Journal of Retailing 87 (4, 2011) 549562

Reducing the Size of Internal Hierarchy: The Case of Multi-Unit Franchising

Rupinder Jindal
Department of Marketing & Entrepreneurship, C.T. Bauer College of Business, University of Houston, Houston, TX 77204-6021, United States

Abstract Successive layers of supervisorsubordinate relationships in organizations often distort information, increase monitoring costs, and lead to a cumulative loss of control. This paper discusses how some organizations can reduce their internal hierarchy by slicing it into two components and substituting the supervisorsubordinate relationship with an independent contract. This substitution allows the organization to shift its lower-level hierarchy to the contractors. These contractors are less likely to indulge in moral hazard, which can further reduce the size of hierarchy required. The paper examines this theory in the domain of multi-unit franchising and tests the hypotheses with a longitudinal data set. 2011 New York University. Published by Elsevier Inc. All rights reserved.
Keywords: Franchising; Channels of distribution; Chain organizations; Inter-organizational relationships; Agency theory; Internal hierarchies

Introduction [In an organization,] layers slow everything down. Take decision-making. The more layers, the more people who have to thump their rubber stamp. The more PowerPoint presentations to be made to bosses and bosses bosses before the rubber stamp. Or take communicating change. Layers make that processhard enough as it islike that childrens whispering game, telephone. Every time a piece of information passes through a person, it morphs a little. Layers do that, too, adding spin, interpretation, and buzz with every telling. Jack Welch, ex-CEO General Electric Company (Welch and Welch 2007, p. 96) Research has long recognized the problems that arise from multiple layers in a hierarchy, that is, a chain of supervisorsubordinate relationships. Subordinates typically satisfy only a fraction of the directives of their supervisors, so successive layers of these relationships often distort information, increase monitoring costs, slow down decision making, and lead to a cumulative loss of control (Aghion and Tirole 1997; Alchian and Demsetz 1972; Calvo and Wellisz 1978; Holmstrom and Tirole 1989; Radner 1992; Williamson 1967). In marketing organizations such as retail chains, sales forces, and franchise systems, these problems may undermine not only organizational

Tel.: +1 713 743 5848; fax: +1 713 743 4572. E-mail address: rjindal@uh.edu

efciency but also the organizations efforts to achieve consistency in brand presentation and customer service. Despite the recognition of the need, most theories of the rm fail to address these problems of hierarchy (Garrouste and Saussier 2005; Gibbons 2005). This paper discusses how slicing a hierarchy into two components and substituting the supervisorsubordinate relationships at the sliced level with independent contracts can help some organizations reduce these problems. This substitution allows an organization to shift its lower-level hierarchy to multiple contractors who can manage these mini-hierarchies more effectively because of their relatively smaller sizes. Such an organization consists of two hierarchical components joined together with an independent contract. Furthermore, due to the compensation design, the contractors monitoring such minihierarchies are less likely to indulge in moral hazard, which further reduces the size of internal hierarchy the organization needs to monitor the monitors. We examine this theory in the domain of franchising. Franchisees usually have incentives to free ride on joint inputs because of spillover effects (Brickley 1999). If left uncontrolled, free riding can damage the value of the brand and hurt both the franchisees and the franchisor. To deter franchisees from free riding, a franchisor can employ monitors to observe their activities. However, because these employee monitors usually draw a substantial portion of their compensation as xed salary, they have incentives to shirk and need to be monitored as well (Alchian and Demsetz 1972; Arrow 1985). Rather than controlling free riding among franchisees, such multiple layers of monitors disseminate moral hazard across the entire hierarchy

0022-4359/$ see front matter 2011 New York University. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.jretai.2011.07.003

550

R. Jindal / Journal of Retailing 87 (4, 2011) 549562

of the organization, leading to distortion of information, slower decision making, and a cumulative loss of control. Hence, franchising provides an appropriate setting to study ways to reduce the size of internal hierarchies. Using data on 3,436 observations from 713 franchisors between 1995 and 2005, we show that one way franchisors can reduce the size of their hierarchies is by adopting multi-unit franchising, that is, allocating multiple outlets to their franchisees. Specically, we show that the franchisors are more likely to adopt multi-unit franchising when the conditions that require a large monitoring hierarchy escalate. By using multiunit franchising a franchisor slices its monitoring hierarchy and substitutes lower-level employee monitors such as district managers with independent entities known as multi-unit franchisees. Multi-unit franchisees compensation depends entirely on outlet performance, which makes them more motivated monitors compared with employee-monitors. Because of their stronger motivation they need relatively less monitoring than employeemonitors, which further reduces the hierarchy the franchisor needs. This research contributes to marketing theory in two ways. First, it contributes to research in franchising by identifying four conditionsdensity of outlets, advertising fee, outlet size, and dispersion of marketsthat inuence a franchisors decision to adopt multi-unit franchising. Although scholars have highlighted the need to incorporate multi-unit franchising into franchise theory to answer the complexity surrounding modern franchising (Kaufmann and Dant 1996, p. 356), this seemingly counterintuitive phenomenon remains a key question of interest (Blair and Lafontaine 2005). Second, this paper contributes to broader research in interorganizational relationships by demonstrating a governance structure that can help some organizations reduce the size of their internal hierarchies. The phenomenon of multi-unit franchising Franchising is a widely used channel for distributing goods and services when the intangibles play a critical role in delivery to the end customer.1 Such goods and services include restaurants, food and nonfood products, lodging, automotive products and services, laundry and dry cleaning services, printing and graphics services, and maintenance services (Blair and Lafontaine 2005). A rm using own outlets to distribute such goods and services needs dedicated outlet managers to run the outlet operations and be the custodian of its brand. However, because the outlet managers are employees they usually have incentives to shirk (Arrow 1985).

To control shirking, the rm enters into a contract with a legally independent entity, a franchisee. The franchisee owns an outlet and operates it with the rms (now franchisor) managerial and operational guidance. The franchisee typically pays a onetime franchise fee at the beginning of the contract period and royalties for the duration of the contract (Kaufmann and Dant 2001). The franchisee is the residual claimant and thus has less incentive to shirk compared with an employee outlet manager (Brickley and Dark 1987; Lafontaine 1992; Rubin 1978).2 Conventional wisdom suggests that to be a dedicated owneroperator, a franchisee be physically present to operate the outlet and monitor the outlet employees (Blair and Lafontaine 2005). Thus, franchisors should allocate only one outlet to each franchisee to own and operate; this phenomenon is known as single-unit franchising. Yet empirical evidence shows that more than half of the franchisors in North America allocate multiple outlets to their franchisees (Bond 2005; Johnson 2006; Wadsworth 2002); this phenomenon is known as multi-unit franchising. Multi-unit franchising has several potential drawbacks. First, because a multi-unit franchisee cannot operate all the outlets on its own, it must hire outlet managers. Because these outlet managers are employees (of multi-unit franchisee), multiunit franchising reintroduces the risk of shirking (Kaufmann and Dant 1996). Second, a franchisor may lose some degree of control over the outlets to multi-unit franchisees (Lowell 2007). Third, a franchisor may lose bargaining power relative to multi-unit franchisees (Kalnins and Lafontaine 2004). Fourth, ownership of multiple outlets within a market may reduce a multi-unit franchisees incentive to behave as aggressively as single-unit franchisees might (McKee, Lovejoy, and Moran 2004). A question thus arises: If multi-unit franchising has so many potential drawbacks, why is it still common? Economic theory suggests that only efcient organizational forms survive in a competitive environment (Anderson 1988; Hirshleifer 1985; Stigler 1958). Literature in population ecology echoes this perspective: organizations that match environmental needs are positively selected and survive, whereas others either fail or change to match those environmental needs (Aldrich 1979). What environmental needs does multi-unit franchising match, and under what conditions might multi-unit franchising be efcient? With some notable exceptions (Bercovitz 2004; Brickley 1999; Kalnins and Lafontaine 2004; Kaufmann and Dant 1996), the rationale for existence of multi-unit franchising still remains unanswered. We relate to the work done by these scholars as we develop our theory and methodology sections.

