1 Introduction

Literature on interfirm relationships has extensively analyzed the choice of the governance mechanism,Footnote 1 in which the decision as to whether contracting should be formal or relational is the key. These are the two basic enforcement mechanisms used to protect the commitment reached in complex (non-spot) contracts (Telster 1980). Formal contracting is based on the fact that the agreed terms are easily verifiable and allow for a third-party authority to oblige the parties to comply with them in case of non-compliance (Klein 1980; Dyer 1997). Relational contracting is based on the contractors’ self-interest in fulfilling their commitments so that there are no reasons to terminate the contract, facilitating the continuity of transactions in the future (Klein 1980; Baker et al. 2002).

The prevalence of relational contracting within and between firms has drawn attention of organizational economists (Baker et al. 2002). Departing from the idea that they are informal agreements sustained by the value of future relationships (i.e., when the long-run benefits from continuing the relationship exceed the benefits of reneging on the agreement for short-term gain), this literature explains how self-enforcement mechanism works and its range of application (Klein 1996) and how and why relational contracting differ in different interfirm relationships such strategic alliances, joint ventures, and networks (Baker et al. 2008).

On the other hand, relational contracting, based largely on social identification and trust (Macaulay 1963; Macneil 1978; Dyer and Singh 1998), has been emphasized as the most important governance mechanism for interfirm cooperative relationships (Ring and Van de Ven 1992, 1994; Gulati 1995a). This is because repeated interactions among the same parties over time (i.e., prior ties) provide parties with valuable first-hand information about the exchange (Gulati 1995a; Mayer and Argyres 2004), allowing them to learn how to efficiently structure their agreements (Reuer and Ariño 2007; Ryall and Sampson 2009; Vanneste and Puranam 2010). Parties learn from each other and therefore develop a mutual understanding, shared values and normative conventions that define how they will work together (Poppo et al. 2016, p. 726). They mainly generate knowledge about the partner’s behavior, familiarity, and interorganizational trust. This is what some authors call knowledge-based trust (Yamagishi and Yamagishi 1994), and others relational trust (Ring 1996; Poppo et al. 2016). All these create an “expectation that alleviates the fear that one’s exchange partner will act opportunistically” (Bradach and Eccles 1989, p. 104), facilitating the use of relational governance mechanisms (Gulati 1995a; Zollo et al. 2002; Gulati and Nickerson 2008).

However, this relevance of relational contracting does not seem to hold in all kinds of interorganizational agreements, among which franchising is an example. Franchising is a type of interfirm relationship in which a firm, the franchisor, grants the right to use its brand and a proven business concept to legally independent firms, the franchisees, under certain conditions and in exchange for financial compensation (entry fee and royalties). Some empirical evidence suggests that, in franchising, formal contracting is the main instrument for enforcing the relationship between franchisor and franchisees (Brickley and Dark 1987; Solís-Rodríguez and González-Díaz 2012), even when they have already renewed their contract several times. Why is relational contracting not the basis for franchising?

More generally, there is another type of prior interaction to which the literature on interfirm relationships has paid hardly any attention. This is when a firm reaches similar agreements on transactions over time, but not with the same partners. Repeated interactions in the same transaction allow firms not only to obtain knowledge about the relevant features of such transactions and, therefore, about potential contractual problems, but also how best to solve them (Argyres and Mayer 2007; Argyres et al. 2007). This learning about the transaction allows for cost-efficient development of formal contracts (Poppo and Zenger 2002; Mayer and Argyres 2004; Ryall and Sampson 2009; Vanneste and Puranam 2010), encouraging the adoption of such mechanisms in interfirm relationships.

The aim of this paper is to explain why formal contracting is more prevalent than relational contracting for enforcing franchise relationship while also demonstrating that prior interactions do not necessarily lead to more relational contracting in interfirm relationships. Following Kim (1997) and Taylor and Greve (2006), we argue that the knowledge yielded by prior interactions may differ depending on whether parties learn about the partner and/or about the transaction. If they yield knowledge about the parties, prior interactions enhance relational contracting; but if they generate knowledge about the transaction, they favor formal contracting. Consequently, their effect on the choice of governance mechanism is dual because it depends on the type of knowledge acquired.

Franchise relationships provide a natural scenario to test these arguments. Because of their characteristics, prior ties in franchising generate much more knowledge about the transaction than about the partner, increasing formal contracts but not facilitating relational contracting. This is because, on the one hand, franchise contracts are usually more repetitive than other interfirm relationships (joint ventures, alliances, outsourcing) because franchisors are continuously signing new contracts with new franchisees. It is therefore easier for franchisors to obtain knowledge about the relevant features and problems of the exchange. Franchisors can be considered to have the chance to improve the contract each time a new franchised outlet is opened. This differentiates franchisors from partners in other types of alliance (see for example the description of pharmaceutical and biotech strategic alliances in Baker et al. 2008), in which contracts are frequently signed with the same partners for related type of transaction (see also Gulati 1995b). On the other hand, each franchisee is usually a different firm (except for multi-franchising (Argyres et al. 2016), which is not frequent among small and medium chains), so franchisors cannot develop trust and familiarity with the franchisees. However, franchisees may develop knowledge-based trust in the franchisor, not because they have interacted previously, but because the franchisor’s repeated interactions in the market with other franchisees have given rise to a reputation which indicates its trustworthiness (Kreps 1990). In short, franchisors can develop learning about the transaction but not about the partner, and franchisees can develop learning about the partner but not about the transaction.

