes (e.g., sales or profits). The principal knows that some effort requiring a certain expertise (e.g., whatever is necessary to formulate an appropriate message that is creatively executed and delivered with the right media) is crucial to achieving the outcomes, but does not know what amount or what type of effort is needed. Further, the principal does not know whether the agent is capable of providing the effort, and realizes that the effort is correlated imperfectly with outcomes. A characteristic of agency relationships is that the agent, while exercising effort, makes decisions or undertakes actions on behalf of the principal.
In agency theory, both the principal and the agent are assumed to be motivated by self-interest and to act rationally (Bergen, Dutta, and Walker 1992). Conflicts arise in the agency relationship because the principal and the agent have different goals and risk preferences that lead them to prefer different courses of action. The goal of both the principal and the agent is maximum utility, each party acting in his or her own best interest (Ellis and Johnson 1993). However, agency theory does not maximize the joint utility of the principal and the agent, but maximizes the principal’s utility, with the agent’s utility serving as one of many constraints. Utility for the principal is related directly to profits, whereas utility for the agent is related inversely to effort and directly to compensation. Hence, the principal is assumed to strive for maximum profits, whereas the agent seeks maximum compensation at minimum effort (Coughlan and Sen 1989; Farley 1964).
Agency theory addresses two broad problems faced by the principal. The first, which arises before the principal enters a relationship with the agent, is the hidden characteristic problem. It essentially involves whether or not the agent has the characteristics (e.g., skills, capacity, or corporate culture) that the principal requires, and what means the principal has for discovering those characteristics. It often arises when an advertiser wants to launch a new advertising campaign, but is without an agency or is in doubt about the current agency’s abilities to create and execute an effective campaign. To solve the problem, the advertiser may brief a few agencies on its needs and invite presentations. The agency judged as best matching the needs of the advertiser will get the account. The second problem, which arises after the principal has entered a relationship with an agent, is the hidden action problem. Here, the advertiser provides incentives to agents to motivate them to behave in a way that would be optimal for the principal. For example, once an agency gets the account the advertiser must ensure that the agency has the incentive to do the best job it is capable of doing. Arrow (1984) proposed the terms “hidden information” and “hidden action” as substitutes for the terms “adverse selection” and “moral hazard,” which were in widespread use and had been borrowed from insurance terminology. Campbell (1995) suggested the term “hidden characteristic” as being more precise than “hidden information,” because information lies at the heart of both “hidden action” and “hidden characteristic” models.
In terms of the hidden action model, the focus of agency theory is the design of the most efficient contract to govern the relationship. That is a challenging task for several reasons. First, the two parties generally have different goals. Perfect concordance of goals is possible (e.g,, the pilot of an aircraft is just as determined as the passengers to arrive safely at the destination), but rare. Second, the parties’ attitudes toward risk often are dissimilar. For example, an employee generally prefers security, whereas an employer or entrepreneur has a greater tolerance for risk. Third, each party has some information that the other does not have-there is an asymmetry of information (Bergen, Dutta, and Walker 1992). For example, a wholesaler knows more about the territory he or she covers than the manufacturer, who in turn knows more about the product than the wholesaler. Fourth, environmental uncertainty is bound to be present; external forces beyond the agent’s control may influence the outcomes (Basu et al. 1985; Levinthal 1988; Umanath, Ray, and Campbell 1996). In the advertiser/advertising agency situation, other promotional, marketing mix, and environmental variables influence sales in addition to advertising. Compounding the problem is the fact that some of the influences are stochastic, making it very difficult to ascertain the actual impact of advertising on sales. Fifth, the principal cannot completely monitor the agent’s effort, at least not without considerable cost outlays for collecting information (Holmstrom 1979). The key constructs from the single hidden action model of agency theory are depicted in Figure 1.
In summary, the aim of the hidden action model of agency theory is to design contracts that align the objectives of the principal and the agent in such a way that the agent is motivated to perform actions that optimize the principal’s profits as well as the agent’s compensation. The model shows how the agent’s self-interest can be harnessed so that it does not become pathological (Campbell 1995, p. 4). In practical terms, the principal must ensure that the compensation package provides enough incentives for the agents that are to be hired. Note that we do not regard advertising per se as an outcome. If advertising determines sales-that is, if advertising and sales are perfectly correlated-then it does not matter whether advertising or sales is regarded as an outcome. In that case a compensation package could be based on either sales or advertising, because (aside from a change of scale) the packages would be equivalent in all respects, especially in their incentive effect on effort. If there is uncertainty, an imperfect correlation between advertising and sales, a compensation package based on advertising will differ from one based on sales in terms of the incentive effect on effort.
Agency relationships are context specific, as a given individual or firm can shift from the role of principal to that of agent across different contexts. Sometimes in a given context an individual or firm can be both principal and agent. In the context of product warranties, for example, a double hidden action model is used (Cooper and Ross 1985). In that model, a consumer is regarded as a principal seeking a reliable product, and the firm as the agent supplying a level of product reliability through its selection of product quality that is unobservable by the consumer. The firm is regarded also as a principal seeking to pay minimum warranty repairs, and the consumer as the agent who can affect the reliability of the product, and hence the amount of repairs paid, through his or her investment in product care and maintenance, which are unobservable to the firm, In our study, we treated the advertiser as the principal and the advertising agency as the agent and use the single hidden action model of agency theory.
Advertising Agency Compensation
Advertising agencies typically are compensated through commission or fees charged for services rendered on the advertiser’s behalf. The commission system in advertising is a tradition of modern business. Arising in the nineteenth century and based on media expenditures (Fox 1984), the standard agency commission has long been 15% of media billings (Adweek 1995). In agency theory, commissions are typically part of an outcome-based contract (Eisenhardt 1989). Clearly, that is not the case with the traditional commission system used by advertising agencies. Advertising agency commissions are based on media purchased, not on sales or campaign performance per se.
