Paper to be presented at the EMNet-Conference on
"Economics and Management of Franchising Networks"
Vienna, Austria, June 26 – 28, 2003
Contract design and range of services in franchising:
An empirical contribution to the matching of franchisors and franchisees
Prof. Dr. Thomas Ehrmann
Institut for Entrepreneurship and Business Development
Director Centre for Franchising and Cooperation, Westfälische Wilhelms-Universität, Leonardo Campus 18, 48149
Münster
Germany
Tel.: +49-(0)251-833 8330
E-Mail: ,
Abstract
A franchise system has to manage a “stable match” (Alvin Roth) between franchisees and franchisors. Given, that in reality there are both franchisees and franchisors of different quality. Both parties have the objective to find their “optimal” match with their “quality partner”. This matching has to solve an information problem: Franchisor can not observe the future qualities of the franchisees and vice versa. So franchisors of a particular quality have to differentiate themselves from their competitors. To attract good franchisees and to overcome the observation problem they can use specific contractual provisions and service offerings. The signaling literature has made predictions how “good” franchisors, who have not yet established their reputation, can signal their type by offering a specific contract, with low fixed fees and high royalties (F. Lafontaine). But very little empirical work has been done to explain how franchisors select their franchisees. As a measure for a successful match I use the average revenue of a franchisee (dependent variable). The following hypotheses were tested concerning the adequacy of contractual provisions and service offerings of a franchisor as a solution to the matching problem:
-the higher the fixed fee the higher the revenue of a franchisee (because of the market value of the trade name).
-the lower the royalty rate the higher the revenue of a franchisee (because of the selection of successful franchisees).
-the higher the degree of freedom for individual franchisees the higher the revenue of a franchisee (because of his necessary entrepreneurial capabilities).
-the higher the initial investment, the higher the revenue of a franchisee (because of the self-selection of highly capable franchisees).
These hypotheses are tested using a questionnaire send to a representative sample of 1211 franchise systems operating in Germany (response rate: 38%). The OLS-regression results offer strong support to the tested hypotheses. The results show that high fixed fees combined with low variable fees and a certain degree of freedom for franchisees are positive indicators for a high individual revenue of a franchisee and hence for a stable match. The paper helps to close some of the empirical gaps left open by the signaling approach. It enhances the understanding of the matching process in franchise systems and gives way to further research on this dynamic process.
- Introduction
A major problem in franchise systems is the mutual selection of franchisors and franchisees. The signalling literature discusses ways by which little-known franchisors behave to attract potential franchisees. Equally important is how franchisors select their franchisees. In real life franchisors as well as franchisees differ in terms of quality. It is in both parties interest to find an optimal stable matching that brings together pairs of franchisors and franchisees that correspond in terms of their respective qualities. A solution to this matching problem must overcome problems of observability: Neither can the franchisor observe the quality of its future franchisees directly, nor is it possible for franchisees to observe the quality of little-known franchise systems. Therefore the main challenge for franchisors is to design contracts that differentiate themselves from franchise systems of lower quality. Furthermore they have to attract those franchisees that correspond to their level of quality. (Sales per franchisee (dependant variable) will serve as a proxy for successful matching as it simultaneously represents both successes of the franchisor and of the franchisee.) Royalty fees and support services from the franchise system to franchisees will serve as independent variables.
In a recent study Brickley (2002) pointed to the gap of empirical evidence on royalties in franchise contracts. Prior work of Lafontaine (1992, 1993) tested a two-sided moral-hazard-model. Two important streams in the theoretical literature are work on signalling theory on the one side (Gallini/Lutz (1992); Brickley (2002); Lafontaine (1993)) and incentive mechanisms and reputation on the other side (Lafontaine (1992, 1993)). The present study presents an explorative contribution towards an integrated view on matching.
The study interprets the consequences of contract design in franchise chains in terms of sales (see Fritz (1995) and Kieser/Nicolai (2002) for further evidence on firm success). Therefore the royalty rate dependant on sales and the initial lump sum fee will be investigated. Furthermore the impact of the franchisors activities and policies on the sales volume of the franchisee will be studied. (Sales per franchisee will serve as a proxy for successful matching as it simultaneously represents both successes of the franchisor and of the franchisee.) First, some results from the theoretical literature will be reviewed and empirical implications derived (2.). Second, hypotheses are generated and tested on a German data set. The data is presented following the literature review (3.) and prior to the empirical results (4.). The paper will be closed by a summary and an outlook (5.).
