investments to create bargaining power: the case of franchising

Strategic Management Journal
Strat. Mgmt. J., 21: 497–514 (2000)
INVESTMENTS TO CREATE BARGAINING POWER:
THE CASE OF FRANCHISING
STEVEN C. MICHAEL*
College of Commerce and Business Administration, University of Illinois at UrbanaChampaign, Champaign, Illinois, U.S.A.
Hybrid organizational forms such as franchise systems join two or more independent parties
under a contract. The ability of each party to achieve its goals depend upon the relative
bargaining power in the relationship established by the contract. Using transaction cost
economics and Porter’s (1980) characterization of sources of bargaining power, this paper
argues that the franchisor can make investments in activities such as tapered integration and
buyer selection to increase its bargaining power and decrease conflict and litigation in a
franchise system. Specifically, tapered integration (owning some units while franchising others),
selecting inexperienced franchisees, and employing a long training program are predicted to
increase the franchisor’s bargaining power and the franchisee’s compliance with franchisor
standards. An empirical analysis of litigation in restaurant franchise systems supports the
theoretical hypotheses. Copyright  2000 John Wiley & Sons, Ltd.
Hybrid organizational forms such as joint ventures, strategic alliances, and franchise systems
are more and more common today. Such organizational hybrids typically are formed by contract
between two or more legally independent parties.
The ability of each party to achieve its desired
outcome depends upon its relative bargaining
power (Porter, 1980) within the framework created by the contract. Most of the factors determining bargaining power (as listed in Porter, 1980)
are hard to change, depending upon characteristics
of the production process, industry characteristics,
or volume of purchases. But investments to
improve bargaining power are possible. One of
the most commonly suggested methods for a firm
to alter its bargaining power is through tapered
integration, making some units and buying others
Key
words: bargaining
power;
franchising;
entrepreneurship; industry analysis; vertical integration
*Correspondence to: Steven C. Michael, College of Commerce and Business Administration, University of Illinois at
Urbana–Champaign, 1206 S. Sixth Street, Champaign, IL
61820, U.S.A.
Copyright  2000 John Wiley & Sons, Ltd.
(Porter, 1980; Harrigan, 1983, 1984, 1985, 1986;
Buzzell, 1983; MacMillan, Hambrick, and Pennings, 1986; Day, 1990). Tapered integration
gives the firm a way to gather information about
costs to be used in purchasing, while still preserving incentives for internal and external suppliers.
A second method is buyer selection (Porter,
1980): choosing buyers that have less intrinsic
bargaining power or are less likely to use it.
Drawing on transaction cost economics and strategic management theory, this paper offers an
explanation and a test of how, in the context of
franchising, firms can make ex ante investments
in tapered integration and buyer selection that
improve their bargaining power ex post and avoid
disputes and litigation in a contractual relationship.
Franchising, the oldest interorganizational form,
provides an excellent venue for such a test. The
franchise system is an organizational structure
governed by contract between a parent company,
the franchisor, and a local outlet, the franchisee,
established to sell products or services under the
parent’s trademark. The franchisor and franchisee
Received 23 December 1996
Final revision received 24 August 1999
498
S. C. Michael
are legally independent but economically interdependent. Issues such as tapered integration and
buyer/franchisee selection have not been examined through the theoretical lens of bargaining
power, although each has been examined in
other ways.
Explanations for tapered integration in franchising, where the franchisor owns some units while
franchising others, have not focused on the benefits associated with altering bargaining power.
Bradach and Eccles (1989) conjecture that owning units can improve information available to
franchisors. Empirical researchers have concluded
that franchisors prefer to own all their stores, but
site heterogeneity makes franchising preferred for
some locations.1 For example, geographic distance prevents effective monitoring, so they franchise distant units (Brickley and Dark, 1987;
Norton, 1988; Carney and Gedajlovic, 1991;
Fladmoe-Lindquist and Jacque, 1995). Areas with
more transient customers present more opportunities to free-ride on the brand name, so the
franchisor owns more units in those areas
(Norton, 1988; Carney and Gedajlovic, 1991). As
compelling as this geographic explanation is, it is
likely incomplete without considering the possible
effect of tapered integration on bargaining power.
First, the franchisor repeats this transaction many
times, as it sells new franchises and recruits new
franchisees. Given the frequency and similarity
of the transaction, it is possible that the opportunity for strategic interaction affects the choice
of whether to own or franchise. The choice for
one transaction may influence another. Second, it
does not explain the use of both owned and
franchised outlets in the same geographic or market area, where sites are most likely to be similar,
which has been documented by Kalnins and
Lafontaine (1996). Third, the explanation is
inconsistent with the practices of franchisors. Survey research reveals that most franchisors with
extensive franchising operations also desire to
own some units and in practice do own some
units (Lafontaine and Kaufmann, 1994: 105; also
Lafontaine, 1992a). According to franchisors,
owned units ‘provide a window on the industry’,
‘facilitate market planning’, or promote ‘control
1
In a theoretical model Gallini and Lutz (1992) argue that
franchisors may own units to signal the profitability of the
franchise concept, but an empirical test by Lafontaine (1993)
did not support that signalling model.
Copyright  2000 John Wiley & Sons, Ltd.
and consistency’ (Lafontaine, 1992a). As a result,
considering whether bargaining power can be
enhanced by tapered integration is suggested by
these observations.
With regard to selection, survival studies have
indirectly considered selection by arguing that
certain franchisee and contractual characteristics
will identify prospective franchisees less likely to
shirk, and selecting for such characteristics will
enhance survival of the franchise system (Shane,
1998). Whether such measures do influence conflict and increase quality has not been observed,
however. In addition, a few studies have examined why individuals become franchisees (e.g.,
Kaufmann and Stanworth, 1995; Peterson and
Dant, 1990), while others have considered conflict
in franchise systems (e.g., Phan, Butler and Lee,
1996; Baucus, Baucus, and Human, 1996). But
no one has examined the role of franchisee selection ex ante in reducing conflict and litigation
in franchise systems ex post.
Investing in bargaining power: Theory
In franchising, the fundamental problem faced by
the franchisor is to assure the quality of the
consumption experience represented by the trademark that is owned by the franchisor and used
by the franchisee (Caves and Murphy, 1976;
Rubin, 1978; Shane, 1996). In this setting, quality
means not just product quality, but conformance
to the franchisor’s operating instructions for both
production and delivery of the standardized product. The use of the trademark in common creates
the risk of opportunism through free riding by
the franchisee (Caves and Murphy, 1976; Rubin,
1978; Klein, 1980). If any single franchisee
relaxes his effort to produce high quality, customers will still patronize his unit, assuming that
his quality is identical to others sharing the trademark. The franchisee can avoid the cost of quality
while gaining from the investments in quality of
the franchisor and of other franchisees.
The franchisor must resolve the conflict
between the franchisee’s desire to put forth less
than full effort for quality and the franchisor’s
demand for quality. Administrative fiat will not
suffice; the franchisee is not an employee. The
franchisee operates a legally independent business
under contract from the franchisor. Instead, conflict must be resolved through negotiations or
legal remedies under the contract. The contract
Strat. Mgmt. J., 21: 497–514 (2000)
Investments in Bargaining Power
law that enables and underpins franchising is
characterized as ‘neoclassical’ contract law
(Williamson, 1991, drawing on Macneil, 1978,
1980). ‘Neoclassical contract law . . . [applies] to
contracts in which the parties to the transaction
maintain autonomy but are bilaterally dependent
to a nontrivial degree.’ (Williamson 1991: 271.)
Neoclassical contract law is generally more elastic
than classical contract law. Rather than specifying
explicit and formal terms for all conditions, the
neoclassical contract creates an ‘adaptive range,’
a framework and a set of boundaries, within
which conflicts are resolved through negotiation
between the parties (Williamson, 1991). Negotiation within the adaptive range rather than literal
adherence to contract terms facilitates adaptation
to change and the preservation of the relationship.
Such a negotiation is always carried out in the
shadow of the law; when the parties cannot agree
within the adaptive range, they resort to the courts
(Macneil, 1980; Williamson, 1991). The
franchisor generally has the right under the contract to terminate the relationship with franchisees
through litigation (Rubin, 1978; Muris, 1981;
Hadfield, 1990; Brickley, Dark, and Weisbach,
1991; Bond and Bond, 1991). The franchisor
will pursue termination and litigation when the
expected gains net of costs from negotiation
within the adaptive range are less than the
expected gains net of costs from litigation. Termination is costly, however, and more and more
states are regulating this practice (Pitegoff, 1989).
As one example, in many states termination
requires ‘good cause,’ a term that lacks clear
statutory definition, and therefore invites litigation
(Pitegoff, 1989). Moreover, termination may be
unwise for competitive reasons; the franchisor
may lose a valuable geographic location, for
example. The costs of litigation and termination
create an incentive for investments in improving
franchisor bargaining power in order to achieve
desired quality within the adaptive range of the
contract.
To increase bargaining power, tapered integration is frequently recommended: ‘A great deal
of bargaining leverage can be gained through
tapered integration’ (Porter, 1980: 125). Tapered
integration is defined as ‘Some portion (but not
all) of the firm’s requirements for an input is
supplied in-house or some portion of outputs is
sold (consumed) in-house.’ (Harrigan, 1984:
645.) Tapered integration is an intermediate
Copyright  2000 John Wiley & Sons, Ltd.