Based on the analysis of 74 contracts in Spanish SME franchise chains, our results indicate that chains that have more franchising experience and, supposedly, greater knowledge about the relevant aspects of the exchange include greater detail in their contracts, formalizing a larger number of contingencies regardless of whether the obligations are for themselves or for the franchisees. This is coherent with the thesis of Mayer and Argyres (2004) and Argyres and Mayer (2007) who argue that firms learn how to contract and manage their relationships over time because they develop contract design capabilities. It may seem surprising that the franchisor, who designs the contract and has most bargaining power (Schwartz 1974; Klein 1980), increasingly formalizes his own obligations in contracts. An explanation is that, by doing so, the franchisor is signaling his commitment to both maintain and/or improve his business model, that is, he is guaranteeing his good behavior (Poppo and Zenger 2002). Moreover, we observe that the franchisor’s trustworthiness (i.e., market reputation) does not affect the formalization of his obligations. This means that, for franchisor’s obligations, formal contracting is not substituted by relational governance mechanisms, as several authors have emphasized (Macaulay 1963; Gulati 1995a; Zaheer and Venkatraman 1995; Dyer and Singh 1998). However, we do observe that franchisor’s reputation has a positive and significant effect on franchisees’ obligations. The greater the franchisor’s reputational capital, the greater his vulnerability to franchisee opportunism, so the franchisor has to introduce more contingencies in the contract in order to control, as far as possible, for potential problems. These findings are consistent with Arruñada et al. (2001).

The contribution of the paper is threefold. First, we contribute to interfirm cooperative relationships literature by claiming that relational contracting is not always the most important enforcement mechanism for these relationships, as had been emphasized in the literature (Ring and Van de Ven 1992, 1994; Gulati 1995a). Our results indicate that the contract is the basic governance mechanism in franchising (Brickley and Dark 1987; Solís-Rodríguez and González-Díaz 2012), so formalization matters, at least in this particular type of long-term interfirm cooperative relationships.

Second, as stated by Ariño et al. (2014, p. 380), “learning from prior relationships is complex and requires additional examination.” To the best of our knowledge, this is the first study to show directly that there are different types of learning from prior relationships and that this affects the choice of contract form: learning about the partner enhances relational contracting, and learning about the transaction favors formal contracting. Most papers analyzing the effect of prior interactions on the choice of governance mechanism have focused on interfirm relationships between the same partners over time, so only consider knowledge acquired about the partner. This may explain why relational contracting has received such emphasis in the literature. Knowledge about the transaction only appears if the parties always collaborate in the same type of business so, as established by Vanneste and Puranam (2010), an effort has to be made to distinguish between the two types of knowledge. We disentangle this relationship between the different types of prior relationship and choice of contract form by taking advantage of the specific characteristics of franchising.

Finally, we contribute to the entrepreneurship literature. A challenge faced by entrepreneurs who wish to franchise their business is to know to choose the right form of contract, not only to avoid conflicts but also to attract the best partners (franchisees). The results of this paper indicate that entrepreneurs must understand that formalization of all the different aspects of the franchise relationship, that is, both their own obligations and those of their partners, is the key for success in cooperative relationships. When they formalize their partners’ obligations, they control for potential opportunistic behavior. When they formalize their own, they are signaling their commitment to the business, so not only will their current partners be satisfied but also their image will be enhanced for potential new partners who may apply to join the network.

The remainder of the article is structured as follows. After this introduction, the second section analyzes the existence of two different types of knowledge arising from prior ties and how their existence influences the choice of governance mechanism. The third section explains how the data were collected and gives the sources and models used. Section four discusses the results and finally, conclusions are drawn.

2 Theoretical background and hypotheses

Two distinctive governance mechanisms have been identified to help parties protect their agreements in an efficient way: relational mechanisms and formal contracts (Telster 1980; Dyer and Singh 1998). The literature on interfirm cooperative relationships has largely focused on the first ones, considering them to be the basis for such relationships (Ring and Van de Ven 1992, 1994; Gulati 1995a). From a theoretical perspective, economists define relational contracts as informal agreements sustained by the value of future relationships (Baker et al. 2002). This can be considered a self-enforcing mechanism if long-run benefits from continuing the relationship exceed the short-term benefits of opportunism and termination of the agreement. Relatedly, interfirm cooperative literature has paid attention to a particular type of relational governance mechanisms, which are mainly based on social identification and trust as their enforcement device (Macaulay 1963; Macneil 1978; Dyer and Singh 1998). Governance emerges from the values and agreed-upon processes found in such social relationships (Macneil 1978; Heide and John 1992; Poppo and Zenger 2002). Prior ties and repeated experiences generate familiarity and trust, implying an expectation of less opportunistic behavior (Bradach and Eccles 1989) which can be enough for enforcing the parties’ agreements (Argyres et al. 2007, p. 9). The use of these self-enforcing mechanisms hardly create ex ante writing costs, enhance flexibility, and reduce ex post opportunism (renegotiation costs) although they are costly enforced by a third party such a court (Bradach and Eccles 1989; Zaheer and Venkatraman 1995; Dyer 1997; Uzzi 1997; Gulati 2007).

However, also an increasing range of repeated exchange activities is organized through interorganizational relationships, in which complex contracting plays an important role (Mayer and Argyres 2004; Argyres and Mayer 2007). A formal contract is “an agreement which is legally enforceable or legally recognized as creating a duty” (Atiyah 1989, p. 40). It outlines the rights and obligations of the parties, the location of decision and control rights, how to act in certain contingencies, how the parties should communicate, and how conflicts should be resolved (Argyres and Mayer 2007). This yields important ex ante writing costs and may reduce ex post flexibility to respond to local circumstances but anticipates many possible conflicts which are solved in advance by mutual agreement. The enforcement mechanism is supervision by a third party. As exchange hazards increase, so must contractual safeguards (Klein et al. 1978; Williamson 1985) in order to minimize the cost and performance losses arising from such hazards (Macneil 1978; Joskow 1988; Heide 1994). This increases the level of contractual completeness, that is, the extent to which relevant contingencies are specified in contracts (Luo 2002; Mesquita and Brush 2008; Vanneste and Puranam 2010).

