International governance mode choice: Evidence from Brazilian franchisors

The purpose of this article is to test a novel, integrative theory of governance mode selection in the context of international franchising from Brazil. Given the Brazilian emerging market context, we added and tested another variable in the model relating to environmental distance between the home (Brazil) and host countries. To test the theory, we employed two logistic regression models and representative data from the Brazilian Franchising Association in addition to the World Bank’s Doing Business index, and the CEPII (Centre d’études prospectives et d’informations internationales). Findings show that three factors influence Brazilian franchisors’ choices regarding international governance modes: the environmental uncertainty of the host countries, their behavioral uncertainty, and the franchisors’ financial capacity. However, the environmental distance does not influence internationalization strategies of Brazilian firms. Emerging markets firms’ governance modes of entry can be adequately explained with the existing model despite contextual differences.


Introduction
One of the most discussed areas in the franchising literature is international governance (Rosado-Serrano et al., 2018). When a company adopts franchising as a way of entering foreign markets, it can choose different types of governance modes that vary in terms of residual income, ownership, and the allocation of decision rights (Jell-Ojobor and Windsperger, 2014). Due to the complexity of these decisions, Jell-Ojobor and Windsperger (2014) proposed an integrative multi-theoretical model to assess both the organizational economics and strategic management factors that might influence the franchisors' choice. They evaluated the model empirically through a case study with two franchisors from Germany and France (Jell-Ojobor and Windsperger, 2017), but the authors suggested conducting quantitative research in different markets.
Following this suggestion, this study quantitatively tests the conceptual model in an emerging market. Brazil offers an interesting context in which to investigate the franchising governance mode choices, both empirically and theoretically. Empirically it is interesting because most of the work on franchising over the last two decades has used data from the United States, Canada, and Western Europe, despite the relevance and growth potential of franchisors headquartered in emerging economies (Aguiar et al., 2017;Lafontaine, 2014;Nielsen et al., 2018;Wright et al., 2005). Few studies of international franchising from Brazil exist, but the market is In franchising joint ventures, the franchisor enters into an equity relationship with the foreign partner to set a joint venture company, making a master franchising or area development agreement with the joint venture company to expand the system. The franchisor and the joint venture partner share the ownership, decision, and residual income rights (Jell-Ojobor and Windsperger, 2014;Rosado-Serrano et al., 2018).
The major low control franchising governance modes are franchisee-operated arrangements. For instance, in direct franchising, the international franchisor sells the business concept to an individual franchisee in the host country. In a master franchising agreement, the franchisor grants to a master franchisee the right to operate franchise units and to give franchise rights to third parties. In an area development agreement, the franchisor grants the franchisees a defined territory in which they can develop units. In these modes, the local partner has only decision and residual income rights (Jell-Ojobor and Windsperger, 2014;Rosado-Serrano et al., 2018).
Moreover, the majority of the entry and governance mode studies have been focused on developed countries, despite the growing relevance of emerging markets, and manufacturing, despite the growing importance of the service sector. Franchise chains from emerging markets have different historical, cultural, economic, and social contexts. Their particular institutional environment requires further investigation to better understand their paths of international expansion (Aguiar et al., 2017;Lafontaine, 2014;Melo et al., 2019;Nielsen et al., 2018).

Specificities of emerging markets
The increasing academic interest in multinationals headquartered in developing and emerging markets has generated a debate on whether the existing theories based on the analysis of such firms in advanced economies can explain these companies' strategies or whether they require new and specialized theories. One view argues that existing theories can explain both developed and emerging market firms' decisions (Rugman, 2010). A second view asserts that emerging market firms behave differently, and new theories are needed (Guillén and García-Canal, 2009;Mathews, 2006). A third view claims that multinationals based in developing and emerging countries can help modify and extend the existing theories (Alon et al., 2011;Cuervo-Cazurra, 2012;Cuervo-Cazurra and Genc, 2008;Cuervo-Cazurra and Montoya, 2018;Ramamurti, 2012).