499
degree of vertical integration, between nonintegration and full integration. (For the purposes of
this paper, franchisors are described as forward
integrating when they operate units, although the
arguments apply to both forward and backward
tapered integration.) Gains in bargaining power
from tapered integration come from two sources,
incentives and information, as Porter (1980)
describes.2 ‘Not only is the threat of further
integration particularly credible, but also partial
manufacture in-house gives them [the firm considering integration] a detailed knowledge of
costs which is a great aid in negotiation.’ (Porter,
1980: 25.) Making some units while buying
others gives considerable power to the procuring
firm while weakening the power of buyers.3
By employing tapered integration, the
franchisor would be likely to raise bargaining
power and become more likely to be able to
reach agreement with the franchisee within the
adaptive range of the contract. Litigation and
termination become less likely. Tapered integration both improves the franchisor’s information
and demonstrates an ability to further integrate if
necessary. First, ownership gives the franchisor
information. The franchisor can more realistically
set standards with knowledge of costs and
demand. With ownership, the franchisor can predict the costs of quality with more precision, and
can judge whether a particular quality standard
imposes costs that cannot be sustained by demand
levels (Lafontaine, 1992a).4 Also, owned units
give information regarding demand levels, customer preferences, and the like (Minkler, 1992;
Lafontaine, 1992a). Such information can credibly
demonstrate knowledge of operations, and provide
a comparison to measure franchisee relative performance (Anand, 1987; Bradach, 1997).5
2
For further discussion of bargaining leverage and bargaining
power, Porter (1976, chapter 2) and Porter (1980, chapters 1
and 6) outline economic mechanisms and Stern and Reve
(1980) suggest sociological mechanisms. The idea of tapered
integration creating bargaining power dates back to at least
Kessler and Stern (1959).
3
Tapered integration contains risks, too; gains from bargaining
power must be traded off possible retaliation by buyers
(Harrigan, 1984; MacMillan, Hambrick, and Pennings, 1986).
4
It is unlikely that a secondary market for such information
will arise independently of ownership of operations, because
those who might sell the information are also the most likely
to be the victims of it (Arrow, 1975).
5
The existence of such hidden information gives the
franchisee scope for opportunism in the model of Mathewson
and Winter (1985). Minkler (1992) has argued that this
asymmetry in information is the reason for franchise systems.
Strat. Mgmt. J., 21: 497–514 (2000)
500
S. C. Michael
Second, the theory above suggests that tapered
integration can be effective when viewed as a
threat of further integration. In the context of
the ongoing relationship between franchisees and
franchisor, ownership of some units suggests to
the franchisee that the franchisor can and will
operate those units if quality declines. Thus buyer
power is weakened. For such a signal to have an
effect, the threat to integrate must be credible
(Ghemawat, 1991). The transaction between
franchisor and franchisee is repeated many times
with little variance; franchise contracts are
broadly similar across franchisees (Lafontaine,
1992a; Lafontaine and Shaw, 1996). Thus action
taken with regard to one franchisee can signal
behavior to other franchisees.6 Ownership of
some units constitutes the strongest possible signal of commitment. The franchisor has made
sunk investments in the units themselves and
the necessary managerial infrastructure to operate
them. Adding another unit presumably does not
appreciably extend managerial resources. Thus
the franchisee recognizes franchisor operation as
a credible threat. Ownership thus signals to franchisees that the franchisor is committed to quality,
that the franchisor can recognize quality, and that
the franchisor can operate a unit if required.
Recognizing this, franchisees increase their effort
to preserve quality.
This application of theory is broadly consistent
with the practice of at least two major franchisors.
Love (1986: 204) describes the motivation of
McDonald’s in opening company-owned stores.
‘Turner [the chief operating officer] wanted some
company controlled outlets as training ground for
his field consultants. But Ray Kroc hoped the
very presence of the four “controlled units,” as
he referred to them, would encourage wayward
McDonald’s franchisees to clean up their act.’
Management of the hotel chain Holiday Inn
applied the same logic. In the 1991 Bass PLC
annual report, management states, ‘A company
managed hotels division was established as a
strategic business unit [in the past year] to lead
the brand by example.’
In summary, from the franchisor’s perspective,
6
This threat can be viewed as an attempt to shape the
expectations of franchisees in order to assure the functioning
of the franchise system. Malmgren (1961) notes that the
problem of shaping expectations within the firm is a necessary
part of management.
Copyright  2000 John Wiley & Sons, Ltd.
the alternative to litigation is tapered integration:
integration and litigation are substitutes. Ownership of units can raise bargaining power, deter
quality degradation, and reduce the need for litigation. This suggests:
Hypothesis 1: As tapered integration rises,
litigation falls, all other things equal.
The hypothesis implies that owning some units
while franchising others is a persistent characteristic of the system. In contrast, the life cycle
model of the franchise system in the literature
argues that systems franchise early but own later.
According to Oxenfeldt and Kelly (1968–69),
franchisors are managerially and financially constrained in their early years, so they aggressively
franchise early in their existence. As they gain
more resources, franchisors will own units themselves, and reacquire previously franchised units,
moving toward complete ownership of the system. Evidence for the life cycle is not well
established (Lafontaine and Kaufmann, 1994;
Dant, Kaufmann, and Paswan, 1992; Norton,
1995). The evidence reviewed in the first part of
the introduction suggest TI is a persistent structural characteristic, hypothesized here to alter
relative bargaining power (although life cycle
effects may generate deviations from a desired
level). The empirical section will test the stability
of tapered integration, and the empirical model
will control for possible lifecycle effects.
A second investment in improving bargaining
power can be made in buyer or partner selection:
‘Buyer selection can . . . minimize the disruptive
power of buyers.’ (Porter, 1980: 109.) Buyers,
here franchisees, have less bargaining power
when faced with high fixed costs of switching,
including investments in training and psychic
costs of severing a relationship (Porter, 1980:
114, 120). Training is likely to raise switching
costs for the franchisee, because franchisees frequently rely on training to learn the specifics of
the business and the industry and by increasing
the system specific human capital the franchisee
develops (Peterson and Dant, 1990; Kaufmann
and Stanworth, 1995; Roha, 1996). In addition,
empirical research in human resources has demonstrated that institutionalized socialization
through training raises psychic costs of switching
by increasing organizational commitment (Jones,
1986; Allen and Meyer, 1990; Ashforth and Saks
Strat. Mgmt. J., 21: 497–514 (2000)
Investments in Bargaining Power
1996). Training and socialization ‘encourage newcomers [such as new franchisees] to passively
accept preset roles and thus maintain the status
quo (Ashforth and Saks 1996: 150).’ This could
be characterized as shaping the franchisees’
expectations so that franchisees ‘follow the system’ to deliver the common consumption experience the trademark represents (Anand and Stern,
1985; Anand, 1987). By shaping expectations and
establishing ex ante the desired behavior, the
franchisor reduces conflict and subsequent litigation (International Franchise Association, 1988;
Poe, 1990; Bond and Bond, 1991), as well as
produces a more efficient use of effort and knowledge (Malmgren, 1961). It seems reasonable to
presume that training, both by raising switching
costs directly and by a process of socialization,
would expand the adaptive range in the mind of
franchisees and increase franchisors’ power, therefore increasing the desire of franchisees to reach
agreement and avoid litigation. This suggests:
Hypothesis 2: As the length of the training
program rises, litigation falls, all other
things equal.
Selecting less informed partners is likely to
raise bargaining power (Porter, 1980: 117). A
prospective franchisee’s industry experience is
likely to be useful in operations, but it may be
more difficult to socialize or to shape the expectations of a prospective franchisee with industry
experience than one who lacks such experience
(Lafontaine, 1992b). Such a franchisee may be
more demanding of the franchisor and less willing
to follow the franchisor’s system.7 As Stanworth
and Kaufmann (1996: 62) note,
‘Franchisors have often gone on record as preferring potential franchisees from outside the operational line of the franchise in question. . . .
[These franchisees] would appear more likely to
be motivated to defer to the franchisor’s knowledge and authority than those with prior experience in the field.’
In addition, a more experienced franchisee may
have additional resources to bring to the franchise
relationship, and also have more outside alterna7
In popular press articles, general business experience is
offered as an explanation for the growth of litigation in
franchise systems (Harris and France, 1997; Dunkin, 1996).
Copyright  2000 John Wiley & Sons, Ltd.
501
tives to operating the franchise, thus creating
bargaining power for the franchisee (drawing on
Porter, 1980: 113). Therefore industry experience
would appear to enhance franchisee bargaining
power, or at least the franchisee’s perception of
that power, and inhibit agreement within the
adaptive range.8 Therefore:
Hypothesis 3: Systems that require previous
industry experience of their franchisees will
experience higher litigation, all other things
equal.