This points to an apparent contradiction in the choice of the mechanism of governance because both relational and formal contracting have been considered the best options for enforcing this kind of repeated long-term relationship. However, according to TCE, this is because interfirm relationships do not all present the same exchange characteristics and, consequently, their optimal organizational form is different. Literature agrees that prior exchange experiences help parties to identify these characteristics and, therefore, the best contract form because they provide parties with valuable first-hand information about the exchange (Gulati 1995a; Mayer and Argyres 2004). But there is no agreement on how prior interactions influence the choice of the mechanism of governance. While many studies indicate that prior interactions allow the emergence of trust and personal ties (Macaulay 1963; Shapiro et al. 1992), facilitating the use of relational governance mechanisms (Gulati 1995a; Zollo et al. 2002; Gulati and Nickerson 2008), others indicate that prior interactions facilitate the development of formal contracts (Poppo and Zenger 2002; Mayer and Argyres 2004; Argyres et al. 2007; Ryall and Sampson 2009; Vanneste and Puranam 2010). How then do prior interactions influence the choice of the mechanism of governance?

2.1 Prior interactions and choice of contract form

Prior interactions provide firms with valuable first-hand information about the exchange (Gulati 1995a; Mayer and Argyres 2004), producing a learning effect (Huber 1991). However, it has not been explicitly indicated that the nature of this learning may differ: from prior interactions parties can learn about the partner and/or about the transaction (Kim 1997; Taylor and Greve 2006). Repeated interactions among the same contractors over time (but not necessarily in the same business) allow firms to develop a better understanding of the partner’s procedures, management system, culture, etc., thus generating familiarity and trust. In other words, repeated interactions with the same partner allow firms to develop learning about the partner (Doz 1996). This learning helps firms to reduce costly negotiations (Zaheer et al. 1998), mitigating ex post coordination, conflict resolution, or information-gathering problems (Reuer and Ariño 2007, p. 317). So, as established by most of the literature on interfirm relationships, partners will use less formal governance structures over time, because the familiarity and trust generated during prior interactions allow for the use of relational governance mechanisms (Gulati 1995a; Zollo et al. 2002; Gulati and Nickerson 2008). Many empirical works support this idea (Bradach and Eccles 1989; Gulati 1995a, 2007; Zaheer and Venkatraman 1995; Dyer 1997; Reuer and Ariño 2003, 2007; Gulati and Nickerson 2008).

However, over time organizations interacting repeatedly in similar transactions (but not necessarily with the same partner) obtain knowledge about the exchange and therefore about the roles and responsibilities of the parties (Argyres and Mayer 2007), developing learning about the transaction. Although there are not so many studies on the effect of this type of learning on contract design (exceptions are Mayer and Argyres 2004; Argyres and Mayer 2007; Argyres et al. 2007; Cochet and Garg 2008; or Vanneste and Puranam 2010), there is plenty of literature indicating the importance of learning within and among organizations (Lieberman 1984; Darr et al. 1995; Argote 1999).

Such learning usually takes place through relatively slow environmental selection of faster learners (Alchian 1950), or the detection and correction of errors in “theories-in-use” (Argyres and Schon 1978), or both. In other words, firms learn by repeating routines and gradually including in them what they learn (Mayer and Argyres 2004). Both transaction cost theory (Williamson 1985) and the literature on organizational learning (Lieberman 1984; Mayer and Argyres 2004) indicate that such learning about the transaction affects contract design. As firms repeat transactions, they develop contract design capabilities: they learn to identify any problems that are likely to arise and how to solve them, and as they do so, they make provisions for them in contracts (Cyert and March 1963). So, as firms acquire experience, they not only learn to better identify the relevant features of transactions but are also able to find the best solution for them (Argyres et al. 2007). All this allows firms to achieve (cost-efficient) development of formal contracts (Poppo and Zenger 2002; Mayer and Argyres 2004; Argyres et al. 2007; Ryall and Sampson 2009; Vanneste and Puranam 2010).

These two types of learning (about the partner and about the transaction) influence the choice of governance mechanism differently. Literature on alliances has extensively studied learning about the partner. Trust emerges from prior collaborative relationships between firms and most interfirm cooperative relationships have in common the development of this trust and social identification through the interaction of personnel across the firms (Hoetker and Mellewigt 2009). Consequently, trust enhances the use of relational mechanisms. This may be why this literature has emphasized the role of this governance mechanism as the most important enforcement mechanism for such relationships.

However, learning about the transaction has mostly been ignored until recently. As stated by Argyres and Mayer (2007, p. 1065), “Economic theories of contracting […] implicitly or explicitly assume that all firms know how to design contract terms that specify roles and responsibilities of the parties […] When parties are bilaterally dependent and when the contract involves complex technology or other kinds of task complexity, properly specifying roles and responsibilities is not always a trivial matter.” The contribution made by these authors is to show that knowledge is needed on key aspects of the transaction and that the key to success may lie therein. Having this knowledge generates heterogeneity in firms’ contract design capabilities, a matter that has mostly been ignored by the literature on contract design.

2.2 A specific type of alliance: franchising

Alliances are a field in which many interorganizational arguments have been tested (Parkhe 1993; Gulati 1995a, 2007; Luo 2002; Reuer and Ariño 2003, 2007; Mayer and Argyres 2004; Ryall and Sampson 2006; Argyres et al. 2007; Gulati and Nickerson 2008; Baker et al. 2008). Franchising, as a type of alliance between two legally independent entrepreneurs, the franchisor and the franchisee, is no exception and has been used to test different (inter)organizational arguments (Brickley and Dark 1987; Lafontaine 1992, 1999; Castrogiovanni et al. 2006; Cochet and Garg 2008).

Franchise relationships present a potential conflict between the parties (Kidwell et al. 2007) because they require a joint effort and costly mutual monitoring. Parties may develop opportunistic behaviors, trying to maximize their personal gain at the expense of the other party. Consequently, the success of a franchise chain will depend on its capacity for avoiding such opportunism. This makes franchising an ideal field for testing our arguments. In addition, franchising has proved to be very much in favor of formal contracts, which are the main instrument used to govern the business-to-business relationship between franchisor and franchisee instead of relational contracting (Brickley and Dark 1987; Solís-Rodríguez and González-Díaz 2012). Given that most repeated interorganizational relationships choose relational mechanisms to govern their exchanges, the franchising choice might seem irrational. However, it is not if we consider the dual learning involved.