Fees are a recent development in advertising agency compensation, emerging in the 1970s-most notably when the Ogilvy agency accepted the Shell account on that basis (Fox 1984). Fee systems are much more complex than commissions, and typically are based on hourly rates, project fees, or monthly retainers, alone or in combination (Lamons 1994). Fee-based compensation arose because some advertisers chose to purchase their media directly or grew uncomfortable with the conflict of interest in the traditional agency commission system whereby agencies might be prone to recommend more media use, instead of other marketing approaches, to increase their compensation. Also, agencies with business-to-business clients preferred fees, as they tended to be light media spenders but demanded a lot in sales support materials. Fee-based systems may have flaws, however. Some advertisers using a fee-based system fear they are overbilled for service (Adweek 1996) or receive a poor value for the time charged.
Regardless of the method used, friction about compensation can and does arise between advertisers and their agencies (Seggev 1992). As pointed out by Ellis and Johnson (1993, p. 76), “One potential problem with both the commission and the fee system is the fact that neither fs based directly on the performance of the advertising campaign….” That problem is an important issue in advertising agency/clients relationships, as reflected in the literature (Hotz, Ryans, and Shanklin 1982; Seggev 1992), and complicates already strained agency/client relationships (Lamons 1992).
For clarification, we define outcome-based contracts and arrangements as agreements between the principal and agent that tie agent compensation to performance, mutually identified and measured by agent and principal. The performance measures may include, but are not limited to, increased sales, results of copy testing and/or tracking studies, and increases in market share. We define behavior-based contracts and arrangements as agreements between the principal and agent that tie agent compensation to actions taken by the agent in the service of the principal. The behaviors may include, but are not limited to, performing market research, copy testing, and negotiating media buys. Behavior-based contracts encompass fee-based contracts.
Such classification of compensation in agency theory is fairly widely used. For an application to channel relationships, see Celly and Frazier (1996). A salary for a salesperson is an example of behavior-based compensation, given that the salesperson must demonstrate certain minimum activities (e.g., showing up for work) to continue earning the salary. A salesperson may also earn his or her entire compensation in commissions on sales, an example of outcome-based compensation. In many instances, salespersons earn both salary and commissions, thus receiving both behavior- and outcome-based compensation.
Growing dissatisfaction among clients with current compensation methods has created a call for more outcome-based compensation for advertising agencies (Calantone and Drury 1979; Dagnoli 1990; Levin 1990). Examples of agencies using outcome-based compensation include Warwick Baker O’Neill, which bases its compensation for its Driver’s Mart Worldwide account on vehicle sales (Gleason 1996; Gleason and Halliday 1996), and Foote, Cone & Belding, which ties compensation for its Mazda account to increased retail sales increases (Johnson 1996). In response to demand for outcome-based compensation based on agency performance, DDB Needham Worldwide announced a “guaranteed results” incentive-compensation plan whereby it would give clients rebates if performance goals were not reached, but would receive bonuses in addition to its base compensation rate if goals were exceeded (Levin 1990).
The recent movement toward linking agency compensation to clients’ results scares some ad executives and intrigues others (Gleason 1996; Selinger 1995). Fear of such linkage is understandable because it shifts greater risk onto the advertising agency and results in reduced compensation if goals are not met. Agencies resist the use of outcome-based agreements unless they believe the campaign goals set by the client are achievable; even so, the practice is thought to be slowly gaining momentum (Gleason 1995). Ad executives who are intrigued view the practice as a way to increase value and accountability (Gleason 1996) by rewarding the contributions of the advertising agency without penalizing it for mistakes made by others. Currently, the two most common methods of tracking performance are measuring market share gains and measuring advertising effectiveness through copy testing and/or tracking studies.
Despite the call for more outcome-based compensation (Gleason 1995), the prevalence of such compensation systems among advertising agencies in the United States is uncertain. One study by the Association of National Advertisers suggests that 11% of advertisers have added an incentive clause to their compensation plans since 1992, and that many are turning from the traditional 15% commission to reduced-rate commissions and labor fees (Gleason 1995; Lucas 1995; Selinger 1995). Yet, outcome-based compensation has not been directly addressed.
On the other hand, the persistence of the traditional contract has not gone unnoticed. Aaker and Myers (1975, p. 8) observed that though the advertising industry has changed considerably since the last century, the standard method of compensation has persisted and efforts to replace it have been largely unsuccessful. Batra, Meyers, and Aaker (1996, p. 13) state that the commission method of compensation is still the method used most often. Gleason (1995), reporting on a study by the Association of National Advertisers, gives the figure for commission-based contracts as 59% in 1995, down from 61% in 1992.
Advertising agency compensation schemes can be placed on a bipolar scale with the most behaviorbased ones on one end and the most outcome-based ones on the other as shown in Figure 2. Traditional agency commission is at the center, or zero point, of the scale-it cannot be classified as either behaviorbased or outcome-based, but has some aspects of both systems and much of neither. Commissions in most business contexts (e.g., salesforce compensation) are typically viewed as an outcome-based form of compensation. However, in the context of advertising agency remuneration, the commission is not based on outcomes of the advertising campaign, but on the amount of media placed, which is an action of the agency. Because the basis for remuneration is behavior, the agency commission is like behavior-based compensation; however, because the agency’s remuneration does not vary with the amount of time or effort spent on the action of media placement, it is unlike behavior-based compensation. Further, it is unlike outcome-based compensation because, for a particular advertising campaign, agency compensation does not vary with outcomes, yet it is like outcome-based compensation in the sense that over time for several advertising campaigns compensation will tend to vary with outcomes if advertising budgets are set as a percentage of sales. Well known theoretical reasons support the percentage-of-sales advertising budgeting rule (Schmalensee 1972) and empirical evidence demonstrates its widespread use (Lancaster and Stern 1983). It is the advertiser who fixes the agency’s compensation by deciding how much to spend on media, but the advertiser is concerned with maximizing his or her profits, not directly with agency remuneration. The agency’s role is to provide the creative content of the advertising that makes reaching the advertiser’s goal possible. Over time, it is in the agency’s best interest to help the advertiser achieve his or her goals, because unsuccessful advertisers tend to reduce advertising budgets and hence agency compensation. In that sense, traditional commission has aspects of behavior-based and outcome-based contracts. It is more outcome-based than fees only, because compensation tends to vary over time with outcomes and not with effort, and it is more behaviorbased than compensation tied to measurable outcomes, because the basis for payment is media placement, an agency action, and not advertising outcomes such as sales or brand awareness.