Regarding the international franchising governance mode choice, the Goldilocks debate engenders the need to investigate if the mainstream theoretical explanations derived from advanced economies are adequate to explain the decisions of emerging market franchisors. The franchisor's governance mode decisions are intrinsically related to the institutional environment of the host markets. The traditional international franchising literature posits that host markets' environmental uncertainty increases the transaction costs associated with monitoring and coordination activities. Therefore, franchisors allocate more ownership and decision rights to local partners to count on their knowledge and experience, choosing high control modes in underdeveloped and weak institutional environments (Hoffman et al., 2016;Windsperger, 2017, 2014).
However, emerging markets have environmental uncertainties borne out of lesser economic development and weaker institutional settings. Malfunctioning or missing institutions prevalent in emerging economies increase transaction costs. The institutional voids (where existing institutional arrangements and support are absent, weak, or fail) require the development of firm capabilities to address and manage these voids and their related transaction costs. Emerging markets multinationals, as a result, are required to be flexible, willing to take risks, and develop a higher tolerance and resilience to instability (Cuervo-Cazurra, 2012;Doh et al., 2017;Khanna et al., 2005;North, 1991North, , 1990Ramamurti, 2012).
The ability to interact with institutional voids in their domestic market may endow emerging market franchisors with greater competencies in dealing with transaction costs as compared to firms from developed economies. The similarity between the emerging markets' institutional environments may help their firms to deal with authoritarian political regimes, corruption, fluctuation in law enforcement, and other related issues (Aguiar et al., 2017;Cuervo-Cazurra, 2012;Cuervo-Cazurra and Genc, 2008;Lanfranchi et al., 2020;Melo et al., 2019).
Thus, the conditions in the emerging markets' home country would affect the ability of their companies to deal with the transaction costs involved in similar environments and influence decisions about host market selection and governance modes. For this reason, emerging market companies would prioritize host countries with similar institutional environments and internalize foreign operations more efficiently, adopting high control modes in other emerging markets.
Nevertheless, despite having typical features such as the experience dealing with uncertainties in their home country, emerging markets are heterogeneous and differ in initial conditions and competitive outcomes. Since they have different levels of institutional and factor market development, the transaction costs and resources needed to achieve competitive advantage vary significantly. The heterogeneity implies that transaction costs also exist between different emerging countries (Estrin et al., 2018;Hoskisson et al., 2013;Kedia et al., 2015;Wright et al., 2005).
Besides, Brazil is a mid-range emerging economy, with institutions, infrastructure, and factor market semi-developed. Firms from mid-range emerging economies consist of a diverse group of companies with significant organizational heterogeneity, and for this reason, they follow different strategic options (Hoskisson et al., 2000(Hoskisson et al., , 2013Wright et al., 2005). As a result, some Brazilian companies seek host countries with similar environments, and others might prioritize better market and institutional conditions in their expansion strategies. The similarity between home and host countries may not be the only factor influencing their governance mode choices. Other factors can play a role in their decisions, as the market potential and the host country environment, despite the differences in their business environments.
Thus, not all behaviors of Brazilian franchisors are necessarily different from those of companies in advanced economies. They may share many characteristics. For this reason, we investigate the Brazilian franchisors' international governance mode decisions in light of multidisciplinary general theories derived from advanced economy studies, including features about the specificities of the emerging markets.

Integrative model and hypotheses
Jell-Ojobor and Windsperger (2014) proposed an integrative model that includes organizational economics and the strategic management variables that can influence the choice of international governance mode. The organizational economics perspective highlights the effect of the choice of governance mode in reducing costs and potentially increasing residual income. The strategic management perspective focuses on the impact of this choice on the company's competitive advantage. Thus, the integrative model considers both concerns related to reducing costs and the issue of competitive advantage.