Learning about the transaction

Franchising provides a natural framework for analyzing the effect that learning about the transaction has on the probability of using formal contracts. Contractors learn about the relevant details of the exchange when they repeatedly interact in similar transactions (Vanneste and Puranam 2010). This repetition happens more frequently in franchising than in other contracts (joint ventures, alliances, outsourcing). The franchisor is permanently signing new contracts with new franchisees so, as time goes by, he learns more about the relevant features of the exchange than parties in other types of contract do.

To our knowledge, the only study on franchising that directly analyzes the influence of learning about the transaction on contract design is Cochet and Garg (2008). This study analyzes the formal contracts of three German franchise chains, observing that (a) chains gradually introduced new terms in their contracts or re-worded existing ones to adapt to problems arisingFootnote 2 and (b) such amendments to contracts enhanced their control over franchisee behavior. However, although the study analyzes a large number of clauses, many others are omitted so part of the information on contractual design is lost (Argyres et al. 2007). Similar results in other fields were obtained by Mayer and Argyres (2004), Ryall and Sampson (2006), and Argyres et al. (2007).

We can therefore establish that the franchisor learns about the transaction from past mistakes. Once such problematical situations have been identified, provision can be made for them in new contracts (Argyres et al. 2007; Cochet and Garg 2008). The more the franchisor learns about the transaction, the fuller, more sophisticated, and more complete the contracts become (Baker et al. 2002; Poppo and Zenger 2002; Ryall and Sampson 2006). Furthermore, the writing costs of such contractual completeness are quite affordable for the franchisor because (a) he repeats the same contract with all new franchisees (Lafontaine and Oxley 2004), (b) he introduces any improvements gradually (Cochet and Garg 2008), and (c) he aims to work with these franchisees for many years (long-term contract) (Brickley et al. 2006).

The hypothesis is therefore as follows:

H1: The greater the learning about the transaction, the more complete franchising contracts become.

Learning about the partner

The second effect, learning about the partner, has been extensively studied in various types of business relationships (Parkhe 1993; Gulati 1995a; Dyer 1997; Dyer and Singh 1998; Zaheer et al. 1998; Luo 2002; Reuer and Ariño 2003; Ryall and Sampson 2006; Gulati and Nickerson 2008; Dekker and Van den Abbeele 2010). The effect observed is that, in a business relation, when the partners know each other from previous experiences, the probability that they will use relational mechanisms is greater.

In franchising, we find an “unusual” situation in comparison with other interfirm relationships: repeated interactions among the same contractors are less frequent. For example, Gulati (1995b) and Baker et al. (2008) sustain that alliances among the same contractors are frequent and they provide some empirical evidence. Conversely, although the franchisor is repeatedly signing franchise contracts, it is almost always with new partners (except for multi-franchising). Consequently, because partners do not know each other from past interactions together, it is difficult for franchisor and franchisee to learn from each other and therefore develop the mutual understanding, familiarity, and trust needed to facilitate the use of relational governance mechanisms. In sum, it seems that learning about the partner does not apply in this case, so relational governance mechanisms cannot be used. This seems to explain the use of formal contracts (Johnson et al. 2002).

However, the franchisee may know a lot about the franchisor prior to signing the contract, despite not having had any interaction with him. This learning comes from indirect prior interactions, i.e., those that the franchisor has had in the marketplace with other franchisees, which have allowed him to build up a market reputation and gain “trustworthy” status (Kreps 1990).Footnote 3 This franchisor’s reputation therefore emits a credible signal about his behavior in the past, from which potential franchisees interpret that he will behave in a similar way in the future (Parkhe 1993; Gulati 1995a, 2007; Poppo and Zenger 2002).

Consequently, the franchisor’s image for potential franchisees will be good and they will want to join the chain because they consider the franchisor is unlikely to breach the terms of the contract as this would damage his market reputation (Klein 1980; Klein and Murphy 1997; Arruñada et al. 2001). In other words, the franchisor’s market reputation serves as a guarantee that he will not behave opportunistically (Williamson 1983; Klein 1996), which may lead the franchisee to trust him. So franchisees can learn about the partner (in this case about the franchisor), and this trust in the franchisor acts as a relational governance mechanism that may make more detailed contracts regarding their obligations unnecessary (Zaheer and Venkatraman 1995; Dyer 1997; Uzzi 1997; Gulati 2007).

So, the franchisor’s reputation means that the franchisee will not be over-concerned about formalizing contingencies regarding the franchisor’s behavior even if problems are anticipated, and will show trust: formalization is substituted by relational mechanisms regarding franchisor’s obligations. Therefore, we establish the following hypothesis:

H2a: Franchisees’ learning about the franchisor (from the franchisor’s market reputation) reduces the level of contract completeness regarding franchisor obligations. So formalization and relational mechanisms act as substitutes for franchisor obligations.

Franchisors, however, cannot learn about franchisees prior to signing the contract. The previous argument about using reputation to infer the counterpart’s behavior does not work both ways. Unlike franchisors, franchisees are usually small entrepreneurs, so their reputational capital is limited (Solís-Rodríguez and González-Díaz 2012) and therefore does not serve as a guarantee for the franchisor. Furthermore, franchisors may develop selection criteria and learn to recruit appropriate franchisees over time (Castrogiovanni et al. 1993; Forward and Fulop 1993), but this offers very limited knowledge about the partner. Such learning helps weed out unsuitable candidates and prevent opportunism (Stanworth 1991; Ramírez-Hurtado et al. 2011), but does not generate the familiarity and trust that prior interactions create. Consequently, the franchisor prefers to include clauses in the contract to prevent opportunism and to ensure that his instructions are followed.

