How Firm Strategies Impact Size of Partner-Based Retail Networks: Evidence From Franchising

Abstract

How do firms’ partnering strategies impact the size of their partner-based retail networks? We draw on agency theory to address this question in the context of franchising. Our econometric analyses (based on 9 years of longitudinal balanced panel data) include assessment of data nonstationarity and estimation of a dynamic panel data model that accounts for unobserved heterogeneity and endogeneity. Our findings indicate that franchisee network size is driven more by franchisor strategies that mitigate agency costs than by strategies that simply lower entry and ongoing costs and barriers for franchisees.

Notes

1 Business format franchise systems (for example, McDonald's, Pizza Hut, KFC, Dunkin' Donuts, Applebee's, H & R Block, Century 21 and ServiceMaster) are franchise systems where the contractual arrangement “includes not only the product, service, and trademark but also the entire business format itself—a marketing strategy and plan, operating manuals and standards, quality control, and continuing two way communication” (U.S. Department of Commerce 1987, pp. 3). In return for the right to use the franchisor's business format, a franchisee typically pays an initial upfront franchisee fee as well as ongoing royalty and advertising fees (Dant and Berger 1996).

2 For expositional clarity, we use the term “franchisee network” to refer to the network of franchisees in the chain and the term “total network” to refer to the totality of franchised and franchisor-owned outlets in the chain.

3 Kosová and Lafontaine (2010) also measured performance of just the franchised component of franchised systems. However, they measured this performance in terms of exit from franchising and the growth rate of the franchisee network rather than franchisee network size. Instead, they use network size as an explanatory variable.

4 A balanced panel includes data for every year for each firm in the panel. In contrast, in an unbalanced panel, observations may be missing for one or more years for one or more firms in the panel.

5 For expositional simplicity and consistency with the extant franchising literature, we hereafter refer to precontractual/hidden information agency problems as adverse selection problems and postcontractual/hidden action agency problems as moral hazard problems.

6 An important distinction between quality signaling in the broader agency theory literature and franchisor quality signaling revolves around what constitutes “quality.” In a general agency setting, quality often refers to fixed traits of the signaling entity that will impact the outcomes from the exchange relationship. In franchising, as suggested by Lafontaine (1993), franchisor quality is better viewed in terms of the future behavior of the franchisor. It is this future behavior, rather than exogenous franchisor characteristics, that will impact franchisee outcomes over the duration of a long term franchise contract. Therefore, franchisor signaling is more meaningful when it provides an indication of the future behavior of the franchisor.

7 We are grateful to two anonymous reviewers whose insights have enabled us to strengthen the theoretical reasoning underlying our hypotheses.

8 The presence and level of an advertising fee reduces the franchisee's share of outlet revenues. Prima facie, this creates greater incentives for franchisee free riding on other inputs. However, this incentive may be offset if the revenue creating impact of the franchisee's inputs is amplified by the greater levels of advertising and brand building undertaken by the franchisor (as a consequence of the higher advertising fee).

9 It can be argued that higher initial fees could lead to greater franchisor moral hazard since the franchisor-specific nature of a franchisee's initial investments exposes it to holdup by the franchisor. Dnes (1992) discounted this argument by noting that franchise contracts are often written to ensure that franchisor appropriation of franchisee assets and investments is restricted to situations of franchisee failure. Indeed, this element of franchise contracts may instead alleviate the franchisee moral hazard problem by motivating franchisees to succeed in operating their franchised outlets.

10 Shane (1998) drew on agency theory to hypothesize a positive relationship between franchise concept complexity and franchise system failure. He posited that increases in complexity result in higher monitoring costs. Note that Shane (1998) looked at a different pool of franchisors (new franchisors only) and a different performance variable (franchise system failure). It is possible that the complexity of a franchise concept may create monitoring challenges for a new franchisor but that, if it survives and operates for some time, the franchisor will develop the ability to effectively undertake this monitoring.