Similarly, a combination of fees and traditional agency commission can be viewed as having more behavior-based aspects than payment by traditional commission alone. Finally, compensation that is tied to measurable outcomes is the least behavior-based method of compensation. We used the bipolar scale in developing and testing our hypotheses.
Hypotheses
In agency theory, the principal can choose to compensate the agent on the basis of (1) the agent’s behavior, (2) realized outcomes, or (3) a combination of the two. In the advertiser/advertising agency situation, we treat both fees and traditional agency commission based on placed media as examples of behavior-based compensation because they are based on the activities, and therefore the effort, of the agency. Outcome-based compensation, in contrast, is based on the outcomes of the advertising effort (e.g., sales or brand awareness) and is calculated accordingly.
Eisenhardt (1985, 1989) and other agency theory researchers have investigated the conditions under which behavior-based contracts are more appropriate than outcome-based ones. They have shown that the efficiency of behavior-based contracts relative to outcome-based ones is likely to increase as a function of (1) the difficulty of measuring the effect of the agent’s effort on outcomes, (2) a decrease in goal conflict between the principal and the agent, (3) an increase (decrease) in risk aversion on the part of the agent (principal), (4) a decrease in monitoring costs, and (5) an increase in environmental uncertainty.
Applying those conditions to an advertising client/ agency context, Ellis and Johnson (1993) developed five propositions, which we adapted into four hypotheses.
Ease of Measuring Agency Effect on Outcomes
The first hypothesis addresses how the difficulty of measuring the effect of an agent’s effort on outcomes influences the nature of the contract between the principal and the agent. If the effect of an agent’s effort is difficult to evaluate, the problems and costs associated with administering outcome-based contracts will be great and behavior-based contracts will tend to be preferred. Because many nonadvertising variables can influence sales, objectively measuring the outcome of agency performance is often difficult. Hence, most clients have relied on behavior-based compensation methods such as commission or fees (Bergen, Dutta, and Walker 1992), which do not require the agency to track campaign effectiveness and which encourage larger advertising budgets (Ellis and Johnson 1993). Firms have also sought greater monitoring of agency actions to align agency incentives more closely to their own objectives (Calatone and Drury 1979). Hence, our first hypothesis suggests that agency compensation would be more outcome-based when the effect of the advertising campaign on outcomes is easy to measure.
H1: As the ease of measuring the effect of an advertising activity on an advertising outcome, or linking a specific advertising outcome to an advertising activity, increases, the likelihood of outcome-based (vs. commission-only) compensation increases and the likelihood of behavior-based (vs. commission-only) compensation decreases.
Our other three hypotheses concern the relationship between the client and its advertising agency.
Client/Agency Goal Conflict
Formal agency theory provides the notion of a “first best” contract that attains the profit maximization result by assuming the agent and the principal are one and the same. Agency theory can be viewed as an attempt to explain why the “first best” result is unattainable. If principal and agent have similar goals, which they would have if they were the same, the latter is likely to behave in accordance with the wishes of the former, regardless of the incentives built into the contract. Goal congruity between a client and its agency results in lower costs of monitoring the agency’s activities (Ellis and Johnson 1993) because of high levels of trust. Business partners with congruous values have been shown to be more committed to their relationship and more likely to trust their partner (Morgan and Hunt 1994). Further, clients would want to be particularly sensitive to the degree to which the agency shares their “vision.” When goal conflict is present, the risk that the principal’s wishes will not be carried out by the agent is greater (Levinthal 1988), increasing the need to monitor the agent or tie compensation to the desired outcome.
H2: As the degree of goal conflict between client and agency increases, the likelihood of outcome-based (vs. commission-only) compensation increases and the likelihood of behavior-based (vs. commission-only) compensation decreases.
Risk Aversion of Advertiser
A risk-averse client would undoubtedly want to shift financial risk to its agency; however, it will be successful only to the extent that it has significant negotiating power in contractual arrangements. An agency must determine during such negotiations both the potential importance of the client’s business and the likelihood that a strong competitor will agree to outcome-based compensation (Ellis and Johnson 1993). Risk, defined as rapid change in an organization, has been addressed in the management literature by Stroh et al. (1996), who found that managers in organizations with high risk aversion received a higher proportion of their compensation in the form of variable pay (i.e., outcome-based). Advertisers characterized as risk averse are thought to be likely to participate in agency activities as a method of monitoring agency actions. Johnson and Laczniak (1991) have suggested that agency/client relationships will in fact improve given increased participation and result in more satisfied advertisers.
H3: The more risk-averse the client, the more likely the use of outcome-based (vs. commission-only) compensation and the less likely the use of behavior-based (vs. commission-only) compensation.