This inclination towards formal contracting accentuates as franchisor’s brand reputation increases. Reputation is an intangible asset that is difficult to imitate, that “summarizes a good deal of information about firms and shapes the responses of customers, suppliers, and competitors” (Teece et al. 1997, p. 521). As stated above, reputation “can inform external constituents about the trustworthiness, credibility and quality of the firm” (Galbreath 2005, p. 981). Therefore, reputation is clearly a source of sustainable competitive advantage for chains. It is therefore reasonable that the greater the franchisor’s market reputation, the greater his concern to protect it against potential opportunistic behavior on the part of franchisees by formalizing their obligations (Arruñada et al. 2001). Therefore, we establish the following hypothesis:

H2b: The greater the franchisor’s market reputation, the more complete franchising contracts become regarding franchisees’ obligations.

3 Methodology

3.1 Data collection

In order to test our hypotheses, between March 2006 and December 2007, we contacted 805 Spanish franchise chains, by telephone and by e-mail, requesting information on their brands and franchise contracts. Foreign chains operating in Spain were not included. This was because, when franchise chains internationalize their operations, it is common practice for them to use the same contract as in their domestic market (Lafontaine and Oxley 2004), so if they were included in the sample, different degrees of contract completeness might stem from different national regulations.

After applying the different techniques that are recommended for increasing the response rate (Fowler 1993; Dillman 2000), 293 franchisors decided to collaborate and 74 of them sent us their franchise contract. This information was completed with general information on the firms, obtained either from the dossiers sent by them or from their web sites, or from professional franchise guides if the former were not available. Table 1 shows the breakdown by sector of activity for the chains in the sample, all of which are currently active in Spain.

Table 1 Sample distribution
Full size table

In order to check that non-response bias was not a problem in our sample, we used the procedure devised by Armstrong and Overton (1977), and compared the first responses received with the last (the latter being considered representative of the firms that did not respond). The results showed that there were no significant differences between them with regard to contract completeness. We also compared the sectors included in the sample with the population (Poppo and Zenger 2002), and found no differences. We can therefore conclude that our sample is representative of the population under study.

3.2 Description of the model and variables

Since our aim is to analyze the effects on contract completeness of both learning about the transaction and learning about the partner, we run an empirical model which is structured as follows:

$$ {\displaystyle \begin{array}{l} COMPLETENESS={\beta}_0+{\beta}_1 LEARNINGTRANSACTION+{\beta}_2 LEARNINGPARTNER+\\ {}+{\beta}_3 FEE+{\beta}_4 DURATION+{\beta}_5 SERVICES+\varepsilon \end{array}} $$

Our dependent variable is the degree of contractual completeness. As indicated previously, completeness measures the extent to which all relevant contingencies are included in the contract (Luo 2002; Mesquita and Brush 2008; Vanneste and Puranam 2010), with contingencies being understood as each of the specific aspects of the relationship that are covered in the contract.Footnote 4 We therefore needed to identify such contingencies or contractual problems, so we read all the contracts carefully. This enabled us not only to identify contingencies that we were already familiar with from the literature and empirical evidence on franchise contracts (such as contract duration, financial conditions, or sales prices) but also others that were unknown to us as they referred to aspects that franchise chains do not normally make public.

Once we had the list, the next step was to process the clauses in the 74 contracts in order to identify what contingencies were covered in each of them. It should be pointed out here that the number of clauses in a contract is not necessarily the same as the number of contingencies covered in it. A single contingency may be covered in one or several clauses or in part of one. This process was undertaken by the authors separately, with a third party helping to resolve any discrepancies. A total of 157 contingencies were identified.

Bearing in mind the definition of completeness given above, the more contingencies covered in a contract, the more complete it is. Therefore, our measure of contract completeness, CONTINGENCIES, was calculated for each contract as the sum of all the contingencies covered in it.

$$ \mathrm{CONTINGENCIES}=\sum \limits_{i=1}^{157}{Y}_1 $$

where Yi equals 1 if the ith contingency was included in the contract and zero otherwise. The summation term therefore ranges from 0 to 157.

In order to test hypotheses H2a and H2b, we needed to identify the contractual completeness that is related to franchisor’s obligations and the contractual completeness that is related to franchisees’ obligations. This is why we used not only the total number of contingencies (CONTINGENCIES) as the dependent variable but also three alternative variables to measure completeness, one related to franchisees’ obligations (for example, paying royalties every month or following the franchisor’s instructions regarding the establishment), one to franchisor’s obligations (for example, providing training or promoting the chain) and one to contingencies (those which are not obligations for either party, such as contract duration).Footnote 5

This measure of completeness is an improvement on the pre-existing ones. Since we had 74 contracts, we were able (without a survey) to identify all the clauses and contingencies included in each contract, so our measure of completeness is much more accurate than those given in prior studies carried out in other areas as they only took certain clauses or contingencies into account (Parkhe 1993; Saussier 2000; Luo 2002; Reuer and Ariño 2003, 2007; Reuer et al. 2006; Ryall and Sampson 2006; Mesquita and Brush 2008) or calculated the measure indirectly using a Likert scale (Poppo and Zenger 2002; Hendrikse and Windsperger 2011; Hendrikse et al. 2015Footnote 6). This is important because, as stated by Goldberg and Erickson (1987), all the clauses in a contract are chosen simultaneously and interact, so empirical studies should consider them together.

In addition, in order to proxy the learning effects, we used the following independent variables:

  1. a)

    Learning about the transaction. We believe that learning about the transaction is directly related to the franchisor’s experience, so we measured this variable as the number of years that the chains have been franchising (EXPERIENCE). We consider this variable to be a good proxy because firms learn to complete their contracts gradually as they learn from experience what is relevant, which takes time (Klein et al. 1978; Williamson 1985; Mayer and Argyres 2004; Ryall and Sampson 2006). In other words, firms learn when they are repeatedly interacting in similar transactions (Vanneste and Puranam 2010), as in franchising (franchisors are continuously signing new contracts with new franchisees). As Mayer and Argyres (2004, p. 398) indicate, “when projects (in this case, opening of outlets) occur almost simultaneously, there is no real chance to incorporate lessons learned from the first.” Moreover, an older system will generally have faced more different situations than new systems, so “[…] over time the contracts between [the parties: franchisor and franchisee] come to serve as repositories of knowledge about how to efficiently work with each other” (Mayer and Argyres 2004, p. 405). Finally, the number of years in the business (in this case, franchising) has been used in many studies as a proxy for experience (Caves and Murphy 1976; Lafontaine and Kaufmann 1994; Minkler and Park 1994; Dant and Kaufmann 2003; Castrogiovanni et al. 2006) and as an indicator of the learning effect (Mayer and Argyres 2004; Cochet and Garg 2008), in both franchising and other types of alliance.