Aside from the preservation of quality, two
common issues of conflict between franchisor
and franchisee are tying sales and geographic
territories. The franchise contract frequently
requires the franchisee to purchase specific inputs
from the franchisor, termed tying. Franchisors
justify the practice as necessary to preserve quality of inputs (e.g., Klein and Saft, 1985), and
one would expect that tying would reduce litigation by improving quality. But tying is perceived by practitioners as a conflict of interest
between franchisor and franchisee, as the
franchisor seeks to realize value not from the
ongoing success of the franchisee but from the
one-time sale of products (Luxenberg, 1986: 266;
Webster, 1986: 189; Love, 1986: 60ff; Purvin,
1994: 195). For example, Keup (1990: 16) colorfully cautions, ‘A franchise is somewhat similar
to the ball point pen. You give the ball point
pen away and make a fine living on selling the
refills to the recipient if he is required to buy
the refills from you and no one else.’ This hostility might lead to litigation given the status of
tying under U.S. antitrust laws: tying is presumed
illegal unless the franchisor can prove the tie is
necessary for efficiency (Scherer and Ross, 1990;
Lynk, 1994). Therefore, some litigation is likely
to occur to determine whether typing is necessary.
On balance, the practitioner hostility and its
ambiguous status in antitrust law suggest that,
despite any gains in quality it may generate, tying
is likely to create additional litigation to determine its necessity. In addition, the franchisor
8
A referee suggested a counterpoint: industry experience will
help the franchisee know how to preserve quality and reduce
litigation. The extent to which this holds depends upon how
idiosyncratic the production system of the franchise is. The
hypothesis as stated presumes relatively high idiosyncrasies,
considering Stanworth and Kaufmann (1996) above.
Strat. Mgmt. J., 21: 497–514 (2000)
502
S. C. Michael
may exhaust bargaining power persuading the
franchisee of the value of the practice. So:
Hypothesis 4: Systems that require franchisees
to buy a higher proportion of inputs from the
franchisor will experience higher litigation, all
other things equal.
Finally, some franchisors grant an exclusive
territory, a geographic monopoly over a particular
market area. Within an exclusive territory, no
other unit (franchised or company-owned) can be
opened without the franchisee’s permission. By
creating a geographic monopoly, the value of the
franchise is increased, and this increased value is
likely to induce greater effort from the franchisee
(Klein and Leffler, 1981).9 Therefore:
Hypothesis 5: Systems that grant franchisees
an exclusive territory will experience lower
litigation, all other things equal.
These hypotheses will be tested after the data
are described.
DATA SOURCES AND MEASUREMENT
To test the theory of tapered integration, an intraindustry regression was performed to examine the
effect of tapered integration on litigation within
franchisor systems in the restaurant industry. The
source for data on each franchisor is the Uniform
Franchise Offering Circular (UFOC), used by the
franchisor to explain the contract’s terms and
conditions, the litigation history, and the franchise
system to the franchisee. The Federal Trade Commission prescribes the format and the information
to be disclosed. The data come from 99 UFOC’s
in the restaurant industry, obtained through personal survey; all offering circulars but two were
in force during 1990 or 1991. In order to test
for validity of this sample, I compare this sample
to sampling frames that are more complete in
their lists of franchises but less detailed in their
data. This database does not differ in royalty
rate or percentage of tapered integration from a
database built from the annual ‘Franchise 500’ in
the January issue of Entrepreneur magazine
9
I am grateful to the referees for suggesting this variable and
sharpening the analysis.
Copyright  2000 John Wiley & Sons, Ltd.
(1990, 1991) or from a data base built from
Franchise Annual (1990).10 The franchise fees
are slightly higher in this data base. To test the
relevance of the sample to recent sources, the
annual ‘Franchise 500’ for 1998 was consulted.
This data base does not differ in royalty rate,
percentage of tapered integration, or franchise fee
by comparison to the 1998 ‘Franchise 500’. Thus
the data set used in this study is representative
of the restaurant franchising population.11
By using a single industry, I controlled for
characteristics of industry technology and market
demand. As Williamson (1989: 172) notes, studies of transaction costs require a deep level of
observation of the phenomenon. ‘Transaction cost
economics operates at a more microanalytic level
of analysis than does orthodoxy.’ As a practical
matter, most studies using the transaction cost
paradigm have occurred in the context of a given
industry. The main variables of interest in transaction cost economics are difficult or impossible
to measure across industries (Shelanski and Klein,
1995). The restaurant industry is the largest
industry using franchising in gross sales ($64
billion dollars) and one of the largest in terms
of franchising presence-47 percent of all sales of
food and drink away from home are sold through
franchise systems (Michael, 1996).
In what follows, I describe the variables in the
study; all were stated explicitly in the UFOC or
derived from it. Various franchisor investments
are hypothesized to increase bargaining power. If
such investments are effective, the franchisor is
more likely to find agreement within the adaptive
range of the contract and require less litigation.
As a result, the dependent variable this paper
seeks to explain is litigation. To measure litigation, I consult the franchisor’s litigation history
in the past three years. All lawsuits involving
conflict with franchisees are used; all represent
attempts by the franchisor to terminate the agreement. Litigation is measured as all lawsuits with
franchisees in which the franchisor has engaged
in the last three years.12
10
Datasets built from these publications have been the primary
source of franchising empirical work (e.g., Lafontaine, 1992b).
11
For further discussion of the sample, please contact the
author for a data appendix.
12
The UFOC lists suits initiated by either franchisee or
franchisor, along with a brief description of the issues. As a
litigation tactic, franchisees who are sued typically ‘counterclaim,’ or sue the franchisor, claiming that promised services
have not been received. Distinguishing franchisor initiated
Strat. Mgmt. J., 21: 497–514 (2000)
Investments in Bargaining Power
The dependent variable, lawsuits, takes on positive discrete values in the time period observed;
they are count data. As is well known, ordinary
least squares (OLS) is inefficient and can be
biased when used on count data. A model commonly used in the social and life sciences for
count data is negative binomial regression (e.g.,
Hausman, Hall, and Griliches, 1984; Cameron
and Trivedi, 1986; Morrison and Schmittlein,
1988; King, 1989a, b; Barron, 1992; White and
Bennetts, 1996; Zorn, 1998; Greene 1997: 931ff).
Count data can be modelled as a Poisson process.
A Poisson process assumes that the event of
interest, say a lawsuit, occurs at some rate over
a duration of time. Then the expected number of
occurrences of the event can be modelled as the
product of the rate times the duration. Empirical
practice has suggested that the assumption of a
constant rate is too restrictive, however. Negative
binomial regression generalizes the Poisson model
by allowing the rate of the underlying process to
vary across observations according to a gamma
distribution.13 Variance in the rate, termed overdispersion, is theoretically explained by the presence of a lack of independence among occurrences of the event for each observation (King,
1989a; Greene, 1997).
To apply negative binomial regression to the
relationship between franchisor and franchisees,
the number of occurrences of lawsuits in the
three year period is assumed to occur at a rate
over time that varies among franchise systems.
Two modifications are required. First, in this
model, the size of the relevant population varies;
each franchise system has a different number
of franchisees, and therefore systems with more
franchisees have more exposure to litigation when
measured as a count of lawsuits. The model must
from franchisee initiated suits was not possible, especially
given that troubled franchisees may anticipate action by the
franchisor and sue first. Therefore, each dispute was counted
as one suit, regardless of whether several suits arose from a
single dispute. Lawsuits that were not the result of an attempt
at termination or enforcement of other contract provisions
were not included in the analysis. In particular, attempts by
franchisees to induce the franchisor to improve its effort or
quality were not included because no such suits occurred in
the time frame studied. Such suits have become more common; see, e.g., Serwer, 1995.
13
The mixture of Poisson processes with rates distributed
gamma yields the negative binomial distribution (derived in
Greene, 1997). The choice of the gamma distribution is made
for mathematical convenience, although to date it has had
considerable success in applications.
Copyright  2000 John Wiley & Sons, Ltd.
503
control for the size of the population of franchisees (potential litigants). Therefore, following
Maddala (1983: 51ff), I control for this difference
among systems by including the log of the number of franchisees. Second, here overdispersion
seems theoretically possible because the probability of one lawsuit within a system may be
affected by the presence of other lawsuits. A
franchisor’s decision to move beyond the adaptive
range to the courts for one franchisee may lower
the costs of taking another franchisee to the
courts; similarly, one franchisee engaging in quality shirking may induce another to do so. Therefore the negative binomial specification seems
appropriate. Additional estimating efficiency can
be gained by modelling overdispersion along with
the mean number of occurrences (Cameron and
Trivedi, 1986; Smyth, 1989; King, 1989b; Barron,
1992; White and Bennetts, 1996; Zorn, 1998). In
this application, I model dispersion as a function
of the independent variables. So the empirical
model uses negative binomial regression to fit the
count of events (lawsuits) normalized by the
number of franchisees to a negative binomial
distribution where the mean and dispersion are
functions of the independent variables.
Among the independent variables, the effect
of tapered integration is measured by the number of units owned by the franchisor. If owned
units do not create bargaining power, the number of franchisor-owned units would have no
effect on litigation. One possible flaw in the
research design is using tapered integration
observed contemporaneously with lawsuits.
Short term fluctuations in the sale of franchises
may cause tapered integration to deviate from
desired levels. To examine the risk of this occurring, I tested whether the ratio of owned to total
units as observed in this data set during either
1990 or 1991 differed from the three-year historical average over the three years 1987–1989
gathered from Entrepreneur magazine.14 For 61
of the firms (62%), data were available for such
a test; the difference of means was not significant (t = −1.07). For 84 of the firms, data were
available to test 1988 differences alone; the
14
In addition to the studies discussed above, this measure has
been used often as the dependent variable in a host of studies,
as reviewed in Dant, Kaufmann, and Paswan (1992). Love
(1986: 292) describes this as a relevant choice variable for
McDonald’s management.