From the organizational economics perspective (transaction cost and agency theories), the authors analyze environmental uncertainty, behavioral uncertainty, and transaction-specific investments. The strategic management perspective (resource-based, organizational capability, and property rights theories) considers system-specific assets, local market assets, financial assets, and intangible assets. Jell-Ojobor and Windsperger (2017) empirically evaluated the integrative model through a case study analysis of the European automotive rental industry, and Hoffman et al. (2020) tested the model in a sample of US franchisors. The findings showed that both organizational economics and strategic management factors are relevant for developed country franchisors' choice of governance mode.

Transaction cost theory and agency theory perspectives on the choice of international franchise governance mode
According to transaction cost theory and agency theory, the choice of governance mode in international markets is related to the franchisors' goal of minimizing costs. Several transaction and agency costs can be associated with international franchising. Examples include the costs of monitoring international operations, transferring knowledge to distant markets, and protecting property rights, as well as possible opportunism or shirking by the local partners. Thus, franchisors choose the governance mode most likely to reduce these costs, given factors such as the environmental uncertainty of the host markets and the behavioral uncertainty of the franchisees (Baena, 2012;Burton et al., 2000;Contractor and Kundu, 1998a;Fladmoe-Lindquist and Jacque, 1995;Grewal et al., 2011;Jell-Ojobor and Windsperger, 2014;Mumdžiev, 2011).
In international markets with different socio-cultural, economic, and institutional environments, the costs related to monitoring, adaptation, information transfer, and property rights protection are higher. The uncertainties and risks of host markets make it difficult to control and coordinate foreign operations. The franchisors' lack of information about the specificities of host countries that suffer from environmental uncertainty requires them to delegate more tasks to their local partners (Hoffman et al., 2016;Windsperger, 2017, 2014).
In other words, the environmental uncertainty of host markets increases the franchisors' transaction costs related to monitoring and coordination activities. Therefore, they will be inclined to allocate more decision rights to franchisees. As a result, in host countries with greater environmental uncertainty, franchisors that choose lower control modes count on their local partners' knowledge about the specificities of those markets and give them the flexibility to deal with changes (Baena, 2012;Burton et al., 2000;Fladmoe-Lindquist and Jacque, 1995;Jell-Ojobor and Windsperger, 2014;Mumdžiev, 2011). Thus, the transaction costs incurred due to environmental uncertainty in host markets increase the likelihood that franchisors will choose lower control modes (Jell-Ojobor and Windsperger, 2017Windsperger, , 2014Mumdžiev, 2011). Hence, we posit the following hypothesis: Hypothesis 1a. The greater the environmental uncertainty of the host country, the more likely a franchisor will choose a low control mode in foreign markets.
We extend the model of Jell-Ojobor and Windsperger (2014) by including features about the specificities of the emerging markets, which might be relevant for internalization or modes of entry. For some scholars, being used to institutional voids can enhance the emerging market firms' ability to deal with transaction costs in similar environments (Cuervo-Cazurra, 2012;Ramamurti, 2012). Under this condition, emerging markets multinationals might internalize foreign operations in other emerging markets that have similar institutional arrangements. In other words, franchisors will choose high control modes since they have experience on operating in environments that are similar to their home country economy, depending less on local partners. Conversely, when the environmental distance between home and host countries increases, franchisors are inclined to allocate more decision rights to local franchisees to count on their knowledge and experience.
Thus, we propose a second model (Model 2) that includes the environmental distance between the home and host country instead of the environmental uncertainty variable of Model 1, according to following hypothesis: Hypothesis 1b. The greater the environmental distance between the home and host countries, the more likely the Brazilian franchisors choose a low control mode in foreign markets.
Successfully transferring information to distant locations is a challenge to franchisors, increasing the level of behavioral uncertainty. When the geographic distance between the home and host countries is greater, the information asymmetry and the risks of opportunistic behavior, free-riding, and shirking are greater (Castrogiovanni et al., 2006;Combs et al., 2004;Fladmoe-Lindquist and Jacque, 1995;Windsperger, 2017, 2014;Mumdžiev, 2011;Watson et al., 2005).