  2. b)

    Learning about the partner. Previous papers, in the field of alliances and joint ventures, have focused on the existence of prior interactions between the partners to proxy this variable (Parkhe 1993; Reuer and Ariño 2007; Ariño et al. 2014). However, as already explained, repeated interactions among the same contractors are less frequent in franchising: franchisors are continuously signing new franchise contracts but with new franchisees (except for multi-franchising). Consequently, neither franchisor nor franchisee can obtain first-hand information about their partner’s behavior and develop trust because there have been no prior interactions. However, such knowledge may also be obtained from prior interactions by the partner in the marketplace with other firms: these signal a reputation which may indicate trustworthiness (Ryall and Sampson 2006). In the case of franchising, franchisees are usually small entrepreneurs without reputational capital so it is of little use to the franchisor for developing trust. However, franchisees can use the franchisor’s market reputation as an indicator that the latter is unlikely to engage in opportunistic behavior (Williamson 1983; Klein 1996).

    Franchisors’ market reputation was proxied using the SIZE variable, that is, the total number of establishments in each chain. This variable has already been used by other authors such as Lafontaine (1992), Agrawal and Lal (1995), Arruñada et al. (2001), or Penard et al. (2003). According to Lafontaine (1992), franchising is appropriate in businesses in which the brand value is enhanced by exposure.Footnote 7 Consequently, we assume that there is a positive correlation between brand value and the number of establishments displaying it.Footnote 8

Control variables

Contractual completeness also depends on potential for opportunism, that is, on contractual hazards. According to Transaction Cost Economics (Klein et al. 1978; Macneil 1978; Williamson 1985; Klein 1995), parties introduce different safeguards in their contracts in order to mitigate them (Lafontaine 1992; Arruñada et al. 2001; Solís-Rodríguez and González-Díaz 2012). We controlled for contractual hazards using two variables: FEE and DURATION. FEE is the up-front fee that the franchisee pays to join the chain (expressed in thousands of euros), and aims to represent the specific investment made by the franchisee. DURATION is the length of the contract in years and reflects the need to anticipate more contingencies in the contract when it has a longer duration.

Finally, SERVICES is used to control for the sector effect. This variable is a dummy taking 1 for franchise chains in the services sector and 0 for those in retail. The aim here is to control for complexity in the exchange, because it might affect the choice of governance mechanism (Bajari and Tadelis 2001). Business concepts that focus on services might be more complex to manage, and might therefore require more contingencies to be covered in the contract.

An important problem in this model is that most of the variables are endogenous, specifically EXPERIENCE, SIZE, FEE, and DURATION. This is because, as stated above, decisions on contract conditions are determined simultaneously as is, therefore, the degree of contractual completeness (Drahozal and Hylton 2003). This means that estimating the model by ordinary least squares (OLS) could lead to biased results (Wooldridge 2002).

To solve this problem, instrumental variables are used. On the one hand, for the FEE variable, many studies indicate that it is affected by the chain’s capital needs, manager control costs, the importance of the franchisor’s effort for chain success, and brand strength (Lafontaine 1992; Sen 1993; Vázquez 2005). The variables used to measure each of these aspects were initial investment, geographical dispersion, size and experience, respectively. The estimated value of FEE using these variables is appropriate as there is one instrument, geographical dispersion, that explains FEE but does not seem to affect completeness.Footnote 9 The problems with EXPERIENCE, SIZE, and DURATION were resolved using lagged values.

Tables 2 and 3 give the descriptive statistics and correlations between variables, respectively.

Table 2 Descriptive statistics
Full size table
Table 3 Correlations
Full size table

4 Results

Tables 4 and 5 present the results of our estimations. Table 4 shows the results using CONTINGENCIES (that is, the total number of contingencies formalized in the contract) as the dependent variable, while Table 5 estimates the models for three different contingency categories or groups: franchisor’s obligations, franchisees’ obligations, and other contingencies. We show different model specifications in each table because there may be problems of multicollinearity between the different variables, specifically between EXPERIENCE and SIZE. By estimating different models, we aim to prevent distortions and guarantee robust results.

Table 4 Learning effect and degree of contractual completeness (CONTINGENCIES)
Full size table
Table 5 Learning effect and type of contingencies
Full size table

Regarding the first hypothesis, it must be stressed that the parameters for EXPERIENCE are, as expected, positive and significant in all cases (Table 4). The most experienced chains, that is, those that might be able to learn most about the transaction complete their contracts in greater detail, drawing up contracts that cover a larger number of contingencies. This result seems to support hypothesis 1, in line with Mayer and Argyres (2004) and Argyres and Mayer (2007), who establish that not all firms have the same contract design capabilities and that they learn to manage their relations with their partners and to design their contracts over time. This may facilitate the (cost-efficient) development of franchise contracts (Poppo and Zenger 2002; Mayer and Argyres 2004; Argyres et al. 2007; Ryall and Sampson 2009; Vanneste and Puranam 2010). Similar results have been obtained by authors such as Mayer and Argyres (2004), Ryall and Sampson (2006), Argyres et al. (2007), or Cochet and Garg (2008).