Strat. Mgmt. J., 21: 497–514 (2000)
504
S. C. Michael
difference of means was not significant
(t = −1.01). A prospective test was also
employed. I tested whether the ratio differed
from the three year average during 1993–1995
gathered from Entrepreneur magazine. For 43
of the firms, data were available; the difference
of means was not significant (t = −0.887). For
56 of the firms, data were available to test 1994
differences; the difference of means was not
significant (t = −1.08). Therefore the tapered
integration observed reflects a long-run structural characteristic over almost a decade for
many franchise systems in the data.15
Other independent variables are measured as
follows. First, I include the length of the franchisor’s training program through which each
franchisee must pass. Second, an indicator variable is employed to determine whether the
franchisor requires previous industry experience
of the franchisee, as in Lafontaine (1992b). To
measure tying, I include the proportion of wholesale purchases by each franchisee that must be
made from the franchisor. Finally, an indicator
variable is added to measure whether the
franchisor has granted to the franchisee an exclusive territory.16
Two control variables are added to the model
to address the risk of institutionalization, the risk
that franchisors may both litigate more and
increase any of the independent variables as a
result of lifecycle effects (Shane, 1998). Both
age and growth of the franchise system are
included. Both are intuitively plausible: older
firms may have more risk of litigation, and
growth in the franchise system may affect
franchisees’ expectations or franchisors’ costs in
monitoring quality. Age is measured as the num15
This appears to contradict Lafontaine and Kaufmann, who
found that PVI changes as a nonlinear function of age. But
the Lafontaine and Kaufmann (1994) data contain mostly
young franchisors; over 60% were under ten years old (1994:
104). By contrast, 73% of franchisors in this data set are over
ten years old. Given that the survey reported in Lafontaine and
Kaufmann (1994) states that most franchisors desire a specific
level of PVI, younger franchisors may have more difficulty
achieving this goal because of resource or demand constraints.
Alternatively, the effect may be generated by industry variation in the desired level of PVI, because Lafontaine and
Kaufmann (1994) use franchisors from multiple industries.
16
Data limitations required me to assume that selection standards are constant over all franchisees, e.g., that the requirement of training and experience today has applied to all
franchisees in the system whenever they entered. Research
has not investigated the stability of these selection criteria
over time.
Copyright  2000 John Wiley & Sons, Ltd.
ber of years franchising, and growth is measured
as the number of units opened in the last three
years. In summary, the empirical model can be
written:
冉
冊 冉
PR(YI = y) =
冉
1/␣
1/␣ + ␭I
1/␣
⌫(y + 1/␣)
⌫(1/␣)y!
␭I
1/␣ + ␭I
冊
冊
y
(1)
with ␭I = exp(XB) and ␣ = exp(XC).
Here y is the number of lawsuits, I indexes
franchisors, ␭ is the rate of lawsuits, ␣ represents
dispersion of lawsuits, ⌫ represents the gamma
function, X is a matrix of independent variables,
and B and C are vectors of coefficients to be
estimated. The X matrix includes owned units,
training weeks, an indicator for experience
required, the percentage of tied products, whether
an exclusive territory is granted, the log of franchised units, the age of the franchise system, and
its three-year growth rate. The equation of interest
is that of ␭, the conditional mean of lawsuits
given the independent variables; its coefficients
are contained in the B vector. In that equation,
the coefficient on Owned Units is predicted to
be negative (Hypothesis 1), the coefficient on
Weeks of Training is expected to be negative
(Hypothesis 2), the coefficient on Experience is
expected to be positive (Hypothesis 3), the coefficient on Percent Tied to be positive (Hypothesis
4), and the coefficient on Exclusive Territory
granted to be negative (Hypothesis 5). The coefficients on the control variables Years Franchising
and log number of franchised units are expected
to be positive. Statistical inference and interpretation on these coefficients are as in the
OLS model.
Means, standard deviations, and correlations
among variables are reported in Table 1.
RESULTS
Main results
A pretest showed that overdispersion exists in
these data, so negative binomial regression is
appropriate. Results of a negative binomial
regression of Equation 1 estimated through
maximum likelihood is presented in Table 2. The
effect on the conditional mean is presented in
Strat. Mgmt. J., 21: 497–514 (2000)
Investments in Bargaining Power
505
Table 1. Descriptive statistics and correlations of the variables
Mean
1. Ratio of owned to
franchised units
2. Litigation
3. Franchised units
4. Owned units
5. Training weeks
6. Experience required?
7. Years franchising
8. Percent tied
9. Exclusive territory used
10. Three year growth in
units (in percent)
0.353
SD
0.291
3.75
6.42
515
1070
230
436
6.93
5.98
0.545
0.500
20.4
12.5
2.54
6.25
0.586
0.495
17%
41%
1
4
5
6
7
8
9
−0.19 1
−0.21 0.48 1
0.24 0.26 0.67 1
0.13 −0.05 0.27 0.16 1
0.32 0.03 0.02 0.24 0.19 1
−0.30 0.32 0.37 0.30 0.07 0.01 1
−0.05 0.08 −0.06 −0.13 0.01 −0.24 0.06 1
−0.02 −0.16 −0.24 −0.22 −0.08 −0.03 −0.18 0.06 1
0.05 −0.14 −0.11 −0.16 0.01 −0.03 −0.37 −0.07 0.23
Multicollinearity is not present. The condition index of all
the independent variables is 4.4, well below 20 as recommended in Greene (1997).
Copyright  2000 John Wiley & Sons, Ltd.
3
1
column 1 and is the focus of discussion. The
effect on dispersion is in column 2.
The equation is significant at the 1% level.17
Results offer support for all hypotheses. For tapered integration to be effective, Hypothesis 1
predicted that owning units would significantly
decrease litigation: it does. The coefficient of
owned stores or tapered integration is negative
and significant at the 1% level. Presumably
increasing tapered integration increases bargaining
power and alters franchisee behavior, decreasing
litigation. Note that this model controls for the
number of franchised units, so the result that
tapered integration decreases litigation cannot be
explained as simply reducing the number of
potential litigants. To address the question of
causality, the number of units owned in 1988,
two years prior to the documents in the data
set, is employed in an alternative regression, not
reported here. Available for 84 of the franchisors,
this historical value for owned units is also negative and significant, with the coefficients on other
variables essentially the same. In addition, an
econometric test was performed to insure that
litigation and ownership structures were not
jointly determined. Hausman’s (1978) test was
applied to test whether the number of owned and
the number of franchised units are correlated to
the error term in the equation: they are not.
Therefore tapered integration is exogenous to
the model.
17
2
With regard to selection, Hypothesis 2 predicted that, by increasing switching costs, training
would reduce litigation. The coefficient on training weeks is negative and significant at the 1%
level, so Hypothesis 2 is supported. Training
Table 2. Negative binomial estimation of litigation in
franchise systems
Conditional Dispersion (2)
Mean of
Litigation (1)
Log (Franchised units)
Owned units
Training weeks
Experience required?
Percent tied
Exclusive territory
granted?
Years franchising
Three year growth rate
Constant
Chi-squared test
(df = 16) for all
coefficients zero
0.4014***
(0.1110)
−0.0013***
(0.0003)
−0.0699***
(0.0178)
0.9444***
(0.3110)
0.0358*
(0.0187)
−0.9192***
(0.2715)
0.0429***
(0.0162)
−0.0076
(0.0054)
−0.8379
(0.8314)
70.3***
−0.8862***
(0.2133)
−0.0026***
(0.0009)
−0.1296**
(0.0602)
1.082*
(0.5854)
−0.019
(0.0463)
−1.019*
(0.5507)
0.0332*
(0.0192)
−0.0448**
(0.0214)
5.842***
(1.264)
Notes: 1) Sample size is 99. 2) Standard error is in parentheses under coefficient estimate. 3) Significance levels are
noted with asterisks: *** is 1%; ** is 5%; * is 10%; all in
two tailed tests. 4) Likelihood ratio test of this model versus
Poisson equals 234 (1 df), significant at p ⬍ 0.001.
Strat. Mgmt. J., 21: 497–514 (2000)
506
S. C. Michael
appears to socialize the franchisee, aligning
expectations between franchisee and franchisor
and reducing litigation. Franchisee experience,
expected to raise franchisee power and therefore
litigation, also had a positive and significant coefficient. Hypothesis 3 is thus supported. An alternative interpretation of this result is that operating
complexity, as reflected by the need for industry
experience of the franchisee, positively affects
lawsuits. To explore this possibility, I added to
the model a variable for the level of initial investment required by the franchisee as a possible
measure of operating complexity, assuming that
more costly franchises were more complex. In
results not reported here, the coefficient on this
variable is insignificant. This test does offer some
control against the interpretation of experience as
operating complexity, but it cannot be rejected
completely by the model.18
The results of Kaufmann and Stanworth (1995)
suggest that prospective franchisees with industry
experience may value training differently than
those without experience.19 Therefore, training
and experience were interacted in a variant of
Equation 1. Interestingly, adding this interaction
and including the two main effects of training
and experience yielded insignificant coefficients
for experience and for the interaction, while training remained significant and negative. Thus it
appears that more training can weaken the effect
of prior industry experience.