Duration of Advertising Client/Agency Relationship
Our final hypothesis addresses the effect of an established working relationship. Hotz, Ryans, and Shanklin (1982) have shown that longevity is an important dimension of the agency/client relationship. Though agency models are generally static, time can be regarded as a surrogate for the changing effect of several constructs in agency theory. Goal conflict, monitoring costs, uncertainty, and the difficulty of measuring the effect of the agent’s effort are likely to decrease over time. In longstanding relationships, reduced goal conflict and reduced monitoring costs favor behavior-based contracts. However, both decreasing uncertainty and decreasing difficulty of measuring the effect of the agents effort favor outcomebased contracts. Ellis and Johnson (1993), focusing on the decrease in goal conflict and cost of monitoring that is brought on by the passage of time, proposed that advertising agency compensation may become more behavior-based as the duration of the client/ agency relationship increases. Work by Stroh et al. (1996), Levinthal (1988), and Holmstrom (1979) tends to support that proposition. However, we argue, and agency theory suggests, that as relationships age, environmental uncertainty and the difficulty of determining the agent’s effect on outcomes also decrease. Such conditions would favor outcome-based contracts. Hence, our four variables would have offsetting effects over time, and a movement away from commission-based compensation toward either behaviorbased or outcome-based contracts would be observed.
H4:As the duration of the relationship between the client and agency increases, the likelihood of outcome-based and/or behavior-based (vs. commission-only) compensation increases.
Method
A national survey of client advertising managers in the United States was undertaken to ascertain their attitudes toward the current relationship with their advertising agency. Preliminary depth interviews were conducted with several agency executives to gain an understanding of the agency side of advertiser/ agency relationships and to assist in the development of the questionnaire. The agency executives interviewed represented agencies that differed greatly in annual billings and included ones specializing in either consumer or industrial advertising.
After questionnaire development, we conducted pretest interviews in person and by telephone with advertisers representing a range of industries (e.g., telecommunications, banking, and retailing). The pretest interviews provided timing estimates for questionnaire completion and feedback for minor changes to the questionnaire format. The questionnaire was constructed and mailed as recommended by Dillman (1978) in a booklet format.
A sample of 2000 advertisers was selected systematically from the 1996 Standard Directory of Advertisers on an nth -count basis (Babbie 1986). Advertisers that indicated using only an in-house advertising agency and those in countries outside the United States were excluded from the sample. As an incentive, a summary copy of the results was offered to respondents who enclosed a business card or provided a return address on the back of the postagepaid envelope.
Survey packets were sent to the advertising manager at the firm, if one was indicated in the Standard Directory of Advertisers, or otherwise to the brand manager. If neither was indicated in the directory, the survey was sent to the vice president of marketing. Addressees were told in the cover letter that the survey was intended for the person with the most interaction with the advertising agency for the brand they managed. That statement was necessary because large companies might have many brands and hence many brand managers, but only one brand manager/advertising manager was to be selected per firm. If a packet recipient was not the person who most often interacted with the advertising agency, he or she was instructed to forward the survey to that person. Such a key informant method is used often in marketing and advertising studies (e.g., Barclay 1991; King, Reid, and Morrison 1997; Sethuraman, Anderson, and Narus 1988) with the intent of interviewing the person best able to respond. If multiple agencies were used for a brand, the respondent was instructed to respond to the survey with reference to the primary advertising agency only.
Two weeks after the initial mailing, a reminder postcard was sent to each advertiser. Of the 2000 mailed, 37 surveys were undeliverable; 14 addressees indicated that company policy prohibited participation and 30 reported that they did not use an independent ad agency. Hence, the effective sample size was 1919. Usable responses were received from 349 respondents. The response rate (18.2%) was deemed acceptable because of the length of the questionnaire (14 pages) and is comparable to that of other surveys targeting the same population (LaBahn 1996).
Measures
The survey instrument was comprehensive and designed to address several advertiser/agency issues. The sections pertaining to agency theory and compensation elicited type of compensation, ease of measuring the effect of advertising, the importance of advertising, goal conflict, frequency of participation, expertise in direct marketing, and demographic data. A 7-point Likert-type scale ranging from “strongly disagree” (1) to “strongly agree” (7) was used to construct multi-item measures for ease of measurement, importance of advertising, and goal conflict. A 7point Likert-type scale ranging from “never” (1) to “always” (7) was used to construct a multi-item, formative measure to determine the frequency of participation with the agency on account-related tasks.
Method of Compensation. Respondents were asked to indicate which method(s) they use to compensate their advertising agency. Five responses were provided: (1) “traditional (percentage of media billings),” (2) “fees,” (3) “partly tied to measurable outcomes such as brand share, new users, repeat users, unaided awareness, sales volume, etc.,” (4) “a combination of (1) and (2),” and (5) “a combination of (1) and (3).” A sixth response, “some other method,” was provided and respondents were asked to elaborate on the method if that response was selected. Of 25 respondents who selected “some other method,” 21 indicated that their agency compensation was based on a combination of fees and measurable outcomes.
H1 holds that the easier it is to attribute advertising to its intended outcomes and measure the intensity of such a link, the greater is the likelihood that compensation schemes are based on those outcomes. We tested Hi by using several variables that were presumed to be good indicators of how easy it is to measure the effect of advertising. They included the advertiser’s perceptions of the ease of measuring the effect of advertising and perceptions of the importance of advertising, and the interaction between the two. They also included the advertiser’s evaluations of the agency’s expertise in direct marketing, whether the offering advertised was a service or a product, and whether the type of customer targeted was a business or a consumer. Those variables were chosen in the belief that the effects of advertising are relatively easier to link to agency effort in the context of direct marketing, the sale of tangible products, and marketing to consumers. The interaction between ease of measurement and importance of advertising was also thought to make it easier to attribute advertising outcomes to agency activities. The importance of advertising as perceived by the client should generally put pressure on the advertising agency to demonstrate the effectiveness of its advertising.
Ease of Measuring the Effect of Advertising. Respondents answered four items designed to measure the ease with which the effect of advertising could be ascertained for the product or service they personally managed. Specifically, the scale measured (1) the sales effect of advertising, (2) the communication effect of advertising, (3) success of the ad campaign, and (4) knowledge that advertising objectives had been accomplished.