Most franchising research is based on either resource constraints theory or agency theory. Resource constraints theory views franchising as a mechanism to ease nancial and managerial constraints on system growth while agency theory views franchising as a mechanism to improve the incentive alignment between the rm and outlet operators. Scholars have proposed that the two theories are not necessarily contradictorya rm needs to both attract resources and align incentivesand might work in concert to explain multi-unit franchising (Combs, Michael, and Castrogiovanni 2004; Kaufmann and Dant 1996). The theory of slicing the hierarchy proposed in this paper, however, is based on agency theory.

Salaried managers may bear partial residual risk if their salary is linked to performance, but franchisees become residual claimants by paying a franchise fee and royalties. These explicit payments make franchisees residual claimants to a greater degree (Mathewson and Winter 1985). Compared with franchisees, salaried managers also are less likely to be concerned about future returns because they do not own the assets (Lutz 1995).

R. Jindal / Journal of Retailing 87 (4, 2011) 549562

551

Conceptual development and hypotheses Franchising involves a two-party relationship between a principal (franchisor) and each of its agents (single-unit franchisees who own and operate an outlet each). Because of the shared brand name though, franchisee investments and efforts have spillover effects (Brickley 1999), which gives franchisees incentives to free ride on joint inputs, such as advertising and investments in maintaining consistent product quality and customer service across all the outlets. As the number of franchisees in a particular territory increases, the incentives of a franchisee to free ride increase as well; such free riding can damage the value of the brand. Because franchisees revenues and the future capital value of their residual claims depend on the value of the brand, franchisees want the franchisor to control free riding in the system. Franchisors want to control free riding too because their revenues and future expansion also depend on the value of the brand (Antia and Frazier 2001; Srinivasan 2006). Research has shown that the threat of termination alone may not be sufcient to deter franchisees because, in most real-life situations, collecting admissible evidence can be expensive and difcult (Dutta, Heide, and Bergen 1999; Lal 1990). One way franchisors control franchisee free riding is by monitoring them (Anderson and Oliver 1987; Eisenhardt 1989). Because franchisors often nd it difcult to monitor all their many and geographically dispersed franchisees themselves, they appoint monitors (e.g., district managers). But these monitors have incentives to shirk because they are salaried employees. For example, during an ethnographic study of restaurant franchisees, Bradach (1998, p. 197) observes: I was once lost for half an hour with a franchise consultant [district manager] who could not nd the location of one of his franchisees restaurants [which he was supposed to be monitoring]. Hence, franchisors need another layer of monitors (e.g., regional managers) to monitor the district managers. Such multiple layers of monitors transform the original two-party franchisorfranchisee relationship into a monitoring hierarchy in which monitors monitor the monitors (Alchian and Demsetz 1972). We propose that franchisors adopt multi-unit franchising in response to conditions that require a larger monitoring hierarchy. Multi-unit franchising broadly takes three formssubfranchising, area development, and incremental allocation in which a single-unit franchisee gradually obtains additional outlets. It is cleaner to juxtapose area development and subfranchising because a franchisor must make these decisions at the time of signing a contract. In incremental allocation of multiple outlets, the franchisor need not anticipate (ex ante) the contractors ability and motivation to operate multiple outlets but instead may observe it (ex post) over a period of time; this simplies the situation considerably and places less strain on the bounds of the principals rationality. Furthermore, organizational structures in incremental allocation and area development are similar. Hence, for the purpose of this study, we do not explicitly model incremental allocation as a mode of multiunit franchising. We do, however, estimate a separate model for adoption of incremental allocation, and the results are similar.

The following subsections explain how franchisors limit their hierarchies by adopting each of the two forms of multi-unit franchising, that is, subfranchising and area development; formal hypotheses follow these explanations. Though the specics may vary, most contracts share these characteristics. Subfranchising In subfranchising, the franchisor enters into a contract with an independent entity and makes it responsible for developing and monitoring multiple outlets in a particular territory. The independent entity (master franchisee) vets prospective franchisees of these outlets (subfranchisees) and selects them with the franchisors approval. Besides the master franchisee, the franchisor also enters into a contract with each subfranchisee to cover certain contingencies, such as the master franchisee quitting the system or being ousted. A master franchisee acts as a mini-franchisor in its territory: assisting subfranchisees in site selection, lease negotiation, construction and design of the outlet, selection, installation, and operation of equipment, staff training, and any other way a franchisor assists single-unit franchisees. The franchisor requires the master franchisee to devote its best efforts to the development and monitoring of subfranchisees. As compensation, the master franchisee gets a share of both the one-time initial franchise fee and royalties collected from the subfranchisees on an ongoing basis. The master franchisee does not receive any form of xed salary (California Electronic Access to Securities and Franchise Information [Cal-EASI] 2010). This compensation design makes both the parties, franchisor and master franchisee, depend solely on the performance of outlets. This incentive alignment is missing in the relationship between a franchisor and its employee monitorsthe franchisor depends solely on the performance of outlets, but the employee monitors draw a portion of their compensation as xed salary independent of the outlet performance. The compensation design improves the master franchisees motivation to monitor the outlets compared with that of an employee monitor such as a district manager. Hence, besides slicing off the lower-level hierarchy, the franchisor needs to monitor the master franchisee less compared with a district manager, which can further reduce the need for a large internal hierarchy. In effect, the franchisor creates mini-franchisors as its contractual partners (see Fig. 1). Area development In area development, the franchisor enters into a contract with an independent entity and allocates it multiple outlets to own and operate in a particular territory. Both the parties agree to a schedule according to which this independent entity (area developer) opens a specic number of outlets in the specied time period. The area developer is responsible for identifying and leasing outlet sites, subject to the franchisors approval. The area developer independently vets and employs outlet managers; the franchisor does not hire or re them. The franchisor requires the area developer to devote its best efforts to the development,

552

R. Jindal / Journal of Retailing 87 (4, 2011) 549562

Fig. 1. Monitoring hierarchies in franchising. Notes: (1) Filled circles indicate residual claimants; unlled circles indicate primarily salaried employees. (2) Solid-line arrows indicate a supervisor-subordinate relationship; dotted-line arrows indicate an independent contractual relationship. (3) A dotted-line box indicates monitoring hierarchy.

management, and monitoring of the outlets, each of which is owned by the area developer. The area developer pays the full franchise fee for the rst outlet but may receive discounts on franchise fees for subsequent outlets. The area developer also pays royalties for the duration of the contract (Cal-EASI 2010). Compared with a master franchisee, an area developer not only gets residual income and decision rights over the outlets but also ownership rights, which confer it a relatively higher level of control over the operational activities at the outlets. Thus, an area developer substitutes for a group of single-unit franchisees who would have been the residual claimants for these outlets. This compensation design improves the area developers motivation to monitor the outlets compared with that of an employee monitor such as a district manager. Furthermore, ownership of multiple outlets in a territory increases the area developers share of spillover benets and, compared with a single-unit franchisee, reduces its incentives to free ride (Brickley 1999; Dant and Gundlach 1999; Kalnins and Lafontaine 2004). Hence, besides slicing off the lower-level hierarchy, the franchisor needs to monitor the area developer less compared with a district manager, which can further reduce the need for a large internal hierarchy. In effect, the franchisor shifts its lower-level hierarchy to the area developer, who acts as an independent mini-hierarchy (see Fig. 1).3

Hypotheses We propose a framework of four factors chosen from the agency-based franchising literature that share a key characteristic: they all are likely to inuence a franchisors decision to adopt multi-unit franchising by increasing the need for a larger monitoring hierarchy. These four factors are: the density of outlets, advertising fee, outlet size, and the geographical dispersion of markets. Besides, we control for a number of other relevant factors that may potentially inuence the adoption of multi-unit franchising. Together, all these factors reect the various aspects of a franchise system such as franchisor characteristics, franchisee characteristics, outlet characteristics, contractual terms, distribution of outlets in a system, and franchisorfranchisee relationship. A franchisee has incentives to free ride if its benets from free riding exceed its costs (Brickley 1999). A franchisee with a relatively large number of neighboring franchisees will have stronger incentives to free ride because a larger number of neighbors would share the costs (Shane 2001). Thus, a franchisor with densely located outlets (where each outlet has a large number of neighbors) would face higher franchisee free riding. One way franchisors can control this free riding is by employing monitors but, as seen earlier, this can lead to a hierarchy of monitoring relationships (Alchian and Demsetz 1972). Both forms of multi-unit franchising allow franchisors to control free riding with smaller monitoring hierarchies.4

3 Similar to company-owned outlets, outlet managers here also have incentives to shirk. But, area managers are considerably smalleron average, an area developer owns four to ve outlets (Wadsworth 2002)and can monitor the outlet managers themselves. This likely keeps shirking at the outlets in check.