The coefficient on the percent of products
required to be purchased by the franchisee from
the franchisor was positive and marginally significant. Hypothesis 4 is supported. Omitting
tying from the model did not affect the signs of
the other model coefficients nor the conclusions
from the hypotheses tests.
The granting of exclusive territories was
expected to lower litigation by raising the value
of the franchise through geographic monopoly in
Hypothesis 5, and the hypothesis is supported.
The indicator variable has a significant and negative coefficient. For the control variables, years
franchising had a positive and significant coefficient; growth had a negative and insignificant
coefficient. Reestimating the model without
growth did not affect the signs of the other
18
I am grateful to a referee for identifying this alternative
explanation.
I am grateful to a referee for suggesting this interaction.
19
Copyright  2000 John Wiley & Sons, Ltd.
model coefficients nor the conclusions from the
hypotheses tests.
Although dispersion is not the subject of this
paper, a few comments on the dispersion coefficients in the second column of Table 2 may
be of interest. Other than the control for size,
coefficients on three variables are significant at
conventional levels. System growth has no significant effect on the number of lawsuits
(insignificant coefficient in the first column), but
it does appear to reduce dispersion, i.e., lawsuits
in high growth systems display less dependence.
One possible explanation is that, in a high growth
system, the franchisor can award additional franchises, and owning multiple units usually leads
to higher profits because of economies of scale
in local supervision (Michael and Moore, 1995).
So the continued goodwill of the franchisor is
more important in growing systems, and franchisees may be less likely to free ride given one
low quality franchisee. Training appears to have
the same effect, as seen by its negative coefficient. This is consistent with our above discussion of training: if training shapes expectations
in such a way as to make high quality performance desirable, then difficulties with one
franchisee are less likely to spread to another.
The number of years franchising appears to raise
dependence. As time passes, the franchisor is
likely to develop a reputation among franchisees
regarding willingness to enforce through the
courts (either positively or negatively), and therefore increase dependence of one suit on another.
Robustness checks
A number of alternative specifications are
employed, reported in Table 3; none affect the
essence of the conclusions above.
The robustness checks are grouped into four
categories: survivor bias, business type effects,
franchisor financial resources, and geographic dispersion. A representative model is reported from
each category in Table 3. For brevity, only the
coefficients on the conditional means are reported.
The first column of the table repeats the baseline
results of Table 2. To test theory and to develop
normative prescriptions for managers, presumably
we are most interested in stable empirical
relationships among established and surviving
franchisors. A representative cross section of franchisors is likely to contain systems that have not
Strat. Mgmt. J., 21: 497–514 (2000)
Investments in Bargaining Power
507
Table 3. Robustness checks for litigation in franchise systems
Log (Franchised units)
Owned units
Training weeks
Experience required?
Percent tied
Exclusive territory given?
Years franchising
Three year growth rate
Base (1)
10 yrs (2)
Indicators (3)
ROA (4)
Termination
State (5)
0.4014***
(0.1110)
−0.0013***
(0.0003)
−0.0699***
(0.0178)
0.9444***
(0.3110)
0.0358*
(0.0187)
−0.9192***
(0.2715)
0.0429***
(0.0162)
−0.0076
(0.0054)
0.3525***
(0.1252)
−0.0013***
(0.0003)
−0.0717***
(0.0186)
1.0348***
(0.3320)
0.0336*
(0.0179)
−0.9375***
(0.2840)
0.0315***
(0.0182)
−0.0107***
(0.0095)
0.1781
(0.1187)
−0.00104***
(0.00031)
−0.0642***
(0.0198)
1.1673***
(0.3716)
0.0633**
(0.0238)
−0.9916***
(0.2448)
0.0571***
(0.0146)
−0.0156**
(0.0070)
Yes
0.4056***
(0.1126)
−0.0012***
(0.0003)
−0.0625***
(0.0209)
0.7889***
(0.3541)
0.0328***
(0.0198)
−0.8651***
(0.2747)
0.0428***
(0.0165)
−0.0080***
(0.0053)
0.3646***
(0.1061)
−0.0013***
(0.0003)
−0.0797***
(0.0185)
0.6311***
(0.3809)
0.0338***
(0.0218)
−1.0578***
(0.2919)
0.0390***
(0.0144)
−0.0090***
(0.0055)
Business type indicators included?
Franchisor return on assets
−0.5011
(0.6195)
Termination state indicator?
Constant
Number of observations
Chi-squared test
0.8379
(0.8314)
99
70.3***
−0.1354
(0.8885)
72
43.0***
−0.0411
(0.7769)
99
78.3***
−0.8189
(0.8334)
99
71.4***
−0.4451
(0.2760)
0.1156
(1.0387)
99
74.7***
Notes: 1) Dispersion results omitted for clarity. 2) Standard error is in parentheses under coefficient estimate. 3) Significance
levels are noted with asterisks: *** is 1%; ** is 5%; * is 10%; all in two tailed tests.
yet overcome the ‘liability of newness’
(Stinchcombe, 1965) to become established businesses; indeed, Shane (1996) observed a 75%
failure rate in a ten year period for new franchise
systems. If practices differ systematically between
new, riskier franchise systems and established
ones, including observations on new franchise
systems may inadvertently suggest a relationship
where none exists among established franchisors.
One screen to identify established franchisors is
to repeat the analysis on firms beyond a certain
age or a certain size, so I re-estimated the equation including only: 1) firms that had survived
five years or more; 2) firms that had survived
ten years or more; 3) firms whose total system
size contained over 25 units; 4) firms whose total
system size contained over 50 units. In all cases,
signs and levels of significance (significant at the
one percent, five percent, ten percent, or not
significant) of the coefficients on the hypothesized
variables remained the same. Column 2 reports
the result of estimating the model with only
Copyright  2000 John Wiley & Sons, Ltd.
franchisors that are ten years old. A second screen
is to examine only franchisors known to survive
until 1998. Nine systems in the data set failed,
using the definition of Shane (1998), delisting
from all major franchising publications. Rerunning the equation with only survivors in 1998
did not alter signs or significance tests of the
other model coefficients. So selection did not
alter the model’s conclusions.
Second, to capture effects due to variation in
business types, indicator variables were added
(pizza, fried chicken, quick service hamburger,
family style restaurant, and the like). As reported
in Column 3, the signs and significance levels
of Equation 1 remained unchanged. Third, strong
financial resources of the franchisor may raise
the franchisor’s propensity to litigate, engaging
in ownership redirection, or raise the franchisees’
willingness to litigate because the franchisor is an
attractive target for a lawsuit. To control for this,
the model is reestimated with three measures of
financial resources of the franchisor: return on
Strat. Mgmt. J., 21: 497–514 (2000)
508
S. C. Michael
assets, return on sales, and debt to assets ratio. (A
few systems have negative equity, so return on
equity is not used.) In no case is the measure of
financial resources significant; all other signs and
significance levels remain unchanged. Therefore
variance in financial resources of the franchisor are
not driving the results. Column 4 reports results
including franchisor return on assets. To test more
directly for ownership redirection, the number of
units reacquired by the franchisor from franchisees,
reported in the UFOC, is added to the variables
of Equation 1. This variable has a negative but
insignificant coefficient, offering no support for
ownership redirection as an explanation for
litigation. Again, all other signs and significance
levels remain unchanged.
Finally, some might argue that the geographic
dispersion of the franchise system affects monitoring costs, so more geographically dispersed
franchisors have less litigation. To measure the
effect of geographic dispersion, I add the number
of states in which the franchise system operates;
the coefficient is insignificant. A more specific
test is also used. State law regarding franchising
termination and litigation varies significantly from
state to state (Beales and Muris, 1995; Pitegoff,
1989). It is possible that geographic heterogeneity
of franchise systems is driving the results, as
franchisors in states with more difficult termination laws litigate less. The only way to control
for this perfectly is to develop a composite mix,
determining how many units are in states with
strict versus loose termination laws. Such state
by state data were not available for this study.
To address this threat to validity, two controls
were applied. First, an indicator variable was
coded one if the franchise system’s headquarters
was located in a strict termination state, as
reported in Table 2 of Beales and Muris (1995)).
Presumably the franchise system would have proportionately more units in its home state, so one
located in a strict termination state has a mix
less favorable to litigation. As reported in column
5, the indicator is not significant. Second, the
equation was estimated using only systems with
operations in more than 20 states. Presumably a
system with operations in 20 states is primarily
making decisions based upon product market
competition and demand for the ultimate product
and not fear of litigation. The indicator is not
significant in this smaller regression either. Therefore state law is unlikely to be driving the results.
Copyright  2000 John Wiley & Sons, Ltd.
DISCUSSION
Investments in bargaining power ex ante through
such mechanisms as tapered integration and
franchisee selection can reduce ex post litigation
and termination in franchise systems. Specifically,
tapered integration, training of franchisees, the
granting of exclusive territories, and the selection
of inexperienced franchisees reduce litigation.
Tapered integration represents an investment to
weaken buyer power through information and
credible commitment to operating units. Selection
of less-powerful individuals as franchisees, and
further weakening that power through socialization and training, also reduces buyer power.
The methodology and the results are robust.