11 There are arguments in the franchising literature for positive effects of the use of master franchising on franchisee moral hazard. For example, Shane (1998) posited and found a positive relationship between the use of master franchising and franchise system failure. However, this view must be considered in the context of the sample of new franchisors studied by him. Shane (1998) noted that the use of master franchising requires codification of enforcement behavior and that this enforcement behavior must either be specified at the time of contracting or be foregone. This is particularly challenging for a new franchisor that is likely to be unable to foresee all possible mechanisms for franchisee shirking. We note, again, that we differ from Shane (1998) in terms of the types of franchisors (new franchisors only versus franchisors of all ages) and organizational performance measures (franchise system failure versus franchisee network size) studied. In our assessment, if a franchisor survives and operates for some time (and, in the process, acquires knowledge about different forms of franchisee shirking), it may be in a better position to codify enforcement behavior in master franchise agreements and realize the franchisee moral hazard reducing benefits of master franchising.

12 No Guides were published for the years 1986, 1987, 1990, and 2000.

13 Although our panel data set covers a number of industries, our fixed effect estimation approach (that accounts for unobserved heterogeneity) alleviates the need for industry-specific control variables in our model.

14 Econometric analysis has traditionally consisted of cross-sectional analysis, time series analysis, or panel data analysis with a small and fixed time series dimension. There has been a growing interest in studying cross-sectional data over time, entailing the use of panel data models with both a large number of cross-sectional units and a large number of time series observations. When working with panel data that has a large time series dimension, we gain additional power over traditional time series analysis from the increased observations in the cross-section dimension, but we must deal with potential nonstationarity in the time series dimension of the data. Recent research has improved our ability to analyze nonstationarity, cointegration, and the spurious regression problem in panel data. These issues have been examined extensively in pure time series (Engle and Granger 1987), but only recently have they been studied in detail in panel data models. These new panel data methods are extensions of the traditional time series methodology, using the additional information gained from the cross-section dimension of the panel. Testing for unit roots in pure time series studies is a common practice among applied researchers. For the nontechnical reader, it is important to point out that the terms unit root, nonstationarity, and random walk (process) mean the same thing and can be used interchangeably. These terms mean that the data series under consideration can be written as y_t = y_t−1 + ε_t, for the case of a random walk without drift or y_t = α + y_t−1 + ε_t, for the case of a random walk with drift (constant term) where y_t is the data series and ε_t is white noise (Hamilton 1994).

15 In the interests of brevity, we have not included the re-estimation results in the paper. These results can be requested from the authors.

16 We are grateful to an anonymous reviewer for another potential explanation for the significant negative relationship between the initial fixed fees and franchisee network size. This explanation revolves around the chronological context of our data. Our data set covers the period 1995–2004. This was the time during which property prices experienced substantial appreciation in the United States and represented an attractive asset class for investments. The return on investment offered by the franchises with high initial fixed fees may have been less attractive compared with property investment alternatives. At lower levels of initial fixed fees, prospective franchisees may not have had similarly attractive property investment alternatives available to them.

Additional information

Notes on contributors

Manish Kacker

Manish Kacker is Associate Professor, Marketing at DeGroote School of Business, McMaster University.

Rajiv P. Dant is Professor of Marketing and Helen Robson Walton Centennial Chair in Marketing Strategy at Price College of Business, The University of Oklahoma and Professor of Marketing, Griffith University.

Rajiv P. Dant

In memory of Dr. Rajiv P. Dant, with respect and gratitude. Dr. Dant passed away after this paper had been accepted and published online by the journal.

Jamie Emerson

Jamie Emerson is Associate Professor of Economics at Perdue School of Business, Salisbury University.

Anne T. Coughlan

Anne T. Coughlan is the Polk Bros. Chair in Retailing, and Professor of Marketing, at the Kellogg School of Management, Northwestern University, Evanston, IL 60208.