4 We address the potential issue of reverse causality in the robustness checks section.

R. Jindal / Journal of Retailing 87 (4, 2011) 549562

553

H1. The greater the density of outlets in a franchise system, the higher is the probability of multi-unit franchising. Most franchisors require their franchisees to contribute toward an advertising fund that is used to establish and maintain the image of the franchisors brand names which is critical to the ongoing attractiveness and the ultimate growth and success of franchise systems (Dant and Berger 1996; p. 1120). Because of the shared brand name though, the contributions to advertising fund have spillover effects, which increase the incentives for franchisees to free ride. These incentives are exacerbated by the franchisees multiple grievances against the advertising fund (Blair and Lafontaine 2005; Dant and Berger 1996). Franchisees (a) fail to see the benets of the fund which is spent primarily on national and regional level advertising, (b) resent the fact that sometimes the franchisor uses the fund to promote a lower price and then forces them to offer this price to customers, (c) feel that the franchisor has simply not done a good job with advertising, or (d) object to the way the franchisor administers the fund. Adopting multi-unit franchising helps a franchisor control such free riding with smaller monitoring hierarchy because multi-unit franchisees are more likely to understand and reap the benets of advertising at the regional level, and likely have more say in how to administer the fund. Hence, H2. The higher the advertising fee in a franchise system, the higher is the probability of multi-unit franchising. Outlet size is a key factor that determines the difculty of operating an outlet (Lafontaine 1992). For example, as the number of employees at an outlet increases, each employee has more incentives to shirk, which requires more effective monitoring (Alchian and Demsetz 1972). Hence, the operator of a large outlet needs to have sufciently strong incentives (Lafontaine and Slade 1997): a single-unit franchisee who is the residual claimant of an outlets earnings has relatively stronger incentives compared with an outlet manager in area development. Moreover, outlets in such franchise systems usually face relatively less intra-chain competitionfor example, any given market is likely to have fewer hotels of a luxury chain than motels of a particular chain, or fewer family restaurants of a particular chain than outlets of a fast-food chain,which reduces the spillover effects to some extent. This, in turn, reduces the potential franchisee free riding curtailing the need for a large monitoring hierarchy, which in turn reduces the need to adopt multi-unit franchising. Hence, H3. The larger the outlet size in a franchise system, the lower is the probability of multi-unit franchising. Geographical distance lowers the subordinates motivation to share information with their supervisors (Maltz and Kohli 1996). Hence, franchisors with markets dispersed over a large area require a larger monitoring hierarchy. Tracing the growth of hierarchy in McDonalds Corporation, Love (1995, p. 387) recounts: The problem was that there were not enough checks on these managers. They were not telling the thinly spread corporate management what they were doing in their regions. [Hence] McDonalds had to create a position of zone manager

. . .. Because a larger monitoring hierarchy aggravates the problems, a franchisor with geographically dispersed markets is more likely to seek ways of reducing the required hierarchy. One such way is to adopt multi-unit franchising. H4. The more geographically dispersed a franchise system, the higher is the probability of multi-unit franchising. Methodology We tested the hypotheses with a longitudinal data set composed of a variety of franchise systems. The longitudinal data allow the use of event history analysis to model the adoption of multi-unit franchising as a function of density, advertising fee, outlet size, and dispersion along with various control variables. Data Ten years of data, from 1995 to 2005, come from Bonds Franchise Guide, a trade publication that collects annual data from franchisors in the United States and Canada. Because potential franchisees often consult this publication, franchisors consider it an important outlet for promotion. The publication collects data based on an annual three-page self-reported questionnaire (see Appendix A for a list of questions relevant to the study), which it sends to the franchisors. Franchisors who do not respond to the rst mailing receive a follow-up package about a month later. To get its information published in the printed guide and the corresponding website, each franchisor has to return the lled-up questionnaire and pay 250 US dollars. Franchisors in the data set represent almost all the sectors in the economy that use franchising including restaurants, food retailing, lodging, personal care services, laundry and dry cleaning services, maid services, printing and graphics services, maintenance services, and various kinds of nonfood retailing. We grouped the franchise systems into 19 categories, according to the products or services they provide (see Table 1). Several empirical studies in franchising literature have relied on the sample of franchise systems included in this publication (Brickley 1999; Martin 1988; Scott 1995; Sen 1993; Shane 2001; Thompson 1994).5 The data however do not provide a balanced panel because some franchisors either do not provide information every year or enter the data set during the observation period. We gave a unique identier to each franchisor the rst time it appears in

5 Bonds questionnaire does not specically ask the respondents to consider their domestic markets or international markets, if any, when lling in the information. It could be argued that when a respondent indicates using multi-unit franchising, it could be in an international market. Mr. Robert Bond, the author and publisher of the trade guide for almost twenty years, asserts that the intent of the overall questionnaire addresses North American operations, and franchisors respond to all the questions keeping in mind their domestic operations only. Furthermore, according to Mr. Bond, international franchises are sold on a country-wide basis and the franchisors would have no knowledge of or involvement with what happens in a country for which they have sold international rights. Hence, they would ll in the questionnaires keeping in mind their domestic operations only. Empirical studies based on these data have also considered them to represent domestic operations only.

554 Table 1 Number of franchisors in various industries. Category

R. Jindal / Journal of Retailing 87 (4, 2011) 549562

1995 19 18 7 26 6 20 21 17 6 9 10 4 3 5 25 12 14 9 60

1996 22 25 9 32 4 25 25 18 10 13 8 2 7 4 26 11 15 10 70

1997 30 21 12 29 8 27 20 24 18 14 7 3 8 7 31 7 14 10 53

1998 33 26 10 25 6 21 23 25 16 10 6 3 9 4 35 11 21 9 57

1999 33 29 8 25 5 23 21 25 13 9 6 3 7 5 37 10 19 7 61

2001 30 21 7 29 9 23 18 20 13 10 5 4 7 9 38 11 20 8 61

2002 36 26 7 33 12 26 22 25 16 9 8 4 7 10 45 13 22 10 63

2003 31 23 6 26 10 24 20 23 12 10 7 3 6 12 38 13 22 5 53

2004 30 23 6 25 11 25 19 23 11 9 7 3 6 12 38 13 21 5 52

2005 30 24 6 22 10 26 19 25 12 9 7 3 7 13 38 16 20 7 56

Automotive products and services, and rentals Building: remodeling, and furniture repair Financial services Business services: advertising, telecom, packaging, mailing, printing, graphics, publications, signs Child development and education Personal development: employment, personnel, education, and training Food retailing: coffee, ice cream, yoghurt, donuts, cookies, bagels Food retailing: quick-service Food retailing: restaurants Food retailing: specialty foods Personal care services: hairstyling, health, tness, beauty Laundry and dry cleaning Lawn and garden care Lodging Maintenance, cleaning, and sanitation Medical, optical, and dental services Real estate services and inspection Recreation: entertainment and travel Non-food retailing: art supplies, sporting goods, clothing, shoes, convenience stores, drug stores, home furnishings, hardware, pet products, photographic products, video and audio, and security and safety Total