The data are taken from primary sources, unlike
most previous franchising research, and modeled
with a technique appropriate to count data and
their underlying distribution. A number of robustness checks using alternative variables has demonstrated that the fundamental relationships hold
even allowing for differing financial resources,
geographies, and samples. And causality appears
to be correctly inferred based on both theory and
econometric technique.
Implications for research
Hybrid organizational forms that combine features
of markets and hierarchies are of considerable
theoretical interest (e.g., Powell, 1987).
Researchers in hybrid organizational forms have
noted the potential and the actuality of conflict
between two independent parties joined together,
and the reduction of conflict has been taken to
be a desirable goal (e.g., Baucus, Baucus, and
Human, 1996). These results suggest a more
nuanced approach to conflict in hybrid forms.
Conflict on a particular issue may be bad, but
compromise may be possible within the adaptive
range, with tradeoffs made in other areas. In
particular, linear measures of conflict and performance may not capture the dichotomy created
by the adaptive range: compromise or terminating
the relationship. Future research in this area might
attempt to delineate more closely the adaptive
range, the issues included and the tradeoffs possible, in particular organizational forms.
The results support the conjecture that
operating
two
organizational
forms
simultaneously (in this case, tapered integration
Strat. Mgmt. J., 21: 497–514 (2000)
Investments in Bargaining Power
through owning some units while franchising
others) can have synergistic effects (Bradach and
Eccles, 1989; Bradach, 1997), here demonstrated
to reduce litigation and termination and presumably improve quality.20 These results extend transaction cost theory by suggesting that the proper
unit of analysis is not the transaction individually
but the set of transactions together.21 This paper
argues that, presented with identical opportunities,
the franchisor may rationally choose to treat identical transactions differently in order to raise his
bargaining power.
This insight regarding the ability of one transaction to affect another enriches current strategic
management thinking on vertical integration. The
theoretical model and framework of Mahoney
(1992a) is certainly the ‘state of the art’ regarding
transaction cost economics and strategic management theory on vertical integration. The article
reviews the major organizational forms including
making and buying, as well as hybrid forms, and
explains variation in these as a function of the
level of transaction specific investment required
and of the effectiveness of input measures or
output measures to provide an accurate basis for
rewarding agents. The model does not have a
role for tapered integration, because it implicitly
assumes a single make-or-buy decision. If that
constraint is relaxed, this paper has argued that
effects exist that make it possible for one transaction to affect another transactor’s behavior. This
paper extends the model in Mahoney (1992a) by
noting that, in the context of a single ‘make-orbuy’ decision, the model is complete. When other
identical decisions are introduced, the effect of
transactions and transactors introduces another
source of variation in behavior and costs, and
tapered integration becomes an alternative.22
The results on training are likely to have particular interest to OB colleagues. As mentioned
20
Bradach and Eccles (1989) conjecture that such forms will
also increase innovation, which this study did not examine.
21
This ambiguity has been noted elsewhere; see Hirsch and
Lounsbury (1996).
22
Jacquemin (1987) makes this point in more general terms.
At present, we have a number of models of both organization
and production that emphasize efficiency. Some models of
production note the effect one firm’s quantity choice has on
another’s; in this sense our models of production are ‘strategic’. Models of organization do not, in general, show such
effects; efficiency has been the dominant paradigm. By arguing that one transaction can affect others, we introduce strategic considerations to organizational design.
Copyright  2000 John Wiley & Sons, Ltd.
509
above, the previous research that has validated
the effect of training has done so generally by
examining employee self-reports up to a year or
so after beginning employment. Although not a
primary purpose of this investigation, the results
do suggest that, first, training may have effects
that persist longer than a year, and, second, that
the training has effects on individuals who are
not strictly employees, such as franchisees.
Although not discussed by Porter (1980), it
appears that the scope and complexity of the
bargain affect bargaining power. The more items
over which bargaining can take place, the more
diffused bargaining power becomes, and the more
likely litigation is to occur. This is perhaps the
most powerful and theoretically interesting explanation for the observed results of exclusive territories and the lack of tied products reducing
litigation. Exclusive territories, whatever their
other effects, eliminate geographic expansion in
the territory as an item of conflict between
franchisee and franchisor, and the lack of tied
products also eliminates an item of conflict. These
have the effect of enhancing ‘discreteness’
(Macneil, 1978), making clear what is expected
of the two parties, and reducing the need for
adaptation (Macneil, 1978; Williamson, 1985:
68ff). Future research should consider whether
deliberately reducing the areas of agreement
required in an organizational form enhances
efficiency.
Bargaining power is broadly complementary
to the two existing theoretical perspectives in
franchising: agency theory and life cycle theory.
Agency theory notes that the agency relationship
between franchisor and franchisee require adjustment of financial incentives, but it argues, or at
least implies, that those incentives, such as royalty rate and franchise fee (Lafontaine, 1992b)
and ex ante rents (Michael and Moore, 1995) are
sufficient to insure quality. Under agency theory,
litigation does not occur; the contract is selfenforcing on the franchisees. By contrast, bargaining power acknowledges the importance of
financial incentives, but also acknowledges that
the contract designed to address the agency problem of franchisee and franchisor is likely to be
incomplete in some important dimension, at least
in the face of uncertainty or change, and therefore
adaptation will be required (Williamson, 1985).
Bargaining power allows for investments ex ante
to influence changes ex post through negotiation
Strat. Mgmt. J., 21: 497–514 (2000)
510
S. C. Michael
in the adaptive range to insure quality beyond
the use of financial incentives, and predicts what
will influence the outcome of that bargaining.
Life cycle theory predicts that litigation will be
primarily a tool of ownership redirection, for
which empirical results here offer no support.
However, life cycle effects may inhibit the franchisor’s desired investment in bargaining power.
For example, resource constraints may impede
the ability of the franchisor to employ tapered
integration in its early years. Theoretical pluralism
(e.g., Mahoney, 1992b) may be the appropriate
approach to franchising and, more generally, issues of the hybrid organization in future research.
This paper suggests several directions for
further research. The paper has focused primarily
on bargaining power of the franchisor. More
detailed measures of bargaining power, experience, and franchisee human capital would be of
value in future research to identify sources of
bargaining power of franchisees. In addition, franchisees frequently own several units, called multiunit franchising; these systems within systems
presumably give their owners more bargaining
power than independents. The role of multi-unit
franchising has begun to be explored by Kalnins
and Lafontaine (1996) and Kaufmann and Dant
(1996). Also, bargaining power is clearly subject
to changes in the legal regime and in the courts;
Brickley et al. (1991) uses changes in law to
explore the market value of franchisors. But certainly more remains to be done.
Implications for managers
The results provide guidance to managers,
especially franchisors implementing a franchise
system. The effect of one transaction on another
makes bargaining power endogenous to a relationship. This suggests that bargaining power can be
and should be managed, as suggested by Porter
(1980). To fully consider the risk of opportunism,
firms entering a relationship must consider if
investment by one or both parties to the agreement can change bargaining power. Further
research on ways to manage bargaining power
would be of interest.
With regard to tapered integration, as argued
in Michael (1996), prospective franchisors choose
whether to franchise or not based upon risk and
human capital considerations in their industry.
The implementation of franchising—how many
Copyright  2000 John Wiley & Sons, Ltd.
units to own—can then be determined by the need
for information and incentives. The geographic
considerations outlined in Brickley and Dark
(1987) and Fladmoe-Lindquist and Jacque (1995)
also play a role. What level of tapered integration
is optimal? In the absence of a detailed model,
conjecture must suffice. The choice will be influenced by the role of information and uniformity
in the production process. Published reports, such
as Franchising in the Economy (International
Franchise Association, 1990: 4), report that on
average franchisors own approximately twenty
percent of their units. McDonald’s, the largest
and most successful franchising company, has a
stated policy of twenty-five percent (Love, 1986).
But the loss of high powered incentives
(Williamson, 1985) argues against most units
being company-owned.
Perhaps more important than the absolute number is the location of the owned units. Tapered
integration is likely to be more successful when
owned units are dispersed among all market areas
where the system competes. A presence in each
market area gives the franchisor on-the-spot information about each market and also demonstrates
the ability, commitment, and infrastructure to
manage units in each market. Concentration of
units in, say, the franchisor’s home state, is
unlikely to be as effective in raising quality and
curbing litigation.
Avoiding litigation, with its cost and uncertainty, is clearly desirable to managers, and the
results also offer some suggestions for reducing
litigation. First, more training is likely to be
better. As discussed above, training is likely to
make franchisees more likely to follow the franchisor’s system and more committed to the
organization. But such training may have a negative effect as well to the franchisor: it may make
the franchisee more valuable to the franchisor
and increase the risk of hold-up. Such a conclusion does not appear to be warranted by the
results; perhaps the effect of socialization outweighs the risk of holdup. For any given
franchisor, however, holdup risk might be created.
How much training is enough? The model does
not imply an optimum level, but some guidance
can be obtained by examining the distribution of
training levels in the data. The mean length of
training is 7 weeks, the median is 5, and the
95th percentile is 15. In addition to transmitting
information, the training should include socialiStrat. Mgmt. J., 21: 497–514 (2000)
Investments in Bargaining Power
zation activities as well (suggested in Jones,
1986), in order to reduce litigation.