Hence, the greater the geographic distance between home and host country, the greater the behavioral uncertainty and the agency problems derived from it. The geographic distance increases the monitoring costs needed to reduce opportunistic behavior and the agency costs related to the selection of poor employees abroad and possible moral hazards. By choosing lower control modes, franchisors shift the costs of these problems to the local partner. The franchise contracts align the incentives of the local partner with those of the franchisor and reduce agency costs (Contractor and Kundu, 1998a;Fladmoe-Lindquist and Jacque, 1995;Grewal et al., 2011). Thus, lower control modes can reduce monitoring costs and mitigate agency problems: Hypothesis 2. The greater the behavioral uncertainty of the host country, the more likely a franchisor will choose a low control mode in foreign markets.

Resource-based theory and organizational capability theory perspectives on the choice of international franchise governance mode
Resource-based theory and organizational capability theory perspectives on the choice of international franchise governance modes focus on the creation of value from the franchisors' assets. Franchisors gain a competitive advantage by exploring and exploiting firm-specific resources and capabilities (Erramilli et al., 2002;Hussain et al., 2018). When a firm enters a foreign market, it must transfer its resources and capabilities to foreign operations. Therefore, it must choose an entry and governance mode that can accomplish these goals effectively (Erramilli et al., 2002;Hussain et al., 2018;Jain et al., 2015;Windsperger, 2017, 2014;Moraes et al., 2008).
The resource-based perspective point outs that franchisors must have sufficient resources such as financial capital, brand name recognition, and managerial and routine-processing know-how before internationalizing (Mariz-Pérez and García-Álvarez, 2009;Perrigot et al., 2004). Therefore, the allocation of these critical resources and capabilities needed to operate in foreign markets, such as system-specific and financial assets, between the franchisor and partners in international markets influences the choice of governance mode (Jell-Ojobor and Windsperger, 2014).
Financial assets are related to the ability of the franchisor and the local partners to fund the operations in the host market. Franchisors that need capital to grow use franchising to obtain resources for their expansion. Therefore, they will rely on investment fees and royalties to fund their expansion. If the franchisor has financial constraints, the local partner's resources will be necessary for the international expansion. The international expansion increases the franchisor's costs significantly (Castrogiovanni et al., 2006;Combs et al., 2004;Windsperger, 2017, 2014).
Lower control modes such as master franchising and area development agreements can make international expansion less expensive for the franchisor. Thus, we posit that franchisors with limited financial resources to internationalize their operations will favor lower control modes and count on their local partners' financial assets. On the other hand, franchisors that have financial resources from their domestic operations might choose high control modes in their international operations (Combs et al., 2004;Windsperger, 2017, 2014;Melo et al., 2015).

Hypothesis 3. The greater the franchisor's financial ability to invest in international expansion, the greater the likelihood of choosing a high control mode in foreign markets.
System-specific assets are related to the franchisor's resources and capabilities. Other factors important in this area are capabilities that are difficult to codify, transfer, and imitate such as business know-how, brand name, marketing, R&D capabilities, advertising, and promotion. When the franchisor's system-specific assets are more relevant than the franchisee's local market assets in creating residual income, the franchisor will retain the decision rights (Castrogiovanni et al., 2006;Fladmoe-Lindquist, 1996;Windsperger, 2017, 2014;Windsperger and Dant, 2006).
One of the essential system-specific assets of the franchisor is brand name. Previous studies have underscored the importance of intangible resources such as brand name and reputation for the success of a franchise network. The research highlights brand reputation as a franchisor-owned strategic asset that gives franchisors a competitive advantage in international expansion (Barthélemy, 2008;Elango, 2007;Fladmoe-Lindquist, 1996;Ghantous and Das, 2018;Gillis et al., 2014;Grewal et al., 2011;Perrigot et al., 2013;Rocha et al., 2018;Shane, 1996). Franchisors that count on a reliable brand name as an essential asset and the competitive advantage derived from it will be more likely to retain a higher level of control (Alon, 2006a(Alon, , 2006bJain et al., 2015). Therefore, franchisors with a prestigious brand name are more likely to keep their decision and ownership rights.