291

336

343

350

346

343

394

344

339

350

the data set to follow it through the years. During the observation period, some franchisors changed the names under which they operated. We used public sources such as trade magazines and franchisor websites to account for such changes to ensure uninterrupted series of data. Furthermore, because the data range only from 1995 to 2005, we do not have information for each franchise system from its date of inception. Hence, some franchisors are already using multi-unit franchising when they appear in the data set for the rst time. It is difcult to ascertain the factors that might have inuenced these franchisors to adopt multi-unit franchising; hence, we excluded all such rms from the data set (Allison 1995). This left 3,436 observations from 713 franchisors. Table 1 shows the number of franchisors in the data set categorized by industry. During the period observed, 125 franchisors (approximately 18%) adopted area development and 50 franchisors (approximately 7%) adopted subfranchising. Among the franchisors adopting area development, 3 abandoned and then readopted area development, 2 abandoned area development and adopted subfranchising, and 13 franchisors abandoned area development all together. Among the franchisors adopting subfranchising, 1 abandoned subfranchising and adopted area development, and 5 franchisors abandoned subfranchising all together. Table 2 shows the number of franchisors adopting multi-unit franchising each year in each industry. Description of variables The annual questionnaire asks franchisors if they use area development or subfranchising. We used this information to

obtain the dependent variable: the adoption of multi-unit franchising.6 Franchisors usually begin operations in their home states and later expand to other states (Kaufmann and Dant 1996). At any time, outlets in most franchise systems thus are unevenly distributed across various markets (McKee, Lovejoy, and Moran 2004). The questionnaire asks franchisors which three states have the largest number of outlets and the number of outlets in each of these states. The data show that the home state is almost always one of these three states. Data further show that, on average, these three states account for a substantial proportion of franchisors total number of outlets: more than a quarter of all outlets in 82 percent of the observations, and more than half of all outlets in 50 percent of the observations. Furthermore, more than a quarter of franchisors operate in three or less number of states, in which case these states account for all the outlets. We use the information about the number of outlets in the three key states to obtain indicators for two predictor variables, density and dispersion. Density increases the severity of potential franchisee free riding in a system, which increases the size of the monitoring hierarchy a franchisor needs. We used the average number of outlets in the key markets to measure density. Because a franchise system is relatively more developed in its key states, this indicator effectively reects the severity of potential franchisee free

For reasons mentioned earlier, we do not explicitly model incremental allocation as a mode of multi-unit franchising. We do, however, estimate a separate model for adoption of incremental allocation, and the results are similar.

R. Jindal / Journal of Retailing 87 (4, 2011) 549562 Table 2 Number of franchisors adopting multi-unit franchising in various industry categories. Category Automotive products and services, and rentals Building: remodeling, and furniture repair Financial services Business services: advertising, telecom, packaging, mailing, printing, graphics, publications, signs Child development and education Personal development: employment, personnel, education, and training Food retailing: coffee, ice cream, yoghurt, donuts, cookies, bagels Food retailing: quick-service Food retailing: restaurants Food retailing: specialty foods Personal care services: hairstyling, health, tness, beauty Laundry and dry cleaning Lawn and garden care Lodging Maintenance, cleaning, and sanitation Medical, optical, and dental services Real estate services and inspection Recreation: entertainment and travel Non-food retailing: art supplies, sporting goods, clothing, shoes, convenience stores, drug stores, home furnishings, hardware, pet products, photographic products, video and audio, and security and safety Total 1995 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1996 2 3 1 6 1 1 6 4 0 0 1 1 1 0 0 0 1 0 4 1997 4 3 1 1 0 2 5 2 6 0 0 2 1 0 3 3 1 0 4 1998 2 3 1 6 0 1 1 6 2 0 0 0 0 1 4 1 2 0 5 1999 3 0 1 1 0 1 3 1 2 1 1 1 0 0 0 0 1 0 5 2001 5 2 0 0 1 1 1 3 4 1 0 0 2 1 5 0 0 2 6 2002 1 0 1 1 0 0 1 3 0 0 0 0 1 0 3 2 0 0 1 2003 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 1 0 2 2004 0 1 0 0 0 2 1 0 0 0 0 0 0 0 0 0 0 0 0

555

2005 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 2

32

38

35

21

34

14

riding, which forces a franchise system to adopt multi-unit franchising. Previous franchising literature has used a similar density indicator (Brickley 1999; Lafontaine 1995; Minkler 1990). It could be argued that the effective density might be different from what this indicator reects because the key states likely vary in sizes. We estimated a model using an indicator adjusted for areanumber of outlets per 100 square milesbut found similar results. We prefer to use the unadjusted measure because it is more robust considering that several states may have large uninhabited areas. In our dataset, more than two-thirds of franchisors charge their franchisees an advertising fee. Due to spillover effects exacerbated further by franchisees multiple grievances, advertising fee increases franchisees incentives to free ride (Blair and Lafontaine 2005; Dant and Berger 1996), which increases the size of the monitoring hierarchy a franchisor needs. In most cases, advertising fee is indicated as a xed proportion of the total sales. Some franchisors, however, indicate a gradient of advertising fee proportional to the amount of total sales (Lafontaine 1992); in such cases, we take the average. For example, if a franchisor has a gradient of advertising fee from 1 to 2 percent, we take the average of 1.5 percent as the advertising fee. Furthermore, some franchisors indicate a at advertising fee independent of total sales. To obtain a proportion in such cases we use the average proportion of advertising fee to royalty rate in the dataset, which is obtained from the cases where both these fees are given as proportion of total sales. On average, royalty rate is 5.5 percent of total sales while the advertising fee is 1.57 percent of total sales, which implies that the advertising fee is

28.5 percent of the royalty rate. When the advertising fee is indicated as a at fee while royalty rate is a proportion of the total sales, we multiply the royalty rate with 0.285 to get advertising fee as a proportion of total sales. In a few cases, both advertising fee and royalty are indicated as at fees independent of total sales. In these cases, we assume the average royalty rate of 5.5 percent, and multiply with the ratio of advertising fee and royalty to get advertising fee as a proportion of total sales. For example, if advertising fee is $100 per week while royalty is $400 per week, we multiply 5.5 percent (average royalty rate) with 100 divided by 400 to get advertising fee of 1.38 percent. To check that the model results are not biased by these assumptions, we re-estimated the model using only the cases where advertising fee is indicated as a proportion of total sales, and nd that the results are similar. Large outlets are relatively more difcult to operate and need an operator with stronger incentives (Lafontaine 1992). Furthermore, the extent of spillovers is likely to be less because of less intra-chain competition, which reduces the potential franchisee free riding and the size of the monitoring hierarchy a franchisor needs. In the dataset, franchisors indicate how many employees, both full-time and part-time, are required to run a franchised outlet.7 The dataset, however, does not provide information

7 Depending on the research question, scholars have used various proxies to capture outlet size such as average sales per outlet, investment in an outlet, and the number of seats in an outlet (Lafontaine and Slade 1997). In a study of restaurant and fast-food industry, Lafontaine (1995, p. 15) used the number

556

R. Jindal / Journal of Retailing 87 (4, 2011) 549562

about (a) do part-time employees differ from full-time employees only in their employment status but otherwise work the same hours, and (b) if not, what proportion of time do they work compared with full-time employees. We assume that part-time employees work half the time of full-time employees. Hence, if a franchisor recommends four full-time employees and four parttime employees, we assume that the recommended staff size at the outlet is six employees. To check that model results are not biased by this assumption, we re-estimated the model using two alternate assumptions: part-time employees differ only in their status but work the same hours as full-time employees (which means a total of 8 employees at the outlet in this example) and part-time employees work only one-third the time of full-time employees (which means a total of 5.33 employees at the outlet in this example); we nd that the results are similar. Geographical distance of markets from the headquarters reduces monitors motivation to disseminate information, which in turn increases the size of the monitoring hierarchy a franchisor needs (Lal 1990). Because the franchisees in key states are more likely to free ride, we used the distance between a franchisors headquarters and its key states to measure the geographical dispersion of a franchise system (Brickley and Dark 1987; Minkler 1990). Specically, the sum of the ying distance in miles between the home state of the franchisor and each of its key states provides a measure of dispersion. The reference point for measuring distance is the main airport in the largest city of each state. We use ying distance for two reasons: rst, ying is the most common mode of business travel, and second, among the various modes of transportation, airplane routes best represent the Euclidean distance between any two locations. When the home state itself is one of the key states, the ying distance, or travel distance within the home state, takes a value of 1 mile. Although crude, this assumption runs counter to the proposed hypothesis and therefore is conservative. To take an example, for a franchisor with headquarters in Texas and with Illinois, New York, and Texas as its key states, dispersion is the sum of ying distance from IAH airport in Houston to ORD airport in Chicago (926 miles), ying distance from IAH airport to JFK airport in New York City (1,410 miles), and distance within the home state (1 mile). The total value of 2,337 miles indicates the dispersion of this franchisor. Descriptive statistics for all the variables are shown in Table 3. Six continuous variables (density, outlet size, dispersion, experience, size, and franchisee investment) are skewed, so to normalize their distributions, we use their log transforms (Greene 2000). Control variables The model controlled for several factors that may inuence a franchisors decision to adopt multi-unit franchising. First,

of seats in an outlet but noted that the appropriate measure here might have been the number of ovens or employees. For our research question, we want to capture the importance of incentives for an outlet operator, which is well reected by the number of employees to be monitored at the outlet.