Importance of Advertising. Four items were used to assess the respondents’ opinion of the importance of advertising for the product or service they personally managed: (1) “Advertising is one of the most important determinants of market success,” (2) “The firms that devote the most resources to advertising are most profitable,” (3) “Firms that do not advertise are not successful,” and (4) “In order to remain competitive, a firm must advertise.”
Other Variables. Expertise in direct marketing was a single-item measure with a 7-point Likert-type scale ranging from (1) “very low” to (7) “very high” in response to a question asking at what level the ad agency performs. Respondents were also asked to indicate the type of offering advertised (product or service) and the type of customer targeted (consumer or industrial).
We tested H2 by using a measure for goal conflict based on the advertiser’s perceptions that the client’s and agency’s general philosophical goals are congruent. Differences in goals can pertain to specific goals for a campaign or philosophical goals of the individuals or organizations. We defined goal conflict as differences in values and treatment of customers and employees, thus addressing philosophical goal conflict.
Goal Conflict. Advertising agencies and their clients have been shown to differ in their perceptions of each other’s ethical practices (Krugman and Ferrell 1981) and objectives (Hotz, Ryans, and Shanklin 1982). That is one context of goal conflict. Another context of goal conflict is more specific, involving the direction of a given campaign. We tested the first type of goal conflict, specifically philosophical goal conflict in terms of shared values (Morgan and Hunt 1994) between the advertiser and the ad agency. Respondents were asked the degree to which they and their agencies share values in general and in specific relation to treatment of customers and employees. The scale was adapted from Morgan and Hunt’s (1994) shared values measure.
We tested H3 by using a measure of the advertiser’s frequency of participation with the agency and use of multiple agencies. Frequency of participation was chosen because advertisers that are more risk averse were assumed to participate more frequently with their agency as a method of monitoring agency actions. The use of multiple agencies was chosen because clients that are more risk averse were assumed to use multiple agencies to diversify their risk.
Frequency of Participation with Agency. Respondents were asked to indicate the frequency with which they participated in the following activities with their ad agency: focus group observation, meetings, campaign development, selection of vendors, production of commercials, approval of ads, and telephone surveys. The scale was adapted from one developed by Mohr and Spekman (1994). It is formative because the construct “frequency of participation” is not reflected by the individual items as such-, rather, the more often advertisers participate in the activities, as denoted by the items, the more they can be said to participate frequently with their agency.
Use of Multiple Agencies. Respondents were also asked whether more than one advertising agency is used by the firm for the product or service the respondent personally manages. We tested H4 by using a self-reported measure of the number of years the advertiser had used the agency. The square of the relationship duration was used to capture possible nonlinear effects. See Figure 3 for a correspondence between the theoretical model/hypotheses and the empirical model.
Reliability and Validity. Cronbach’s alpha and confirmatory factor analysis were used to assess the measures. Cronbach’s alpha was .88 for ease of measuring advertising effects, .79 for importance of advertising, and .81 for goal conflict. Using LISREL 8 (Joreskog and Sorbom 1993), we created a measurement model to test the reliability and validity of each scale. All of the items converged exceedingly well on their respective constructs, the lowest t-value being 9.50. Significant tvalues meet the criterion for convergent validity (Anderson and Gerbing 1988). Composite reliability ranged from .713 (importance of advertising) to .833 (goal congruity). Those results and the construct reliability of each scale are reported in Table 1.
Discriminant validity was assessed by comparing the variance extracted estimate of each construct with phi squared as shown in the matrix at the bottom of Table 1. Discriminant validity was supported across all constructs. In each case, the variance extracted estimate exceeded phi squared Tornell and Larcker 1981), with the smallest variance extracted estimate being.385 for the importance of advertising and the largest being .630 for goal conflict. Additional support for discriminant validity is provided by examining the 95% confidence interval to determine whether the correlation plus or minus two standard deviations contains the value 1.00 (Anderson and Gerbing 1988). In no case did the confidence interval include that value.
The model provides an acceptable fit to the data (X^sup 2^= 112.46, d.f. = 41, p = .0; GFI = .94; CFI = .94; RMSEA = .072), as shown in Table 1. Overall, the constructs have convergent and discriminant validity. Neither confirmatory factor analysis nor Cronbach’s alpha was generated for the frequency of contact scale because it is a formative measure.
Results
The sample consisted of 349 respondents, most of whom held the title of advertising manager (42%), communications director (39%), or vice president (15%). The majority (69%) were 35 to 54 years of age. On average, they had held their current position for 5.5 years and controlled an advertising budget of $13 million. Approximately 22% of their advertising budget was spent in-house, with 38% using multiple advertising agencies for the product or service they personally managed (see Table 2). On average, the relationship with their primary agency had lasted 8.6 years.
Advertisers were asked how their agency was compensated for its services. Table 3 shows that most advertisers reported fees (37%) or a combination of fees and commission (40%) as the most frequent method of agency compensation. Less than I percent (.3%) reported compensation to be tied exclusively to measurable outcomes, and 7.1% used a combination of measurable outcomes and some other method. Because of the small number of advertisers that tied compensation exclusively to measurable outcomes, they were combined with ones that used a combination of measurable outcomes and some other method in our analysis. The traditional commission-only method of compensation was used by 15.9% of advertisers. Clearly, some form of behavior-based compensation, either fees or a combination of fees and commission, predominated the types of compensation systems employed, as it was used by 77% of advertisers.