Table 4 shows that chains with longer experience, which have therefore learnt more about the transaction, design more complete contracts, introducing a larger number of contingencies. But what types of contingencies are introduced by such chains? Table 5 shows the use of three different groups of contingencies: franchisor’s obligations, franchisee’s obligations, and other contingencies. The EXPERIENCE variable is also positive and significant in all the cases. This indicates that the more experienced chains introduce more contingencies reflecting all the different aspects of the relationship. These results again support hypothesis 1. Over time franchisors learn to identify and introduce in their contracts all the different potential conflicts and contingencies that are relevant in their relationships with franchisees. Conflicts may arise because both franchisor and franchisee have the potential to engage in opportunistic behavior (Brickley and Dark 1987; Lafontaine 1992; Shane 1998; Combs and Ketchen 1999; Bercovitz 2000) so it is necessary to include their obligations in the contract. But conflicts may also arise because of general aspects of the relationship, for instance, contract duration, how the parties will communicate, the applicable law or methods for conflict resolution.

It may seem surprising that the greater the franchisor’s experience in franchising (and, therefore, his learning about the transaction), the greater the number of franchisor obligations included in the contract. The franchisor is the best-established party (Klein 1980) in the contract negotiation process, that is, he has most bargaining power (Schwartz 1974; Klein 1980): the franchisor owns the business that is being traded and is responsible for drawing up the contract and writing the operations manual. Consequently, we should at least expect franchisor obligations to not increase. But why does the franchisor create obligations for himself as a result of learning? A possible explanation is that, by formalizing his obligations, the franchisor is signaling his commitment to both maintain and/or improve his business model and to comply with what has been agreed with the franchisee. This ties in with the idea of Poppo and Zenger (2002) that formalization of the transaction conditions may also act as a guarantee of his good behavior, that is, “well-specified contracts […] promote more cooperative, long-term, trusting exchange relationships” (Poppo and Zenger 2002, p. 708).

Hypothesis H2a establishes that franchisee learning about the franchisor from the franchisor’s market reputation reduces the level of completeness of franchisor obligations. As we can see in Table 5, the parameters for SIZE are not significant when we use the number of franchisor obligations as the dependent variable. Because we use SIZE as the proxy for franchisor market reputation, i.e., a self-enforcing mechanism, this result indicates that even when relational contracts might be available, they do not reduce the degree of completeness (i.e., third party enforcement). In other words, these mechanisms of enforcement do not work as substitutes for franchisor’s obligations.

There are two explanations for this result. First, market reputation may perhaps not be such a credible signal for franchisor’s trustworthiness as has been argued theoretically (Williamson 1983; Klein 1996). This might be because the indirect knowledge gained about the franchisor from his market reputation informs franchisees about different aspects of the franchisor’s behavior than what they can obtain from prior interactions. Market reputation cannot generate the familiarity, trust, and mutual understanding that would come from working together. The latter generates a credible expectation about the partner’s behavior because it is based on direct interactions. However, reputation is based on indirect interactions, so is not so reliable. Therefore, the commitments offered to the franchisee by the franchisor’s reputation are not credible enough for the franchisee to be able to depend on it alone, and formalization becomes necessary. Second, we are not aware of any franchise chain that does not use formal contracting. So, given that formal contracts are always present, Hart (2001) argues that their net effect on relational contracting is ambiguous, which might also explain our lack of correlation between contract completeness about the franchisor’s obligations and relational contracting. On the one hand, the more complete the contract is, the smaller the benefits for the parties from breaching the relational agreements and the greater their incentives to abide by such relational agreements. However, because the formal contract is already developed, the penalty for breaching the relational contract is low since parties can always govern their relation based on the formal contract alone.

When we use the number of franchisee obligations as the dependent variable, we observe that SIZE has a positive and significant impact on them. That is, the greater the franchisor’s reputation, the larger the number of contingencies regarding franchisees’ obligations. This result supports hypothesis H2b. The underlying argument is the franchisor’s fear that the brand name value might be expropriated. The greater the franchisor’s reputational capital, the greater the formal protection against potential opportunistic behavior by franchisees because the damage that franchisees can inflict on the franchisor is greater (Arruñada et al. 2001). This result indirectly supports the idea that learning from reputation is not a two-way argument in the franchisor-franchisee relationship. Since all franchisees are new for the franchisor and are usually small entrepreneurs (so their reputational capital is limited), the franchisor cannot develop indirect learning about them. Therefore, he cannot use relational mechanisms based on trust. Consequently, the only mechanism the franchisor has to control franchisee behavior is the contract, and he does so by introducing more contingencies reflecting franchisee obligations (Mellewigt et al. 2007).

Finally, regarding control variables, we observe that both FEE and DURATION have the expected sign and are significant in all the models (Table 4). These results corroborate the idea that the greater the contractual hazards, the greater the contract completeness in order to attenuate potential opportunism (Klein et al. 1978; Macneil 1978; Williamson 1985; Lafontaine 1992; Klein 1995; Arruñada et al. 2001; Solís-Rodríguez and González-Díaz 2012). However, the SERVICES variable is not significant in any of the models, indicating that belonging to the retail or services sector does not affect contract completeness.

In sum, all these results support the idea that formalization is the basic governance mechanism in franchising. On the one hand, the fact that franchisors are continuously signing new contracts with new franchisees, so have experience in franchising their business, means that franchisors learn about its relevant features, and this is reflected in the inclusion of more contingencies in the franchise contract. This seems logical because, as established by TCE and the literature on organizational learning, as firms gain experience, they learn about the characteristics of their transactions and, therefore, about how to better design their contracts (Klein et al. 1978; Williamson 1985; Mayer and Argyres 2004; Ryall and Sampson 2006). Their experience means that not only do they learn about the relevant features of the transaction but they also identify them more accurately and can find better solutions (Argyres et al. 2007). On the other hand, while a franchisor’s reputational capital may act as a credible signal of his good behavior for the franchisee, it does not reduce the degree of completeness regarding the franchisor’s obligations. This indicates that the indirect knowledge that the franchisee can obtain from the franchisor’s reputation is insufficient to guarantee that the agreement will serve as a self-enforcing mechanism. Therefore, it does not substitute contract formalization, as several authors have suggested for other interfirm relationships (Macaulay 1963; Gulati 1995a; Zaheer and Venkatraman 1995; Dyer and Singh 1998).