it controlled for any inuence of resource constraints a franchisor may have. Some scholars propose that franchisors require nancial and managerial resources to support their single-unit franchisees in opening and operating their outlets (Blair and Lafontaine 2005; Combs and Ketchen 1999). Single-unit franchisees may nd it difcult to open and operate without such support because they usually do not have the resources and skills to obtain these services on their own. Multi-unit franchisees, as operators of multiple outlets, can leverage their size to obtain these services cost effectively. Hence, a resource-constrained franchisor may be more likely to adopt multi-unit franchising. Furthermore, by adopting multi-unit franchising such a franchisor may gain access to multi-unit franchisees resources to grow faster, gain operational efciencies, or build economies of scale to compete against other rms (Oxenfeldt and Kelly 1968). Therefore, we used four binary variables to control for any potential inuence of resource constraints: whether a franchisor provides nancial assistance to its franchisees, whether a franchisor assists franchisees in selecting suitable sites for their outlets, whether a franchisor assists franchisees in negotiating leases with the site owners, and whether a franchisor assists franchisees in evaluating their outlet operations. Second, literature in inter-organizational relationships contends that pre-contractual vetting to minimize adverse selection may also reduce the extent of post-contractual monitoring required to control moral hazard (Ouchi 1979; Stump and Heide 1996). Hence, vetting of potential single-unit franchiseesbrand custodians at the outletsmay help reduce free riding in the franchise system. Trade press indicates that although most franchisors vet their potential single-unit franchisees, the importance of vetting varies. Because the extent of franchisor control over the vetting of brand custodians in the two modes of multi-unit franchising is differentless control over outlet managers in area development but more control over subfranchisees in subfranchising,the importance of vetting may inuence which mode of multi-unit franchising, if any, a franchisor adopts. Hence, we used two variables to control for any potential inuence of pre-contractual vetting: the importance of conducting personal interviews with potential single-unit franchisees and the importance of preparing their psychological proles. Both the variables are reported on a 15 scale, where 1 denotes unimportant and 5 denotes very important. Third, the model controlled for potential inuence of the rms experience, size, and growth rate. Before franchising, some rms rene their business formats for a few years using only company-owned outlets. A rm that has taken longer to franchise may not have acquired the skills required for franchisee selection and training and may be more likely to adopt multi-unit franchising. We measured experience as the number of years a rm takes to start franchising after its inception. A large system also may be more likely to adopt multi-unit franchising, as it may be less apprehensive about creating powerful contractual partners. Therefore, we controlled for size, which is measured as the total number of outlets for each franchisor. Furthermore, scholars have shown positive correlation between growth in a franchise system and the use of multi-unit franchising (Kaufmann and Dant 1996). They argue that this relationship

Table 3 Descriptive statistics and correlation matrixa (n = 3,436). Variable 1. MUF 2. Density 3. Advertising fee 4. Outlet size 5. Dispersion 6. Financial assistance 7. Assistance with site selection 8. Assistance with lease negotiation 9. Assistance with eld oper. eval 10. Imp. of personal interview 11. Imp. of psych. prole 12. Experience 13. Size 14. Growth 15. Franchisee investment 16. Competition 17. Interval Mean 0.19 29.14 1.51 7.93 1705.47 0.65 0.89 0.70 0.89 4.39 3.07 7.90 259.71 0.12 249.91 41.65 1.16 0.88 1.16 12.23 959.57 0.39 1082.58 35.54 0.51 S.D. 79.32 1.73 13.46 1429.72 1 1.00 0.02 0.09 0.01 0.01 0.07 0.09 0.10 0.07 0.01 0.03 0.02 0.03 0.01 0.02 0.08 0.06 2 1.00 0.12 0.01 0.19 0.03 0.10 0.10 0.06 0.03 0.05 0.16 0.54 0.05 0.02 0.09 0.01 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

1.00 0.13 0.10 0.02 0.08 0.08 0.07 0.10 0.04 0.02 0.05 0.01 0.13 0.15 0.02

R. Jindal / Journal of Retailing 87 (4, 2011) 549562

1.00 0.01 0.13 0.07 0.11 0.09 0.02 0.05 0.00 0.01 0.02 0.45 0.17 0.03

1.00 0.04 0.14 0.17 0.03 0.15 0.10 0.07 0.48 0.04 0.10 0.15 0.00

1.00 0.02 0.16 0.06 0.00 0.03 0.02 0.10 0.00 0.02 0.07 0.04

1.00 0.45 0.04 0.19 0.04 0.06 0.17 0.04 0.13 0.09 0.01 1.00 0.12 0.17 0.06 0.00 0.17 0.03 0.07 0.10 0.01 1.00 0.13 0.04 0.01 0.01 0.02 0.06 0.10 0.01 1.00 0.26 0.00 0.02 0.06 0.15 0.07 0.01 1.00 0.01 0.11 0.01 0.17 0.04 0.01 1.00 0.11 0.02 0.09 0.04 0.00

1.00 0.07 0.08 0.02 0.01

1.00 0.02 0.01 0.05

1.00 0.02 0.02 1.00 0.00

1.00

Notes: Standard deviation has not been calculated for dichotomous variables. Spearman rank correlations are reported for pairs of dichotomous variables and their pairs with continuous variables, and Pearson product moment correlations are reported for pairs of continuous variables. a Correlations greater than .033 (absolute value) are signicant at the .05 level.

557

558

R. Jindal / Journal of Retailing 87 (4, 2011) 549562

is related to resource constraints of the franchisor, and propose that agency theory and resource constraints theory work in concert to explain multi-unit franchising. We derived the growth rate from information about system size and controlled for both system growth and its square term. Fourth, a franchisor may nd it difcult to recruit enough eligible single-unit franchisees. For example, some franchisors stipulate the nancial strength of prospective single-unit franchisees; if only a few entities fulll those stipulations, franchisors may be forced to allocate them multiple outlets. We measured the scale of required franchisee investment as an average of three items (Brickley and Dark 1987; Lafontaine 1992): (1) minimum net worth (in 000 dollars) required to be a franchisee in the system, (2) equity capital (in 000 dollars) required to own an outlet, and (3) total investment (in 000 dollars) required to start the outlet. Fifth, we controlled for any inuence of market competition (Gal-Or 1995). Competition is the number of other franchisors in the industry to which a franchise system belongs. Actual competition, however, could be relatively more severe or less severe because all these franchisors may not be competing in each market and/or some rms in the industry may not be using franchising. Sixth, industries differ in several characteristics that may affect the cost or difculty of monitoring, such as the proportion of repeat customers (e.g., high in cleaning services versus low in sign making), the location of operations (e.g., xedlocation restaurants versus mobile lawn care), and the level of programmability of outlet employees tasks (e.g., high in quickservice restaurants versus low in auto repair). Hence, the model controlled for industry-level effects. Model and estimation Event history analysis enables the study of occurrence of events over a period of time. Compared with other statistical methods, this method is better suited to account for censored observations and time-varying covariates (Greene 2000). Therefore, we specied and estimated a repeated events model to estimate the probability of a franchisor adopting multi-unit franchising during the observation period (Box-Steffensmeier and Jones 2004).8 This model accounts for a franchisor adopting one or both modes of multi-unit franchising; or abandoning and then readopting the same or the other mode. Repeated events Cox regression model adjusts the variance of the parameter estimates by clustering on franchisor to account for the repeated nature of the data. The hazard rate for the jth cluster and kth event is hk (t ) = h0 (t ) exp( xkj ) (1)