Testing Hypotheses
Discussion
A common lament is that formal theories are more a way of explaining observations than the product of systematic and rigorous investigation. For example, Joskow (1975, p. 273) opined, “Somehow one gets the distinct feeling that the important messages are being carried by the informal theories, stories, and behavioral observations, and the formal theories are trotted out ex-post to demonstrate that some kind of formal apparatus can explain or incorporate some of what is actually being observed.” Rigorously testing a theory, confronting it with empirical facts, and letting the chips fall where they may results in much richer insights and understanding of the problems that the theories were designed to explain. However, in the case of advertising agency compensation, it was unclear before our study what the facts were. The trade press had implied that outcome-based compensation was all the rage, a popular assumption that seemed to cry out for an agency theory explanation.
Our results provide support for three of the four hypotheses that we tested. Results for two of the three hypotheses suggest that in relation to commission-only compensation, outcome-based compensation would be used more when (HI) the outcome is easier to measure and to link to agency activities and (H3) the client is more risk averse. For those two hypotheses the support is partial, whereas H4 has full support, affirming that in relation to commission-only compensation, either behavior-based or outcome-based compensation increases with the duration of the relationship between client and the agency.
The results do not support H2, that in relation to commission-only compensation, there will be more outcome-based and fewer behavior-based contracts as the goal conflict in the advertiser/agency relationship increases. However, the goal conflict measure in our study was operationalized as goal congruity, which can be defined as a philosophical similarity of values and treatment of employees and customers. Hence, goal conflict specifically pertaining to an advertising campaign or other pressing operational concerns was not measured. A more “tactical” measure of goal conflict might have yielded different results.
We regard HI to be a key hypothesis from an agency theory perspective. Its support, even with the small number of advertisers (n = 25) that compensated their agency on the basis of measurable outcomes, is remarkable. However, we also had a result running counter to agency theory, suggesting that as advertising effects become easier to measure, compensation becomes more behavior-based. We interpret it to mean that because measuring advertising effects is neither cheap nor trivial, ease of measurement alone will not prompt clients and agencies to move toward outcome-based compensation. Advertising must also be an important element of the marketing mix.
Our findings fully support H4, but we stress that other theories-for example, diffusion theory, which states that imitation and bandwagon effects accentuate industry trends (Abrahamson and Rosenkopf, 1993) – can also explain the phenomenon. However, we emphasize that the hypothesis itself goes beyond Eisenhardt (1989) and Ellis and Johnson (1993). They restrict the effect of relationship duration to increasing behavior-based contracts by taking into account only the decrease in goal conflict and monitoring costs. However, uncertainty is likely to decrease, and measurement of the effect of the agent’s effort on outcomes is likely to become easier, favoring a trend toward outcome-based contracts. The result seems to suggest that the prudent approach for a client and an agency that do not know each other is to start with a standard commission-only contract, which contains elements of both behavior-based and outcome-based contracts. Then, as they get to know each other, and depending on their specific context, they can move toward either a behavior-based or outcome-based contract or one that has features of each.
Silk and Berndt (1993, p. 68) have documented scale economies in advertising agencies, which they believe are consistent with “the trend away from reliance by large advertisers on fixed commission rates as the basic method of compensating their agencies.” Agency theory and our empirical results can contribute to discussion of that trend in two ways. First, the scale economies argument explains only a trend within commission-only contracts from a fixed rate of commission, the traditional 15%, to a downward sliding scale rate that can vary from 13 to 8%. Second, scale economies do not explain why commission-only contracts are modified by fees and/or outcome-based clauses, or even replaced entirely by them.
From the Civil War to the turn of the last century, two key developments in advertising practice forever changed the industry. One was the advertising agency N. W. Ayer’s introduction of the “open contract” and the other was the provision of copy writing and other creative services by advertising agencies. Both of those developments, and especially the latter, can be explained by market forces and transaction cost economics. An agency theory explanation simply provides some additional insight to historical processes that were neither straightforward nor inexorably preordained.
Prior to the “open contract” or commission-only method of compensation, advertising agencies were essentially wholesalers or agents of the media, earning their compensation by buying low and selling high. From the perspective of the hidden action model, the advertiser knows what is paid to the advertising agency and what advertisements the media publish, but does not know what the advertising agency keeps for itself and what is pays the media. Through the “open contract” the terms of the sale of the media space become common knowledge, and are no longer a question of hidden action.
Before advertising agencies provided creative services, copy writing had been left to the advertiser on the logical assumption that he or she knew most about his or her own business. As Wood (1958, p. 231) makes clear, up to the turn of the century advertisers hired their own copywriters, who generally worked on a freelance basis. Even N. W. Ayer is quoted by Pope (1983, p. 138) as saying in 1893, “The writing of advertisements is not our business. It is only a small part of it, and is done by us only for regular customers.” Yet, by 1910 the provision of copy services had become a standard part of agency service. In terms of the hidden action model, it is necessary to view the advertising agency as the principal and the advertiser as the agent. The agency realizes that its efforts in media buying will be jeopardized in the long run if the message is not right. By providing creative services, the agency ensures greater control over the quality of the message and does not have to leave it to the actions of the advertiser.
Two interesting questions are why commissionbased contracts have persisted so long and why, when they are modified, it is with fees rather than with outcome-based clauses. The answer to the latter question, in terms of agency theory in our particular context, is that monitoring costs are lower than the cost of measuring the agent’s effect on outcomes. In other words, it is easier and cheaper for an agency to hire a cost accountant than it is for an advertiser to measure reliably the effects of advertising, and therefore it is cheaper and easier to move in the direction of fees. Besides, as Gleason (1995) says, “Some advertisers, after insisting on a performance-based compensation system, balk at full disclosure of the data used to measure performance.”
At the turn of the twentieth century the real question is whether agency theory can predict any likely developments in advertising agency compensation. On the basis of the theory and our empirical work, and despite the mavens of the trade press, outcomebased compensation seems likely to go only so far as sound, economic, and generally accepted methods of measuring advertising effectiveness are developed. Direct, database, and internet marketing along with the profusion of information technology undoubtedly will accelerate the process to some extent, but even so the measurement of advertising effects is not a trivial undertaking. It requires expertise and trust in that expertise-qualities that are associated with behavior-based modes of compensation.