Another system-specific asset critical to the franchising system is business know-how, including factors such as store layout, product mix, operating procedures, location selection priorities, monitoring capabilities, advertising, and R&D. Such know-how is unique to the franchisor, and thus considered a proprietary advantage. As with brand name, business know-how is difficult to transfer across firm boundaries and cannot be easily imitated. Therefore, it is quite likely to generate rents (Jell-Ojobor and Windsperger, 2017Windsperger, , 2014Shane, 1996). Thus, we propose the following hypothesis: Hypothesis 4. The more relevant the system-specific assets of the franchisor are, the greater the likelihood of choosing a higher control mode in foreign markets.

Sample
To test our hypotheses, we used a quantitative, descriptive approach with a representative sample of data from the Brazilian Franchising Association -ABFfrom 2018. According to ABF, in 2018, there were 145 Brazilian franchisors with international operations in 114 countries. The database used in this study provides information about 108 Brazilian franchisors, 74.5% of the total number of Brazilian franchisors operating abroad, and the 55 countries where those franchisors had operations in 2018. Thus, in total there are 267 observations in our study.
In addition, we also used the 2018 data from the World Bank's database Doing Business, which contains information regarding the institutional environment in the countries in which Brazilian franchisors have operations. We also utilized a database from CEPII (Centre d'études prospectives et d'informations internationales) with data about the geographic distance between home and host countries. We adopted the country classifications of the World Economic Outlook from the International Monetary Fund as to whether they were advanced economies or emerging and developing economies. All of our measures refer to 2018. Table 1 summarizes the variables adopted, measures, and sources of data. The dependent variable is a dummy indicator of high control and low control modes in each foreign market. The variable received the value of 1 if the franchisor adopted at least one high control mode-company-owned outlets. Those that adopted only low control modes in each market-franchisee-owned outlets, master franchisee outlets, and area developers' outlets-received the value of 0.

Variables and measures
The hypotheses involve several independent variables. Environmental uncertainty is a composite measure obtained from the World Bank's Doing Business database (Ang and Michailova, 2007;Dikova and Van Witteloostuijn, 2007;Michailova and Ang, 2008). As the business environment is a crucial factor in international franchising (Baena, 2012;Fladmoe-Lindquist and Jacque, 1995;Grewal et al., 2011), information from this source was suitable for assessing environmental uncertainty.
To compute the proxy for environmental uncertainty, we adapted to the franchising context previous composite measures adopted by entry mode studies to reflect the institutional environment (Ang and Michailova, 2007;Dikova and Van Witteloostuijn, 2007;Michailova and Ang, 2008), selecting specific indicators from the World Bank's database (ease of starting a business, protecting minority investors, enforcing contracts, and access to credit) used as driving factors influencing international franchising governance mode decisions (Baena, 2012(Baena, , 2013Contractor and Kundu, 1998a;Grewal et al., 2011;Lanfranchi et al., 2020).
Institutions and regulations related to the application of the law and contract enforcement are important as they protect the franchise chains' intangible assets, such as brand and know-how. Previous evidence shows the relationship between contract enforcement efficiency and international expansion via franchising (Baena, 2012(Baena, , 2013Contractor and Kundu, 1998a). An environment condition that facilitates business transactions, such as friendly credit and bureaucratic procedures, and good conditions to Table 1 Variables, measures and sources of data.

Variables
Measure Sources Dependent variable High-control modes (1)/lowcontrol modes (0) Dummy indicator: Franchisors that adopt at least one high-control mode in each market (1)/franchisors that adopt only low-control modes in each market (0) Brazilian Franchising Association database (2018) Independent variables Environmental uncertainty Composite measure of the indicators: "ease of starting a business", "getting credit", "protecting minority investors", and "enforcing contracts" Dummy indicator of the country classification between "Emerging and developing economies" (1) and "Advanced economies" (0) World Economic Outlook -IMF small investors, is also crucial to franchise operations (Grewal et al., 2011;Lanfranchi et al., 2020). We averaged each country's rank position for the selected indicators, and the resulting composite measure provides information about the country's performance relative to the performance of other countries in a way that higher scores represent greater environmental uncertainty ("Doing Business", 2019).