where h0 (t) is the baseline hazard rate function, and xkj are the regression parameters and covariates (Box-Steffensmeier and Jones 2004). The model uses a robust covariance matrix given by V = I1 G GI1 where I1 is the estimated covariance matrix and G is a matrix of the group efcient score residuals. The data set has three characteristics that needed to be accounted for during model estimation. First, more than one franchisor may adopt multi-unit franchising in a year. Known as tied events, such an occurrence is purely statistical and does not imply any relationship between franchisors or their decisions. We used Efron approximation to modify the partial likelihood function to account for such tied events (Allison 1995; Cleves, Gould, and Gutierrez 2004). Second, sometimes data are not available for some rms for one or more years. Other things being equal, the probability of an event increases with the length of this interval. If time-varying covariates are associated with interval length, the estimated coefcients may also be biased. Therefore, we controlled for the length of the interval by including it as a covariate. The square term of interval controlled for any potential nonlinear effect (Allison 1995). Third, rms sometimes differ in ways that are not fully captured by a model. Such unobserved heterogeneity produces dependence between the multiple observations of the same rm, which may attenuate the estimates of covariate effects. The model accounts for any unobserved heterogeneity (Box-Steffensmeier and Jones 2004). Results The results from the model estimation appear in Table 4. These results support Hypothesis 1: The greater the density of outlets in a franchise system, the higher is the probability of multi-unit franchising ( = 0.57, p = 0.03). The results also support Hypothesis 2: The higher the advertising fee in a franchise system, the higher is the probability of multi-unit franchising ( = 0.16, p < 0.01). Hypothesis 3 is supported too: The larger the outlet size in a franchise system, the lower is the probability of multi-unit franchising ( = 0.24, p = 0.08). Finally, the results also support Hypothesis 4: The more geographically dispersed a franchise system, the higher is the probability of multi-unit franchising ( = 0.14, p = 0.02). Results indicate that franchisors providing assistance with eld operation evaluation may be more likely to adopt multiunit franchising. Franchisors that consider conducting personal interviews with potential franchisees very important are less likely to adopt multi-unit franchising. This result could partly be due to relatively higher proportion of franchisors adopting area development in the datasetbecause area development allows a franchisor less control over the selection of outlet managers, franchisors may be less likely to adopt area development. Results further indicate that the faster the growth of a franchise system, higher is the probability of multi-unit franchising, which is in line with the ndings of Kaufmann and Dant (1996). Limitations and robustness checks Although the data used in this study are comprehensive in their coverage of various industries, are available over a

We prefer a semiparametric specication to parametric specication for two reasons: (1) It is more robust because it does not specify a baseline hazard, and (2) a parametric model is cumbersome to estimate when the data include a large number of time-varying covariates (Cleves et al. 2004). In this study, several explanatory and control variables vary over time.

R. Jindal / Journal of Retailing 87 (4, 2011) 549562 Table 4 Results from repeated-events cox regression model for the likelihood of multiunit franchising. Coefcient Explanatory variables Log (Density) Advertising fee Log (Outlet size) Log (Dispersion) Control variables Financial assistance Assistance with site selection Assistance with lease negotiation Assistance with eld operation evaluation Imp. of personal interview Imp. of psychological prole Log (Experience) Log (Size) Growth Growth squared Log (Franchisee investment) Competition Interval Interval squared Industry category Automotive products and services Building: remodeling, and furniture repair Financial services Business services Child development and education Personal development Food retailing: coffee, ice cream, yoghurt Food retailing: quick-service Food retailing: restaurants Food retailing: specialty foods Personal care services Laundry and dry cleaning Lawn and garden care Lodging Maintenance, cleaning, and sanitation Medical, optical, and dental services Real estate services and inspection Recreation: entertainment and travel Log likelihood Wald chi-square Notes: Two-tailed tests. * p < 0.10. ** p < 0.05. *** p < 0.01. 0.57** 0.16*** 0.24* 0.14** 0.31 0.92 0.19 1.77* 0.36*** 0.14* 0.02 0.32 2.43*** 0.76*** 0.12 <0.01 0.37 0.14 0.61 0.23 1.64* 0.01 0.31 1.53 0.74** 0.82 1.97*** 0.64 0.30 1.82*** 1.88*** 0.48 0.45 0.24 0.14 0.82 325.13 113.01*** S.E. 0.26 0.05 0.14 0.06 0.23 0.60 0.25 0.98 0.12 0.08 0.04 0.20 0.57 0.29 0.15 <0.01 0.57 0.14 0.76 1.16 0.93 0.72 0.41 1.04 0.32 0.76 0.48 0.66 1.06 0.64 0.37 1.22 0.73 0.57 0.66 0.83

559

longitudinal period of time, and have been shown to be representative of the overall franchising sector, the empirical analysis has some limitations. First, we theorized that a franchisor adopts multi-unit franchising to reduce the size of monitoring hierarchy required for increased density; however, the use of multi-unit franchising arguably might lead to greater density. There is ample empirical evidence to support our theory (Lowell 2007; McKee, Lovejoy, and Moran 2004). Trade press indicates that an area developer usually opens only one outlet in its rst year of operations, very similar to a single-unit franchisees operations. Therefore, it seems unlikely that the adoption of area development leads to greater density in the system, at least at the time of adoption, which corresponds to the dependent variable

in this study. Trade press also indicates that a multi-unit franchisee may in fact open outlets at a slower pace compared with a group of single-unit franchisees (Siebert 2004), which again suggests multi-unit franchising is unlikely to lead to increased density, at least in the short term. Furthermore, we conducted a Granger causality test to nd support for our theorizing. Marketing scholars (e.g., Chintagunta and Haldar 1998) have used this test to determine similar causal relationships. The Akaike information criterion suggested a two-period lag model as the most appropriate. The Durbin Watson statistic ruled out any autocorrelation. The results showed that whereas the lagged values of density Granger-cause the use of multi-unit franchising (F (2, 1072) = 3.45; p = 0.03), the lagged values of the use of multi-unit franchising do not Granger-cause density (F (2, 1072) = 0.73; p = 0.48). Despite the empirical evidence and the results of Granger causality test, longitudinal data at the outlet level could provide more denitive evidence. Second, because the use of multi-unit franchising is reported on an aggregate system-wide basis in the data set, multi-unit franchising arguably might not have been adopted in any of the key markets at all. Instead, franchisors might have adopted multiunit franchising as a way to enter and quickly gain operational efciencies in a new market or in an existing non-key market. The results of control variables indicate that franchisors do not seem to adopt multi-unit franchising to gain access to multiunit franchisees resources, which could be used for growing quickly or gaining operational efciencies. Nonetheless, to rule out the possibility that franchisors might have adopted multiunit franchising to enter a new state, we estimated the model using only the observations where the franchisors did not enter a new state during the observation period. The results supported the ndings of the study. Additionally, to rule out the possibility that franchisors might have adopted multi-unit franchising in one of the non-key markets, we estimated the model using only the observations where franchisors operated in three or less number of states during the entire observation period. The results again supported the ndings. Ideally, the data should have provided all relevant information for each franchisor in each individual market in which it operates over the entire observation period. However, we are not aware of any franchising database that provides such market-level longitudinal information. Third, the data used in the study had been collected using selfreported questionnaires. Observations that do not have data on all the variables included in the model are dropped during estimation. Some scholars question the representativeness of such a sample (Allison 2002). Discussion The results show that franchisors are more likely to adopt multi-unit franchising in response to conditions that require a large monitoring hierarchy. Four such conditions are the higher density of outlets, higher advertising fee, smaller outlet sizes, and wider geographical dispersion of markets. Density increases the severity of free riding in a franchise systema franchisee with more neighboring franchisees has stronger incentives to free ridewhich requires a franchisor to build a large