The use of experienced franchisees increases
litigation, but, as shown by Shane (1998), it also
increases the chance of survival of a new franchise system. A tradeoff is suggested, then,
between reducing litigation and increasing survival. Therefore, franchise systems who have
overcome the ‘liability of newness’ (Stinchcombe,
1965) are advised to recruit primarily inexperienced franchisees. These more established systems are presumably at less risk of failure. On
the other hand, new franchise systems are advised
to recruit experienced industry hands as new franchisees, to increase survival, even at the risk of
increasing litigation in the long run. If the system
fails, there is no long run.
Tying is a legally problematic device that
appears to increase litigation, despite its potentially quality-enhancing effect. Using alternatives
such as exclusive suppliers or requiring purchasing from designated vendors is likely to remove
the perceived conflict of interest, strengthen the
relationship, and reduce litigation. Exclusive territories represent a significant commitment of the
franchise system to the individual franchisee. But
the effect of reducing litigation is likely to offset
some of that cost.
Conclusion
Investments in bargaining power ex ante through
mechanisms such as tapered integration and
franchisee selection can reduce ex post litigation
and termination in franchise systems. Specifically,
tapered integration, training of franchisees, the
granting of exclusive territories, and the selection
of inexperienced franchisees reduce litigation.
Tapered integration represents an investment to
weaken buyer power through information and
credible commitment to operating units. Selection
of less-powerful individuals as franchisees, and
further weakening that power through training,
also reduces buyer power. Overall, the length and
strength of the franchising relationship can be
affected by managerial and contractual variables.
Franchising has the advantage of relative homogeneity and simplicity for theory-testing, but
nothing in the theory directly precludes application to other industries and situations. In particular, it highlights that investment in bargaining
power is likely to be especially effective in the
Copyright  2000 John Wiley & Sons, Ltd.
511
context of organizational forms characterized by
neoclassical contract law and an adaptive range,
such as joint ventures and strategic alliances.
Research to understand further the usefulness of
bargaining power through tapered integration, selection, and other means in other hybrid organizational forms is desirable.
Acknowledgements
Many thanks to Richard Caves, Pankaj Ghemawat, Al Silk, Jim Combs, Roger Koenker, Hun
Lee, Jim Wade, Chris Zorn, and two referees for
helpful comments and suggestions. Thanks to Jeff
Kolton and Frandata Corporation for access to
some of the data used in this research. I am also
grateful to Lou Stern who first introduced me to
the opportunities and problems in channels of
distribution. Remaining errors are mine alone.
Funding for this project came from the Harvard
Business School Division of Research.
REFERENCES
Allen, N. J. and J. P. Meyer (1990). ‘Organizational
socialization tactics: A longitudinal analysis of links
to newcomers’ commitment and role orientation,’
Academy of Management Journal, 33, pp. 847–858.
Anand, P. (1987). ‘Inducing franchisees to relinquish
control: An attribution analysis,’ Journal of Marketing Research, 24 (May), pp. 215–221.
Anand, P. and L. W. Stern (1985). ‘A sociopsychological explanation for why marketing channel members relinquish control,’ Journal of Marketing
Research, 22 (November), pp. 365–376.
Arrow, K. (1975). ‘Vertical integration and communication,’ Bell Journal of Economics, 6 (Spring),
pp. 173–183.
Ashforth, B. E. and A. M. Saks (1996). ‘Socialization
tactics: Longitudinal effects on newcomer adjustment,’ Academy of Management Journal, 39,
pp. 149–178.
Barron, D. N. (1992). ‘The analysis of count data:
Overdispersion and autocorrelation.’ In P. V.
Marsden (ed.), Sociological Methodology, Vol 22.
Blackwell, Cambridge, MA, pp. 179–220.
Baucus, D. A., M. S. Baucus, and S. E. Human (1996).
‘Consensus in franchise organizations: A cooperative
arrangement among entrepreneurs,’ Journal of Business Venturing, 11, pp. 359–378.
Beales III, J. H. and T. J. Muris (1995). ‘The foundations of franchise regulation: Issues and evidence,’
Journal of Corporate Finance, 2, pp. 157–198.
Bond, R. E. and C. E. Bond (1991). The Source Book
Strat. Mgmt. J., 21: 497–514 (2000)
512
S. C. Michael
of Franchising Opportunities. Dow Jones-Irwin,
Homewood, IL.
Bradach, J. L. (1997). ‘Using the plural form in the
management of restaurant chains,’ Administrative
Science Quarterly, 42 (2), pp. 276–303.
Bradach, J. L. and R. G. Eccles (1989). ‘Price, authority, and trust: From ideal types to plural forms,’
Annual Review of Sociology, 15, pp. 97–118.
Brickley, J. A. and F. H. Dark (1987). ‘The choice of
organizational form: The case of franchising,’ Journal of Financial Economics, 18 (June), pp. 401–420.
Brickley, J. A., F. H. Dark and M. S. Weisbach (1991).
‘The economic effects of franchise termination
laws,’ Journal of Law and Economics, 34 (April),
pp. 101–131.
Buzzell, R. D. (1983). ‘Is vertical integration profitable?,’ Harvard Business Review, 83(1), pp. 92–
102.
Cameron, C. A. and P. K. Trivedi (1986). ‘Econometric
models based on count data: Comparisons and applications of some estimators and tests,’ Journal of
Applied Econometrics, 1, pp. 29–53.
Carney, M. and E. Gedajlovic (1991). ‘Vertical integration in franchise systems: Agency theory and
resource explanations,’ Strategic Management Journal, 12 (8), pp. 607–629.
Caves, R. E. and W. F. Murphy (1976). ‘Franchising:
Firms, markets, and intangible assets,’ Southern Economic Journal, 42 (April), pp. 572–586.
Dant, R. P., P. J. Kaufmann and A. K. Paswan (1992).
‘Ownership redirection in franchised channels,’
Journal of Public Policy and Marketing, 11,
pp. 33–44.
Day, G. S. (1990). Market Driven Strategy: Processes
for Creating Value, Free Press. New York.
Dunkin, A. (1996). ‘Franchising: A recipe for your
second career?’, Business Week, (March 4),
pp. 128–129.
Fladmoe-Lindquist, K. and L. L. Jacque (1995). ‘Control modes in international service operations: The
propensity to franchise,’ Management Science, 41,
pp. 1238–1249.
Gallini, N. T. and N. A. Lutz (1992). ‘Dual distribution
and royalty fees in franchising,’ Journal of Law,
Economics, and Organization, 8 (October),
pp. 471–501.
Ghemawat, P. (1991). Commitment: The Dynamic of
Strategy. Free Press, New York.
Greene, W. H. (1997). Econometric Analysis, 3rd edn.
Prentice-Hall, Englewood Cliffs, NJ.
Hadfield, G. K. (1990). ‘Problematic relations:
Franchising and the law of incomplete contracts,’
Stanford Law Review, 42, pp. 927–992.
Harrigan, K. R. (1983). Strategies for Vertical Integration. D. C. Heath, Lexington, MA.
Harrigan, K. R. (1984). ‘Formulating vertical integration strategies,’ Academy of Management Review,
9, pp. 638–652.
Harrigan, K. R. (1985). ‘Vertical integration and corporate strategy,’ Academy of Management Journal, 28,
pp. 397–425.
Harrigan, K. R. (1986). ‘Matching vertical integration
Copyright  2000 John Wiley & Sons, Ltd.
strategies to competitive conditions,’ Strategic Management Journal, 7 (6), pp. 535–555.
Harris, N. and M. France (1997). ‘Franchisees get
feisty,’ Business Week, (Feb. 24), pp. 65–66.
Hausman, J. A. (1978). ‘Specification tests in econometrics,’ Econometrica, 46, pp. 1251–1272.
Hausman, J., B. H. Hall and Z. Griliches (1984).
‘Econometric models for count data with an application to the patents-R&D relationship,’ Econometrica, 52, pp. 909–938.
Hirsch, P. M. and M. D. Lounsbury (1996). ‘Rediscovering volition: The institutional economics of
Douglass C. North,’ Academy of Management
Review, 21, pp. 872–884.
International Franchise Association (1988). What You
Need to Know When You Buy a Franchise. International Franchise Association, Washington, DC.
International Franchise Association (1990). Franchising
in the Economy, 1988–1990. International Franchise
Association, Washington, DC.
Jacquemin, A. (1987). The New Industrial Organization. MIT Press, Cambridge, MA.
Jones, G. R. (1986). ‘Socialization tactics, selfefficacy, and newcomers’ adjustments to organizations,’ Academy of Management Journal, 29,
pp. 262–279.
Kalnins, A. and F. Lafontaine (1996). ‘The characteristics of multi-unit ownership in franchising:
Evidence and implications for theory,’ manuscript,
University of Southern California.
Kaufmann, P. J. and R. P. Dant (1996). ‘Multi-unit
franchising: Growth and management issues,’ Journal of Business Venturing, 11, pp. 343–358.
Kaufmann, P. J. and J. Stanworth (1995). ‘The decision
to purchase a franchise: A study of prospective
franchisees,’ Journal of Small Business Management, 33 (4), pp. 22–33.
Kessler, F. and R. M. Stern (1959). ‘Competition,
contract, and vertical integration,’ Yale Law Journal,
69, pp. 1–129.
Keup, E. J. (1990). The Franchise Bible. Oasis Press,
Grants Pass, OR.