5 Conclusions

This paper has tried to explain why relational contracting is not as relevant in franchising as in other interfirm cooperative relationships, in which it has been emphasized as the most important governance mechanism (Ring and Van de Ven 1992, 1994; Gulati 1995a). We argue that the literature to date has considered that prior exchange experiences show a single effect on the choice of the governance mechanism. However, we point to the existence of two different dimensions of learning from prior interactions: learning about the transaction, which arises from repeating the same transactions (but not necessarily with the same partner), and learning about the partner, which results from repeated interactions with the same partner over time (but not necessarily in the same business). We claim that the knowledge generated from these two types of learning based on prior ties differs and determines the choice of contract form, which is mainly between formal and relational contracting.

The differences between franchising and other alliances provide a natural framework for disentangling these learning effects of prior ties. On the one hand, franchise contracts are usually more repetitive than other contracts (joint ventures, alliances, outsourcing) because franchisors are continuously signing contracts with new franchisees. It is therefore easier for franchisors to learn more about the relevant features of the exchange than parties in other types of interfirm contract that involve fewer repetitions. Consequently, this greater learning about the transaction enables the franchisor to detail parties’ commitments in the contract at an affordable cost, increasing the level of contractual completeness (H1). On the other hand, franchisors sign at least as many contracts as parties in other types of alliance, but almost always with new partners (except for multi-franchising). Consequently, because partners do not know each other from past interactions together, it seems that learning about the partner does not apply in this case as in other interfirm agreements. However, franchisees can indirectly develop such learning, not from previous interaction with the franchisor, but because the franchisor’s reputation in the market is a credible signal of his lack of opportunistic behavior (Williamson 1983; Klein 1996), enabling relational contracting. We argue that contracts will therefore be less complete with regard to franchisor’s obligations because relational contracting substitutes the formalization of such obligations (H2a). Conversely, franchisors cannot learn about their franchisees prior to signing a long-term contract with them, not even indirectly. Franchisees are usually small entrepreneurs, so they do not have a market reputation that could be used by the franchisor as a credible signal of appropriate behavior. Consequently, franchisors prefer to limit franchisees’ opportunism by detailing franchisees’ obligations, particularly when franchisor’s brand reputation is large (H2b).

We checked these arguments using 74 different franchise contracts. Our results clearly support our first hypotheses, that is, the greater the learning about the transaction (measured as the chain’s experience in franchising), the greater the contractual completeness of the franchise contract. Specifically, we observe that more experienced chains (a) introduce more contingencies in their franchise contracts and (b) these contingencies cover all the different aspects of the franchisor-franchisee relationship (franchisee’s obligations, franchisor’s obligations, and other contingencies). These results support the argument that not all firms have the same contract design capabilities and that they undergo a gradual learning process on how to manage their relations with their partners and how to design their contracts (Mayer and Argyres 2004; Argyres and Mayer 2007; Argyres et al. 2007). This also suggests that entrepreneurs who wish to franchise their business model and want the franchisor/franchisee relationship to be conflict-free need to learn how to translate their experiences and knowledge of day-to-day problems into new contractual provisions and safeguards. It might seem surprising that experienced franchisors formalize their own obligations increasingly over time, particularly because they are known to have the greatest bargaining power in the negotiation process (Schwartz 1974; Klein 1980). However, this is because they are signaling their commitment to the business model and are therefore unlikely to behave opportunistically (Poppo and Zenger 2002). This may work as a recruiting policy to attract and retain the best franchisees. By formalizing their obligations, franchisors not only keep their current partners satisfied but also improve their image among potential new partners who may therefore show interest in joining the network.

Our results do not support hypothesis H2a because we observe that franchisee learning about the franchisor (resulting from the franchisor’s market reputation and, therefore, indicating his trustworthiness) does not reduce the degree of contractual completeness regarding the franchisor’s obligations. This indicates that relational governance mechanisms do not substitute formalization of franchisor’s obligations. A possible explanation is that franchisees do not see franchisors’ self-enforcing mechanisms based on market reputation as such a credible signal as legal contracts. We also observe that franchisor reputation has a positive and significant effect on franchisee obligations and other contingencies. This result supports the idea that the greater the franchisor’s reputational capital, the greater the tendency to introduce more contingencies in the contract in order to control all these potential problems as far as possible. These findings are consistent with Klein (1980) and Arruñada et al. (2001).

In sum, all these results support the idea that formalization seems to be the most prevalent governance mechanism in franchising instead of relational contracting. Franchisors prefer formal contracting because it is feasible and affordable for them (they learn gradually how to design the contract and they share writing costs out among many franchisees), and the formalization of their own obligations sends a credible signal about their commitment to the chain. Furthermore, relational contracting is hardly used either because franchisees do not have enough reputational capital or because franchisees do not consider franchisors’ self-enforcing mechanisms as such a credible signal (of franchisors’ behavior) as formalization for such a long-term and complex contract. In other words, formalization is always necessary to enforce franchise agreements, regardless of the presence of relational contracting.

These results leave important research questions open. First, relational contracting seems to play an important role in multi-franchising. Although most chains use the same contract for both multi-unit and single-unit franchisees, the advantage of having had prior ties (i.e., more than one store) seems to be the reduction of judicial conflicts among parties (Argyres et al. 2016). Assessing the role of relational governance mechanisms in franchising is an important item in our research agenda. Second, Transaction Cost literature mostly agrees that organizations whose transactions are appropriately governed are more likely to perform better than those whose transactions are inappropriately governed (Silverman et al. 1997; Nickerson and Silverman 2003). Checking if the formalization practices adopted according to the theory enhance franchising performance also constitutes an important item in our research agenda. Third, Hart’s (2001) argument about the dual effect of formalization on the interest of breaching relational contracting is a challenge that cannot be directly tested with our data set, but could be testable if information is added about the relevance of relational contracts in different franchise chains.

Finally, we know that this study is not without limitations. Although we believe that the way we measure completeness is a contribution, the variables used to proxy relational contracting should be improved. Furthermore, we present a static picture of a dynamic reality, so it would be necessary to observe the evolution of this completeness in the same set of franchise chains over time.