560

R. Jindal / Journal of Retailing 87 (4, 2011) 549562

monitoring hierarchy. Due to spillover effects exacerbated further by franchisees multiple grievances, advertising fee increases franchisees incentives to free ride, which requires a franchisor to build a large monitoring hierarchy. Operators of large outlets need to be provided stronger incentives to monitor larger number of employees. Franchise systems with large outlets also have reduced spillover effects due to relatively lower intra-chain competition. Thus, a franchisor with larger outlets does not need to build a large monitoring hierarchy. Geographical dispersion of markets lowers subordinates motivation to share information with their supervisors, which again increases the size of hierarchy a franchisor needs. Agency-based literature has looked at the relationship between free riding, monitoring, and multi-unit franchising. Scholars have primarily applied logit models to estimate the cross-sectional relationships of density and dispersion with the use of multi-unit franchising (Bercovitz 2004; Brickley 1999). Kalnins and Lafontaine (2004) examined the allocation of outlets to multi-unit franchisees. We contribute to this literature by formulating multi-unit franchising as a governance structure that a franchisor adopts to reduce the problems of hierarchy. We looked at the integrative role of four agency-based variables by applying event history analysis to a longitudinal data set composed of almost all the sectors in the economy that use franchising. Specically, we proposed that by using multiunit franchising a franchisor slices its hierarchy and substitutes lower-level employee monitors with independent entitiesarea developers or master franchisees. Because their compensation depends entirely on outlet performance, these independent entities are more motivated to monitor the outlets compared with employee monitors. Hence, besides slicing off the lower-level hierarchy, the franchisor needs to monitor the multi-unit franchisees less compared with employee monitors, which can further reduce the size of required internal hierarchy. Whereas a master franchisee acts as a mini-franchisor in its territory, an area developer acts as an independent mini-hierarchy. Thus, area development consists of two hierarchical components joined together with an independent contract. Their relatively smaller size compared to the franchisor (Wadsworth 2002) ensures that area developers are relatively less burdened by the problems of hierarchy. This research suggests several avenues for further study. First, franchising is a stylized governance structure as it involves fairly standard and relatively complete take-it-or-leave-it contracts. It would be interesting to replicate this study in other settings where organizations may prefer hierarchies, even in adverse conditions, due to incomplete contracts. Second, this study focused on franchisors reasons to adopt multi-unit franchising; it would be interesting to integrate the franchisees reasons for operating multiple outlets too (Darr, Argote, and Epple 1995). Third, while this study has highlighted the efciency gained by adopting multi-unit franchising, there could be efciency loss due to multi-unit franchisees relative lack of experience and resultant inconsistency in brand presentation and customer service. It would be interesting to compare the performance of outlets under different governance structures. Fourth, attening of hierarchyby providing performance-based compensation and

other long-term incentives to employeesrepresents an alternative approach to reduce the size of internal hierarchy (Rajan and Wulf 2006). It would be interesting to explore conditions that may lead organizations to choose one approach over the otherslicing or atteningto achieve a leaner hierarchy. Fifth, although both area development and subfranchising help reduce the size of hierarchy, they reveal some key differences. Subfranchising relies on the more efcient control of franchisee free riding by master franchisees; area development, on the other hand, gives residual claims of multiple outlets to area developers and substitutes their outlet managers shirking for franchisee free riding. Which form of moral hazard is a franchisor more likely to endure at the outlet level: shirking or free riding? It would be interesting to study factors that may inuence the preference for one mode over the other. Because the monitoring hierarchy in a franchise system is comparable to the internal hierarchy in a typical organization, the idea of slicing and shifting lower-level hierarchy to contractors can be extended to organizations in other settings. For example, when rms outsource their business processes, in effect, they shift their lower-level hierarchies to independent contractors and reduce the size of their internal hierarchies. This view offers an alternative argument for outsourcing, which has been seen primarily as a way to cut labor costs. Conventional wisdom suggests that outsourcing becomes less attractive as the labor cost savings decline. This research however suggests that besides lower labor costs, outsourcing can deliver sustainable savings by allowing a rm to reduce the size of its internal hierarchy. Thus declining labor cost savings may make offshoringwhere rms in developed countries move jobs to less developed countries to save on labor costsless attractive but outsourcing is likely to remain attractive. This may manifest in offshored jobs returning to the home country but remaining outsourced, that is, still done by independent contractors. Furthermore, this view suggests that contractors be considerably smaller than the rms; such small contractors are likely less burdened with the problems of hierarchy. If contractors were to get too large, they may also face similar problems of hierarchy, and in an attempt to reduce these problems they themselves may start outsourcing. This alternative argument for outsourcing gets some support; for example, Azoulay (2004) found that pharmaceutical rms outsourcing their clinical trials based their decision not on the trade-off between market and hierarchy, but on the choice between the hierarchy of their own rm and the hierarchy of their contractors.

Acknowledgments The author dedicates this article to his PhD dissertation chair Professor Erin Anderson. The author thanks the other members of his dissertation committee: Professors Hubert Gatignon, Francine Lafontaine, Paddy Padmanabhan, Javier Gimeno, and Scott Shane. The author also thanks Professors Shantanu Dutta, Bart Weitz, and Jim Hess for their critical comments. A special mention is reserved for Professor Edward Blair for his feedback during the development of the manuscript. Finally, the author

R. Jindal / Journal of Retailing 87 (4, 2011) 549562

561

thanks Professor Rajiv Dant and anonymous JR reviewers for their helpful comments. Appendix A. Annual self-reported questionnaire used by Bonds Franchise Guide9
Franchisor information Franchise Trade Name Address Business description Franchisor background Year the company was founded First year as franchisor Total operating outlets in the U.S. Total operating outlets in Canada What 3 states/provinces have the largest number of operating outlets? How many operating outlets are located in these states/provinces? In qualifying a potential franchisee, please rank the following criteria from Unimportant to Very Important (15 scale) Personal interview Psychological prole Including the owner/operator, how many employees are recommended to staff the average franchised unit? Full-Time Part-Time Financial requirements Even though the cash investment may vary substantially by individual outlet, what is the range of equity capital (up-front cash) required? What is the range of total investment required? What is the minimum net worth required of the franchisee? How much is the on-going advertising fee? % or Terms of contract Do you have area development agreements? Do you have subfranchising contracts? Franchisor support provided Is nancial assistance available? Do you assist the franchisee in site selection? Do you assist the franchisee in lease negotiations? Do you provide eld operations evaluation for franchisee outlets?

References
Aghion, Phillipe and Jean Tirole (1997), Formal and Real Authority in Organizations, Journal of Political Economy, 105 (1), 129. Alchian, Armen A. and Harold Demsetz (1972), Production, Information Costs, and Economic Organization, American Economic Review, 62 (5), 77795. Aldrich, Howard E. (1979), Organizations and Environments, Englewood Cliffs, NJ: Prentice-Hall. Allison, Paul D. (1995), Survival Analysis Using the SAS System: A Practical Guide, Cary, NC: SAS Institute Inc. (2002), Missing Data, Thousand Oaks, CA: Sage Publications. Anderson, Erin (1988), Strategic Implications of Darwinian Economics for Selling Efciency and Choice of Integrated or Independent Sales Forces, Management Science, 34 (5), 599618. Anderson, Erin and Richard L. Oliver (1987), Perspective on Behavior-Based versus Outcome-Based Salesforce Control Systems, Journal of Marketing, 51 (4), 7688. Antia, Kersi D. and Gary L. Frazier (2001), The Severity of Contract Enforcement in Interrm Channel Relationships, Journal of Marketing, 65 (4), 6781.

Only the relevant questions are shown.