Limitations and Conclusion
We applied the single hidden action model from agency theory to advertising agency compensation. However, two of its limitations must be mentioned. First, the framework implicitly assumes a long-term relationship and rational behavior within such a time horizon. Those assumptions may be too simplistic because sometimes agencies are fired if marketing objectives are not met, and account turnover is fairly rapid. In such cases, a short-term time horizon clearly is operating alongside the long-term one, and the agency’s behavior, especially its response to incentives, depends on what time horizon it is following. Second, there is a strong interaction between marketing variables, and the success of one depends on the performance of others. One can view the agency as the principal that delegates to the client the responsibility for product quality, distribution, and sales to ensure that advertising works. Therefore, a double rather than a single hidden action model may be the most realistic representation of the agency/client relationship, with the client and agency each being principal and agent for different activities.
Other limitations are related to some of the construct measures. We measured the method of agency compensation as a categorical variable. Researchers may want to consider asking advertisers what percentage each method of compensation comprises of the total compensation paid to the agency. Such an approach would further inform the stream of research. Our goal conflict measure was defined as a philosophical similarity of values and treatment of employees and customers. Researchers should consider other measures of goal conflict specifically related to advertising campaigns or marketing strategy.
Our study provides the first empirical support for agency theory in the context of advertising agency compensation. Given that advertisers are dissatisfied with the status quo of traditional commission compensation, more are likely to consider outcome-based compensation in the future. However, agency theory implies that such a trend will be contingent on improving methods for measuring advertising effects. The strength of agency theory for practitioners is that it puts the focus on a few key informational variables that are needed for the provision of incentives. As a consultant put it, “The key question is how to make agency compensation a tool to [motivate] the agency to accomplish specific marketing or advertising goals” (Gleason 1995, p. 44). Not bad for a definition of agency theory.
References
Aaker, David A. and John G. Myers (1975), Advertising Management, Englewood Cliffs, NJ: Prentice Hall.
Abrahamson, E. and L. Rosenkopf (1993), “Institutional and Competitive Bandwagons: Using Mathematical Modeling to Explore Innovation Diffusion,”Academy ofManagement Review, 18 (July), 487-517.
Adweek (1995), “Payment Plans,” 36 (November 13), 44. – (1996), “Check, Please,” 37 (November 4), 64-66.
Anderson, James C. and David W. Gerbing(1988), “Structural Equation Modeling in Practice: A Review and Recommended TwoStep Approach,” Psychological Bulletin, 103 (May), 411-425.
Arrow, K. J. (1985), “The Economics of Agency,” in Principals and Agents: The Structure ofBusiness, J. Pratt and R. Zeckhauser, eds., Boston: Harvard Business School Press, 37-51.
Babbie, Earl (1986), The Practice of Social Research, 4th ed., Belmont, CA: Wadsworth.
Barclay, Donald W. (1991), “Interdepartmental Conflict in Organizational Buying: The Impact of the Organizational Context,” Journal of Marketing Research, 28 (May), 145-159.
Basu, Amiya K., Rajiv Lai, V. Srinivasan, and Richard Staelin (1985), “Salesforce Compensation Plans: An Agency Theoretic Perspective,” Marketing Science, 4 (Fall), 267-291.
Batra, Rejeev, John G. Myers, and David A. Aaker (1996), Advertising Management, Upper Saddle River, NJ: Prentice Hall.
Bergen, Mark, Shantanu Dutta, and Orville C. Walker, Jr. (1992), “Agency Relationships in Marketing. A Review of the Implications and Applications of Agency and Related Theories,” Journal of Marketing, 56 (July), 1-24.
Calantone, Roger and Donald H. Drury (1979), “Advertising Agency Compensation: A Model for Incentive and Control,” Management Science, 25 (July), 632-642.
Campbell, Donald E. (1995), Incentives.- Motivation and the Economics ofInformation, New York: Cambridge University Press. Celly, Kirti Sawhney and Gary L. Frazier (1996), “Outcome-Based
and Behavior-Based Coordindination Efforts in Channel Relationships,”Journal of Marketing Research, 33 (May), 200-210@ Cooper, Russell and Thomas W. Ross (1985), “Product Warranties
and Double Moral Hazard,” Rand Journal of Economics, 16 (Spring), 103-113.
Coughlan, Anne T. and Subrata K. Sen (1989), “Salesforce Compensation: Theory and Managerial Implications,” Marketing Science, 8 (Fall), 324-342.
Dagnoli, Judann (1990), “Campbell Cuts Marketing Budget,” Advertising Age, 61 (July 23), 1, 48.
Dillman, Don A. (1978), Mail and Telephone Surveys: The Total Design Method, New York: Wiley.
Eisenhardt, Kathleen M. (1985), “Control: Organizational and Economic Approaches,” Management Science, 31 (February), 134-49. – (1989), “Agency Theory: An Assessment and Re
view,”Academy of Management Review, 14 (January), 57-74. Ellis, R. Susan and Lester W. Johnson (1993), “Observations: Agency Theory as a Framework for Advertising Agency Compensation Decisions,” Journal of Advertising Research, 33 (September/October), 75-80.
Farley, John U. (1964), “An Optimal Plan for Salesmen’s Compensation,”Journal of Marketing Research, 1 (May), 39-43.
Fornell, Claes, and David L. Larcker (1981), “Evaluating Structural Equation Models with Unobservable Variables and Measurement Error,” Journal of Marketing Research, 18 February), 39-50.
Fox, Stephen (1984), The Mirror Makers, New York: William Morrow. Gleason, Mark (1995), “Role of Performance Gains in Agency Pay,” Advertising Age, 66 (November 27), 3, 44.