King, G. (1989a). ‘Variance specification in event count
models: From restrictive assumptions to a generalized estimator,’ American Journal of Political
Science, 33 (August), pp. 762–784.
King, G. (1989b). ‘Event count models for international
relations: Generalizations and applications,’ International Studies Quarterly, 33, pp. 123–147.
Klein, B. (1980). ‘Transaction cost determinants of
“unfair” contractual arrangements,’ American Economic Review, 70, pp. 356–362.
Klein, B. and K. B. Leffler (1981). ‘The role of market
forces in assuring contractual performance,’ Journal
of Political Economy, 89 (August), pp. 615–641.
Klein, B. and L. F. Saft (1985). ‘The law and economics of franchise tying contracts,’ Journal of Law
and Economics, 28, pp. 345–361.
Lafontaine, F. (1992a). ‘How and why franchisors do
what they do: A survey Report.’ In P. Kaufmann
(ed.), Franchising: Passport for Growth and World
of Opportunity: Sixth Annual Proceedings of the
Strat. Mgmt. J., 21: 497–514 (2000)
Investments in Bargaining Power
Society of Franchising. International Center for Franchise Studies, Lincoln, NE.
Lafontaine, F. (1992b). ‘Agency theory and franchising:
Some empirical results,’ Rand Journal of Economics,
23 (Summer), pp. 263–283.
Lafontaine, F. (1993). ‘Contractual arrangements as
signalling devices: Evidence from franchising,’ Journal of Law, Economics, and Organization, 9,
pp. 256–289.
Lafontaine, F. and P. J. Kaufmann (1994). ‘The evolution of ownership patterns in franchise systems,’
Journal of Retailing, 70 (2), pp. 97–113.
Lafontaine, F. and K. Shaw (1996). ‘The dynamics of
franchise contracting: Evidence from panel data,’
National Bureau of Economic Research Paper
#5585.
Love, J. F. (1986). McDonald’s: Behind the Arches.
Bantam Books, New York.
Luxenberg, S. (1986). Roadside Empires: How the
Chains Franchised America. Penguin, New York.
Lynk, W. J. (1994). ‘Tying and exclusive dealing:
Jefferson Parish Hospital v. Hyde (1984).’ In J. E.
Kwoka Jr. and L. J. White (eds.), The Antitrust
Revolution (2nd edn.). Harper Collins, New York,
pp. 376–399.
MacMillan, I. C., D. C. Hambrick and J. M. Pennings
(1986). ‘Uncertainty reduction and the threat of
supplier retaliation: Two views of the backward
integration decision,’ Organization Studies, 7,
pp. 263–278.
Macneil, I. R. (1978). ‘Contracts: Adjustment of longterm economic relations under classical, neoclassical,
and relational contract law,’ Northwestern University
Law Review, 72, pp. 854–905.
Macneil, I. R. (1980). The New Social Contract. Yale
University Press, New Haven, CT.
Maddala, G. S. (1983). Limited-Dependent and Qualitative Variables in Econometrics. Cambridge University Press, Cambridge, UK.
Mahoney, J. T. (1992a). ‘The choice of organizational
form: Vertical financial ownership versus other
methods of vertical integration,’ Strategic Management Journal, 13 (8), pp. 559–584.
Mahoney, J. T. (1992b). ‘The adoption of the multidivisional form of organization: A contingency
model,’ Journal of Management Studies, 29,
pp. 49–72.
Malmgren, H. B. (1961). ‘Information, expectations
and the theory of the firm,’ Quarterly Journal of
Economics, 75, pp. 399–421.
Mathewson, G. F., and R. A. Winter (1985). ‘The
economics of franchise contracts,’ Journal of Law
and Economics, 28 (October), pp. 503–526.
Michael, S. C. (1996). ‘To franchise or not to franchise:
An analysis of decision rights and organizational
form shares,’ Journal of Business Venturing, 11
(January), pp. 57–71.
Michael, S. C. and H. J. Moore (1995). ‘Returns to
franchising,’ Journal of Corporate Finance: Contracting, Governance, and Organization, 2,
pp. 133–155.
Minkler, A. P. (1992). ‘Why firms franchise: A search
Copyright  2000 John Wiley & Sons, Ltd.
513
cost theory,’ Journal of Institutional and Theoretical
Economics, 148, pp. 240–259.
Morrison, D. G. and D. C. Schmittlein (1988). ‘Generalizing the NBD Model for customer purchases:
What are the implications and is it worth the effort?’,
Journal of Business and Economic Statistics, 6
(April), pp. 145–159.
Muris, T. J. (1981). ‘Opportunistic behavior and the
Law of Contracts,’ Minnesota Law Review, 65,
pp. 521–590.
Norton, S. W. (1988). ‘An empirical look at franchising
as an organizational form,’ Journal of Business, 61
(April), pp. 197–218.
Norton, S. W. (1995). ‘Is franchising a capital structure
issue?’, Journal of Corporate Finance, 2, pp. 75–101.
Oxenfeldt, A. R. and A. O. Kelly (1968–69). ‘Will
successful franchise systems ultimately become
wholly-owned chains?’, Journal of Retailing, 44,
pp. 69–83.
Peterson, A. and R. P. Dant (1990). ‘Perceived advantages of the franchise option from the franchisee
perspective: Empirical insights from a service franchise,’ Journal of Small Business Management, 28
(3), pp. 46–61.
Phan, P. H., J. E. Butler and S. H. Lee (1996).
‘Crossing mother: Entrepreneur-franchisees’ attempts
to reduce franchisor influence,’ Journal of Business
Venturing, 11, pp. 379–402.
Pitegoff, T. M. (1989). ‘Franchise relationship laws:
A minefield for franchisors,’ Business Lawyer, 45,
pp. 289–331.
Poe, R. (1990). ‘Take the franchise test: Seven traits
of the perfect franchisee,’ Success, 37 (October),
pp. 62–66.
Porter, M. E. (1976). Interbrand Choice, Strategy, and
Bilateral Market Power. Harvard University Press,
Cambridge, MA.
Porter, M. E. (1980). Competitive Strategy. Free Press,
New York.
Powell, W. W. (1987). ‘Hybrid organizational arrangements: New form or transitional development?’, California Management Review, 30, pp. 67–87.
Purvin, R. L. (1994). The Franchise Fraud. John
Wiley, New York.
Roha, R. R. (1996). ‘Making it, franchise style,’
Kiplinger’s Personal Finance Magazine (July),
pp. 71–74.
Rubin, P. H. (1978). ‘The theory of the firm and the
structure of the franchise contract,’ Journal of Law
and Economics, 21 (April), pp. 223–233.
Scherer, F. M. and S. Ross (1990). Industrial Market
Structure and Economic Performance HoughtonMifflin, Reading, MA.
Serwer, A. E. (1995). ‘Trouble in franchise nation,’
Fortune, 131 (March 6), pp. 115–129.
Shane, S. (1996). ‘Hybrid organizational arrangements
and their implications for firm growth and survival:
A study of new franchisors,’ Academy of Management Journal, 39, pp. 216–234.
Shane, S. A. (1998). ‘Making new franchise systems
work,’ Strategic Management Journal 19(7),
pp. 697–707.
Strat. Mgmt. J., 21: 497–514 (2000)
514
S. C. Michael
Shelanski, H. A. and P. G. Klein (1995). ‘Empirical
research in transaction cost economics: A review
and assessment,’ Journal of Law, Economics, and
Organization, 11, pp. 335–361.
Smyth, G. K. (1989). ‘Generalized linear models with
varying dispersion,’ Journal of the Royal Statistical
Society B, 51, pp. 47–60.
Stanworth, J. and P. Kaufmann (1996). ‘Similarities
and differences in UK and US franchise research
data: Towards a dynamic model of franchisee motivation,’ International Small Business Journal, 14
(3), pp. 57–70.
Stern, L. W. and T. Reve (1980). ‘Distribution channels
as political economies: A framework for comparative
analysis,’ Journal of Marketing, 44 (Summer),
pp. 52–64.
Stinchcombe, A. L. (1965). ‘Social structure and
organizations.’ In J. G. March (ed.), Handbook of
Organizations. Rand McNally, Chicago, IL,
pp. 142–193.
Copyright  2000 John Wiley & Sons, Ltd.
Webster, B. (1986). The Insider’s Guide to Franchising. AMACOM, New York.
White, G. C. and R. E. Bennetts (1996). ‘Analysis of
frequency count data using the negative binomial
distribution,’ Ecology, 77, pp. 2549–2557.
Williamson, O. E. (1985). The Economic Institutions
of Capitalism. Free Press, New York.
Williamson, O. E. (1989). ‘Transaction cost economics.’ In R. Schmalensee and R. D. Willig (eds.),
The Handbook of Industrial Organization, Vol I.
Elsevier Science Publishers B. V. New York,
pp. 135–182.
Williamson, O. E. (1991). ‘Comparative economic
organization: The analysis of discrete structural alternatives,’ Administrative Science Quarterly, 36,
pp. 269–296.
Zorn, C. J. W. (1998). ‘An analytic and empirical
examination of zero-inflated and hurdle Poisson specifications,’ Sociological Methods and Research
26, p. 3.
Strat. Mgmt. J., 21: 497–514 (2000)