Arrow, Kenneth J. (1985), The Economics of Agency, in Principals and Agents: The Structure of Business, Pratt J. W. and Zeckhauser R. J., eds. Harvard Business School Press, 3751. Azoulay, Pierre (2004), Capturing Knowledge Within and Across Firm Boundaries: Evidence from Clinical Development, American Economic Review, 94 (5), 1591612. Bercovitz, Janet E.L. (2004), The Organizational Choice Decision in BusinessFormat Franchising: An Empirical Test, in Economics and Management of Franchising Networks, Windsperger J., Cliquet G., Hendrikse G. and Tuunanen M., eds. New York: Springer-Verlag, 3868. Blair, Roger D. and Francine Lafontaine (2005), The Economics of Franchising, New York: Cambridge University Press. Bond, Robert E. (2005), Bonds Franchise Guide, Oakland, CA: Source Book Publications. Box-Steffensmeier, Janet M. and Bradford S. Jones (2004), Event History Modeling: A Guide for Social Scientists, Cambridge, UK: Cambridge University Press. Bradach, Jeffrey L. (1998), Franchise Organizations, Boston, MA: HBS Press. Brickley, James A. (1999), Incentive Conicts and Contractual Restraints: Evidence from Franchising, Journal of Law and Economics, 42 (2), 74574. Brickley, James A. and Frederick H. Dark (1987), The Choice of Organizational Form: The Case of Franchising, Journal of Financial Economics, 18 (2), 40120. California Electronic Access to Securities and Franchise Information (Cal-EASI) database (2010), (accessed July 12, 2010), [available at http://www.corp.ca.gov/CalEASI/caleasi.asp]. Calvo, Guillermo A. and Stanislaw Wellisz (1978), Supervision, Loss of Control, and the Optimum Size of the Firm, Journal of Political Economy, 86 (5), 94352. Chintagunta, Pradeep and Sudeep Haldar (1998), Investigating Purchase Timing Behavior in Two Related Product Categories, Journal of Marketing Research, 35 (1), 4353. Cleves, Mario A., William W. Gould and Roberto G. Gutierrez (2004), An Introduction to Survival Analysis Using STATA, College Station, TX: Stata Corporation. Combs, James G. and David J. Ketchen Jr. (1999), Can Capital Scarcity Help Agency Theory Explain Franchising? Revisiting the Capital Scarcity Hypothesis, Academy of Management Journal, 42 (2), 196207. Combs, James G., Steven C. Michael and Gary J. Castrogiovanni (2004), Franchising: A Review and Avenues to Greater Theoretical Diversity, Journal of Management, 30 (6), 90732. Dant, Rajiv P. and Paul D. Berger (1996), Modelling Cooperative Advertising Decisions in Franchising, Journal of the Operational Research Society, 47 (9), 112036. Dant, Rajiv P. and Gregory T. Gundlach (1999), The Challenge of Autonomy and Dependence in Franchised Channels of Distribution, Journal of Business Venturing, 14 (1), 3567. Darr, Eric D., Linda Argote and Dennis Epple (1995), The Acquisition, Transfer, and Depreciation of Knowledge in Service Organizations: Productivity in Franchises, Management Science, 41 (11), 175062. Dutta, Shantanu, Jan B. Heide and Mark E. Bergen (1999), Vertical Territorial Restrictions and Public Policy: Theories and Industry Evidence, Journal of Marketing, 63 (4), 12134. Eisenhardt, Kathleen M. (1989), Agency Theory: An Assessment and Review, Academy of Management Review, 14 (1), 5774. Gal-Or, Esther (1995), Maintaining Quality Standards in Franchise Chains, Management Science, 41 (11), 177492. Garrouste, Pierre and Stephane Saussier (2005), Looking for a Theory of the Firm: Future Challenges, Journal of Economic Behavior and Organizations, 58 (2), 17899. Gibbons, Robert (2005), Four Formal(izable) Theories of the Firm, Journal of Economic Behavior and Organization, 58 (2), 20045. Greene, William H. (2000), Econometric Analysis, Upper Saddle River, NJ: Prentice Hall. Hirshleifer, Jack (1985), The Expanding Domain of Economics, American Economic Review, 75 (6), 5368.

562

R. Jindal / Journal of Retailing 87 (4, 2011) 549562 Minkler, Alanson (1990), An Empirical Analysis of a Firms Decision to Franchise, Economic Letters, 34 (1), 7782. Ouchi, William G. (1979), A Conceptual Framework for the Design of Organizational Control Mechanisms, Management Science, 25 (9), 83348. Oxenfeldt, Alfred R. and Anthony O. Kelly (1968), Will Successful Franchise Systems Ultimately Become Wholly-Owned Chains?, Journal of Retailing, 44 (4), 6983. Radner, Roy (1992), Hierarchy: The Economics of Managing, Journal of Economic Literature, 30 (3), 13821415. Rajan, Raghuram G. and Julie M. Wulf (2006), The Flattening Firm: Evidence from Panel Data on the Changing Nature of Corporate Hierarchies, Review of Economics and Statistics, 88 (4), 75973. Rubin, Paul H. (1978), The Theory of the Firm and the Structure of the Franchise Contract, Journal of Law and Economics, 21 (1), 22333. Scott, Frank A. Jr. (1995), Franchising vs. Company Ownership as a Decision Variable of the Firm, Review of Industrial Organization, 10 (1), 6981. Sen, Kabir C. (1993), The Use of Initial Fees and Royalties in Business Format Franchising, Managerial and Decision Economics, 14 (2), 17590. Shane, Scott (2001), Organizational Incentives and Organizational Mortality, Organization Science, 12 (2), 13660. Siebert, Mark (2004), How Fast Should You Grow? (accessed January 12, 2011), [available at http://www.entrepreneur.com/franchises/ franchisingyourbusinesscolumnistmarksiebert/article74168.html]. Srinivasan, Raji (2006), Dual Distribution and Intangible Firm Value: Franchising in Restaurant Chains, Journal of Marketing, 70 (3), 12035. Stigler, George J. (1958), The Economies of Scale, Journal of Law and Economics, 1, 5471. Stump, Rodney L. and Jan B. Heide (1996), Controlling Supplier Opportunism in Industrial Relationships, Journal of Marketing Research, 33 (4), 43141. Thompson, R. Steve (1994), The Franchise Life Cycle and the Penrose Effect, Journal of Economic Behavior and Organization, 24 (2), 20718. Wadsworth, Frank H. (2002), Multi-Unit Owners Study, Washington, DC: IFA Educational Foundation. Welch, Jack and Suzy Welch (2007), Lay Off the Layers, BusinessWeek, June, 96. Williamson, Oliver E. (1967), Hierarchical Control and Optimum Firm Size, Journal of Political Economy, 75 (2), 12338.

Holmstrom, Bengt R. and Jean Tirole (1989), The Theory of the Firm, In Handbook of Industrial Organization, Vol. 1, Schmalensee R. and Willig R. D. eds. Amsterdam: North Holland, 61127. Johnson, Darrell (2006), State of Franchising: Economic Review, Presented at the Multi-Unit Franchising Development Conference & Expo, Las Vegas, NV. Kalnins, Arturs and Francine Lafontaine (2004), Multi-Unit Ownership in Franchising: Evidence from the Fast-food Industry in Texas, RAND Journal of Economics, 35 (4), 74761. Kaufmann, Patrick J. and Rajiv P. Dant (1996), Multi-Unit Franchising: Growth and Management Issues, Journal of Business Venturing, 11 (5), 34358. and (2001), The Pricing of Franchise Rights, Journal of Retailing, 77 (4), 53745. Lafontaine, Francine (1992), Agency Theory and Franchising: Some Empirical Results, RAND Journal of Economics, 23 (2), 26383. Lafontaine, Francine (1995), Pricing Decisions in Franchised Chains: A Look at the Restaurant and Fast-Food Industry, Working Paper No. 5247, NBER. Lafontaine, Francine and Margaret E. Slade (1997), Retail Contracting: Theory and Practice, Journal of Industrial Economics, 45 (1), 125. Lal, Rajiv (1990), Improving Channel Coordination Through Franchising, Marketing Science, 9 (4), 299318. Love, John F. (1995), McDonalds: Behind the Arches, New York: Bantam. Lowell, Brett (2007), Which Franchising Structure Fits Your System? (accessed January 17, 2011), [available at http://www.franchise.org/ Franchise-News-Detail.aspx?id=33626]. Lutz, Nancy A. (1995), Ownership Rights and Incentives in Franchising, Journal of Corporate Finance, 2 (12), 10331. Maltz, Elliot and Ajay K. Kohli (1996), Market Intelligence Dissemination across Functional Boundaries, Journal of Marketing Research, 33 (1), 4761. Martin, Robert E. (1988), Franchising and Risk Management, American Economic Review, 78 (5), 95468. Mathewson, George F. and Ralph A. Winter (1985), The Economics of Franchise Contracts, Journal of Law and Economics, 28 (3), 50326. McKee, Lynette, Ronald Lovejoy and Barbara Moran (2004), Overview of Multi-Unit Franchising, Presented at the Annual Convention of the International Franchise Association, Las Vegas, NV.