– (1996), “Driver’s Mart Pacing Agency Pay to Sales,” Advertising Age, 67 (May 6), 3, 54.
– and Jean Halliday (1996), “Warwick Oks Used Car Pay-for-Sales Plan,” Advertising Age, 67 (May 27), 4.
Greene, William H. (1990), LIMDEP, New York: Econometric Software.
– (1993), Econometric Analysis, New York: Macmillan. Holmstrom, Bengt (1979), “Moral Hazard and Observability,” The
Rand Journal of Economics, 10 (Spring), 74-91.
Hotz, Mary R., John K. Ryans, Jr., and William L. Shanklin (1982), “Agency/Client Relationships as Seen by Influentials on Both Sides,”Journal of Advertising, 11 (1), 37-44.
Johnson, Bradley (1996), “Mazda Drives Ahead with FCB,” Advertising Age, 67 (February 5), 32,
Johnson, Jean L. and Russell N. Laczniak (1991), “Antecedents of Dissatisfaction in Advertiser-Agency Relationships: A Model of Decision-Making and Communication Patterns,” Current Issues and Research in Advertising, 13 (1), 45-59.
Joreskog, Karl and Dag Sarbom (1993), LISREL 8: A Guide to the Program and Applications, Chicago: SPSS.
Joskow, P., (1975), “Firm Decision-Making Process and Oligopoly Theory,” American Economic Review, Papers and Proceedings, 65, 270-279.
King, Karen Whitehill, Leonard N. Reid, and Margaret Morrison (1997), “Large-Agency Media Specialists’ Opinions on Newspaper Advertising for National Accounts,” Journal of Advertising, 26 (Summer), 1-17.
Krugman, Dean M. and 0. C. Ferrell (1981), -Me Organizational Ethic of Advertising: Corporate and Agency Views,” Journal of Advertising, 10 (1), 21-30.
LaBahn, Douglas W. (1996), “Advertiser Perceptions of Fair Compensation, Confidentiality, and Rapport,” Journal of Advertising Research, 36 (March/April), 28-38.
Lamons, Bob (1992), “Agency-Client Relationships Grow More Adversarial,” Marketing News, 26 (October 12), 13, 21. _@ (1994), “Fee Systems Determine Agency Perfor
mance,” Marketing News, 28 (July 4), 14.
Lancaster, Kent M. and Judith M. Stern (1983), “Computer Based Advertising Budgeting Practices of Leading U. S. Consumer Advertisers,”Journal of Advertising, 12 (4), 4-9.
Levin, Gary (1990), “DDB Expands Pay Plan,” Advertising Age, 61 (July 9), 42@
Levinthal, Daniel (1988), “A Survey of Agency Models of Organizations,” Journal of Economic Behavior and Organization, 9 (March), 153-185.
Lucas, Allison (1995), “The Changing Face of Compensation,” Sales & Marketing Management, 147 (July), 20.
Maddala, G., (1988), Introduction to Econometrics, New York: Macmillan.
Mohr, Jakki and Robert Spekman (1994), “Characteristics of Partnership Success: Partnership Attributes, Communication Behavior, and Conflict Resolution Techniques,” Strategic Management Journal, 15 (February), 135-152.
Morgan, Robert M. and Shelby D. Hunt (1994), “The CommitmentTrust Theory of Relationship Marketing,” Journal of Marketing, 58 (July), 20-38.
Pope, Daniel (1983), The Making of Modern Advertising, New York: Basic Books.
Schmalensee, Richard (1972), The Economics of Advertising, Amsterdam: North-Holland.
Seggev, Eli (1992), “Advertising Effectiveness Measurement for Contribution-Based Compensation,” Journal of Advertising Research, 32 (July/August), 73-78.
Selinger, Iris Cohen (1995), “Big Profits, Risks With Incentive Fees,” Advertising Age, 66 (May 15), 3.
Sethuraman, Rajagopalan, James C. Anderson, and James A. Narus (1988), “Partnership Advantage and Its Determinants in Distributor and Manufacturer Working Relationships,” Journal of Business Research, 17 (December), 327-347,
Silk, Alvin J. and Ernst R. Berndt (1993), “Scale and Scope of Advertising Agency Costs,” Marketing Science, 12 (Winter), 53-72.
Stroh, Linda tC, Jeanne M. Brett, Joseph P. Baumann, and Anne H. Reilly (1996), “Agency Theory and Variable Pay Compensation Strategies,” Academy ofManagement Journal,39 (June), 751-767.
Umanath, Narayan S., Manash R. Ray, and Terry L. Campbell (1992), “The Effect of Uncertainty and Information Asymmetry on the Structure of Compensation Contracts: A Test of Competing Models,” Management Science, 42 (June), 868-874.
Wood, James Playsted (1958), The Story of Advertising, New York: Ronald Press.
Zbar, Jeffery D. (1994), “Usable Framework Sought for Compensation Issue,” Advertising Age, 65 (July 25), S-24.
Deborah F. Spake (Ph.D., University of Alabama) is an Assistant Professor at the Department of Marketing, Haworth College of Business, Western Michigan University.
Giles D’Souza (Ph.D., University of Texas at Dallas) is an Associate Professor at the Department of Management and Marketing, Culverhouse College of Commerce and Business Administration, University of Alabama.
Robert M. Morgan (Ph.D., Texas Tech University) is an Associate Professor and Reese Phifer Fellow of Marketing at the Department of Management and Marketing, Culverhouse College of Commerce and Business Adminstration, University of Alabama.
Tammy Neal Crutchfield (Ph.D., University of Alabama) is an Assistant Professor at the Stetson School of Business and Economics, Mercer University.
Copyright American Academy of Advertising Fall 1999
Provided by ProQuest Information and Learning Company. All rights Reserved