- Signalling, reputation and entrepreneurial freedom: theoretical foundations and empirical implications
- Theoretical foundations
Manufacturers (as principals) have multiple possibilities regarding the organizational form of their distribution system. The extreme positions are a manufacturer-owned distribution system on the one hand and retailing through independent entities on the other hand. Franchising is a hybrid form of retailing combining a contractual relationship of the manufacturer with independent retailers. The choice between a manufacturer-owned and independent retail system is determined by its contractual adequacy of the respective solutions. Assuming that the agent’s effort as well as the observability of output data (e.g. sales) plays a key role in choosing the contractual regime (Lafontaine/Slade (1998)), the following holds: With rising importance of the agent’s effort vertical integration becomes more interesting relative to franchising if output is more easily observed (vice versa). If output data (e.g. sales) are more easily detected as the input (i. e. effort) then franchising becomes more interesting (Lafontaine/Slade (1998)).
From a contract theory perspective franchising is seen as a solution to the one-sided incentive problem referring to the agent not performing in a way that is in the best interest of the manufacturer. With risk-neutral franchisees and non-credible capital constraints a solution to the one-sided incentive problem implies no royalty fees (Holmström (1979), Kreps (1990)). The relative importance of the franchisor for sales in the system is not relevant for an explanation of the royalty fee being dependant on sales.
An extended analysis considers the double-moral-hazard which integrates the incentive problems of both the franchisor and the franchisees. (The following is based on the model of Bhattacharyya/Lafontaine (1995)).[1] The model assumes that franchisor and franchisees influence the demand for the product by their respective efforts:
,
with a, A representing the level of effort chosen by the franchisee and franchisor respectively; g (a, A) represents the effort-output function. is a random variable with an expected value of zero that cannot be observed by the contracting parties. The effort of franchisor and franchisee enhance sales with:
and .
Maximising the franchisor’s expected profit (where f is the royalty fee, F the initial lump sum fee, and C the cost):
, subject to the franchisee’s profit being at least his reservation profit:
.
In the optimum marginal profits are equal to marginal costs:
.
The franchisees participation constraint is just fulfilled. For a detailed analyses of the optimal contract the production function has to be specified (Bhattacharyya, F. S./Lafontaine, F. (1995)). It is assumed that the effort-output relation can be represented by a neoclassical production function (Cobb-Douglas):
,
where and being the sales elasticity with respect to the effort of franchisee and franchisor respectively. is a parameter to represent the sales potential in the market.
The cost functions are specified by:
and .
The cost elacticities indicate by how much (in percent) costs of effort rise with the effort level. The optimal running royalty fee is then given by:
.
The royalty fee is equal for all franchisees that exhibit the same cost function with respect to its power (Bhattacharyya./Lafontaine (1995); Hempelmann, B. (2000)). The royalty fee is independent of the market size (K) but dependent on the effort levels of the contracting parties.
Depending on the activities and the business framework in which f and F are paid their values will vary accordingly (Brickley (2002)/Lafontaine (1992)). Correspondingly these variables can have differing effects on the success of the franchisee. In the model of Bhattacharyya and Lafontaine,(1-f) and f stand for the activities of franchisee and franchisor respectively. Depending on the relative importance of f for the generation of sales, f will be influenced in favour of the franchisor or the franchisee.
F should take a level to extract all of the downstream rents from the franchisees; where a running royalty fee is assumed (Lafontaine (1992), Dnes (1992), Dnes (1996), Klein (1995)). F should correspond to the net present value of the franchisee’s future profits (for every given level of sales and royalty fees (Lafontaine (1992)). In other words: F should be correlated positively to the total value of the franchise system. This total value represents the characteristics of the local market, the value of the brand, the number of franchisees as well as the market power of the system. f on the other side stands for the activities, e. g. marketing effort etc., that can be influenced by franchisors and franchisees to ascertain the market success. The question to raise is what kind of effects result from which part of the fees.
From the above reasoning follows that F and f should be correlated negatively when no rents remain with the franchisee (Lafontaine (2002), Ehrmann (2002)). The effect of different levels of f and F on the matching of franchisors and franchisees is ambiguous. This question is particularly interesting for potential franchisees who are keen on buying into a system without being able to measure their future entrepreneurial success ex ante (Klein (1995), Kaufmann/Lafontaine (1994a)). From this point of view the two parts of the fee structure are particularly important.
2.2. Empirical implications
2.2.1. General Reasoning
When selecting each other franchisors and franchisees do not observe the other parties’ qualities. Observable are f and F. Furthermore can sales of franchisees already in the system and sales of the overall system be observed. Observable is also the level of effort by the franchisor supporting franchisees more or less and leaving them different degrees of entrepreneurial freedom. A good and stable matching is realised when no franchisees or franchisor would prefer being matched to another party (Roth (1999)).
Then the following holds: good but little-known systems have to differentiate from less attractive systems. To signalise their respective quality, they should offer a contract V = V(f,F) that low quality franchisors cannot copy. The quality of good franchisors and franchisees will then be observable by higher sales than in comparable but low quality systems.
2.2.2. Empirical implications
In order for a matching to succeed incentives and reputation are important. By combining incentive and reputation arguments further hypotheses regarding the effect of different fee structures on the success of franchisees can be formulated. The chances for success of a franchisee should be reflected by the total value of the franchise system, measured by the brand value or market power. A high value should correspond to a high initial lump sum fee. At the same time a high initial lump sum fee should represent a self-selection criterion for good franchisees who expect to recoup the outlay in the future (Sen (1993)). Following this reasoning the following hypotheses should hold:
H1: The higher the initial lump sum fee, the higher the franchisee’s sales level.
Given that the initial lump sum fee valued by the market is a prerequisite for the franchisee’s success, then the level of the royalty fee is important for the achieved sales volume. Lower running fees allow franchisees c.p. to offer lower prices to customers than higher taxed franchisees. Higher royalty fees are than a reason for lower sales. Good franchisees should also offer other services (marketing) at low prices. Thereby good franchisees have good incentives for profit making. Thus, the following statement should hold:
H2: The lower the royalty fee, the higher the franchisee’s sales level.
Furthermore the assumption is that good franchise systems equally guarantee the success of the franchisor by selecting good franchisees as they offer good franchisees opportunities for success. Services that are delivered by the system should relate positively to the sales volume of franchisees:
H3: The higher the sum of delivered services, the higher the franchisee’s sales volume.
A classical task of franchise systems is the central administration of business planning which is a part of the support services that are well suited to be centralised. Therefore:
H4: Business planning activities should have a positive effect on the franchisee’s sales volume.
From the reputation and incentive reasoning follows, (as opposed to the signalling theory,) that franchisors attribute high importance to the selection of good franchisees. This may imply that these franchisees are entrepreneurial and able to take decisions on their own. Besides tasks that are suitable for central planning some may only be executed effectively on site. It may be that a higher degree of organization, i.e. more policies and manuals, affects franchisee’s sales negatively. Therefore:
H5: The lower the degree of organization, the higher sales by franchisees.
3. The data
The underlying data of the study where provided by the Peckert-Group. The data was initially collected for the Jahrbuch Franchise und Lizenz 2002 (FAZ-Verlag) that gives an overview on 500 franchise systems out of 1240 that were initially addressed. The franchise systems that were included in the study (Table 1a) were extracted from the initial sample according to the variables studies here. The distribution of the systems on different industries is typical for franchise systems. Through the focus on the retailing and services industries franchise systems are generally open for lateral entry.
4. Empirical results
For the falsification of the above stated hypotheses regarding signalling on the one hand and incentives and reputation on the other hand the following model was tested by multiple linear regressions:
+7 ALT with
U = sales by a franchisee
EG = initial fees
LG = royalty rate
SL = sum of services
KPLA = business planning instruments
ORG = degree of organization
INVEST = requirements on the initial investment by franchisees
ALT = age of the system
The degree of organisation was established using responses on a five-point scale (high-low) regarding the standardisation of procedures through the franchisor. The sum of all services was identified by summing (unweighted) up all services offered by the franchisor.
As business building instruments the following subset of support services was considered:
Group: General
-enterprise resource planning system
-Research & Development
-Market analyses
-QM-systems
-Collective purschase
Group: “Building up of the company”
-investment planning
-human resource management
Group: „Development of the company“
-innovation manegement
-intercompany comparison
-budgeting / controlling
-consulting
-marketing consulting
The sum of positive responses was added (unweighted). The results figure in table 2.
Coefficients(a)Non-standardized coefficients / Standardized
coefficients / T / Significance
Model (R-Square= 0,35)
F = 7,758* / B / Standard error / Beta
1 / (Intercept) / 2662,022 / 683,517 / 3,895 / ,000
Initial fees / 45,337 / 9,735 / ,457 / 4,657 / ,000
Royalty rate (% / sales) / -105,321 / 31,667 / -,321 / -3,326 / ,001
Sum of services
(Sum of General., Building, Development) / -79,065 / 29,935 / -,519 / -2,641 / ,010
Business planning instruments (see annexe) / 155,871 / 69,077 / ,453 / 2,256 / ,027
Degree of organization / -411,151 / 171,642 / -,244 / -2,395 / ,019
a Dependent variable: Gross sales T €
Table: 2
* sign. at 0.1% level
Descriptive StatisticsMean / Standard deviation / N
Gross sales T € / 659,08 / 974,13 / 78
Initial fees / 13,45 / 9,83 / 78
Royalty rate (% / sales) / 5,14 / 2,97 / 78
Sum of services
(Sum of General., Building, Development) / 21,3846 / 6,3985 / 78
Business planning instruments (see annexe) / 7,9744 / 2,8283 / 78
Degree of organization / 3,95 / ,58 / 78
The coefficients confirm the hypotheses resulting from the incentives and reputation arguments. They indicate that good franchise systems are looking for very active and entrepreneurial franchisees. An interesting result is that both support services (exceeding business planning in its classical-centralised form) and a high degree of organization affect sales negatively. One could object that the results are biased towards “old” systems that do not have to signal their quality by the contract V (f, F) as “new” systems have to.
For checking this argument a further model was tested including age as an independent variable. Should the coefficient of the variable age be positive and significant there would be an indication for such a bias. Table 3 shows that there is a positive sign attached to the variable age. However the correlation is weak and not significant. Therefore, there is no case for a bias in the data.
Coefficients(a)Non-standardized coefficients / Standardized
coefficients / T / Significance
Model (R-Square = 0,414)
F = 7,239* / B / Standard error / Beta
1 / (Intercept) / 3019,303 / 727,978 / 4,148 / ,000
Initial fees / 53,197 / 10,413 / ,524 / 5,109 / ,000
Royalty rate (% / sales) / -155,326 / 41,033 / -,398 / -3,785 / ,000
Sum of services
(Sum of General., Building, Development) / -84,025 / 34,348 / -,537 / -2,446 / ,017
Business planning instruments (see annexe) / 137,847 / 80,439 / ,380 / 1,714 / ,092
Degree of organization / -418,045 / 179,359 / -,242 / -2,331 / ,023
Age of franchise system / 7,604 / 12,612 / ,061 / ,603 / ,549
a Dependent variable: Gross sales T €
Table: 3
* sign. at 0,1% level
Descriptive StatisticsMean / Standard deviation / N
Gross sales T € / 710,96 / 1023,54 / 69
Initial fees / 13,49 / 10,08 / 69
Royalty rate (% / sales) / 4,90 / 2,62 / 69
Sum of services
(Sum of General., Building, Development) / 21,3478 / 6,5436 / 69
Business planning instruments (see annexe) / 8,1304 / 2,8228 / 69
Degree of organization
/ 3,97 / ,59 / 69Age of franchise system / 8,8841 / 8,1863 / 69
The algebraic sign of the variable initial lump sum fee indicates that franchisees have a good idea about their own skills as well as concerning the market value of the franchise system. The royalty fee variable indicates that incentive and price effects dominate signalling effects. The negative effects of a higher royalty fee on the franchisees sales illustrate the costs of such a kind of signalling. It seems to be the case that signalling of good quality to the franchisees could be achieved more cost efficiently (licensing etc.) than by rising the running royalty fee.