We measured behavioral uncertainty using the geographic distance between the home and host countries (Fladmoe-Lindquist and Jacque, 1995; Jell-Ojobor and Windsperger, 2017Windsperger, , 2014, based on the geographic definition of the "distance between two points on the surface of the earth, as given by latitudinal and longitudinal coordinates" (Beugelsdijk et al., 2018). To obtain the information needed for these calculations, we used a database from CEPII (Centre d'études prospectives et d'informations internationales) (Mayer and Zignago, 2011). We then calculated the geographic distance in kilometers between the 55 countries' capitals and Brazil's capital (Brasília).
System-specific assets consisted of the franchisors' brand name and business know-how. The proxy used to measure brand name is the existence of an advertisement fee applied to franchisees. Many empirical studies that evaluate the relevance of the brand to the international franchising from a resource-based perspective have used this proxy. The more important the franchisor's brand name assets, more revenues from advertising fees are required to maintain the brand value through investments in communication and advertising (Lafontaine and Shaw, 2005;Mariz-Pérez and García-Álvarez, 2009;Perrigot et al., 2013;Windsperger, 2004;Windsperger and Yurdakul, 2007). The proxy used to measure the business know-how, also based on previous studies (Windsperger and Yurdakul, 2007), is the number of training courses and technical support areas that the franchisors offer to the franchisee. The number of training courses and technical support areas required is greater if the franchisor's business know-how is more specific and challenging to transfer.
The franchisees' investment, along with the on-going taxes charged by the franchisors in the domestic market, contributes to the franchisor's financial capacity to fund its international expansion. The franchisors report the minimum and maximum initial investment required to the Brazilian Franchising Association, and we calculated the average initial investment. The royalties are an on-going payment made by the franchisee. As there are several ways to charge royalties such as a fixed monthly fee, a percentage of gross sales, or a percentage per transaction, we considered just the existence of regular on-going payments, regardless of the collection mode. Thus, we devised a composite measure to proxy for the franchisors' financial assets that considered the existence of both an initial investment and on-going payments.
To measure institutional distance, international business researchers have usually used data from the World Bank, the Heritage Foundation EFI, and ICRG scores (Beugelsdijk et al., 2018). In order to include the effects of the home market and measure the distance between the home and host countries, we used the environmental uncertainty composite measure previously calculated, based on the four business environment indicators (ease of starting a business, protecting minority investors, enforcing contracts, and access to credit) to compute the Euclidean distance between each of the 55 host countries and Brazil's position in the composite measure. For cultural and institutional distances, Euclidean distance results do not differ from other distance indexes (Beugelsdijk et al., 2018), and it is widely employed in international business studies (Barkema and Vermeulen, 1997;Drogendijk and Slangen, 2006). The smaller the absolute value of the distance between Brazil and a specific country, the more similar their business environment. For instance, as Brazil is a mid-range economy, it is nearer to other BRICS countries, such as India and South Africa, and more distant from both developed countries as Australia and poor ones as Guinea-Bissau.
In addition to these independent variables, we also controlled for four other variables. First, we considered the franchisors' age, calculated as the number of years since their founding. As firms get older, they gather more experience in partner and site selection, monitoring, and control, which is essential to develop international operations (Alon et al., 2012(Alon et al., , 2013Alon and McKee, 1999;Burton et al., 2000;Dikova and Brouthers, 2016;Madanoglu et al., 2017;Ni and Alon, 2010).
Second, we controlled for the size of the network, measured by the number of domestic operations of the Brazilian franchisors. Several studies have documented that a firm's size affects its choice of mode of entry and governance. The size indicates the availability of resources that a firm can commit to the international operations and also its ability to manage distant localized partners, influencing the governance mode decision (Alon and McKee, 1999;Brouthers and Hennart, 2007;Burton et al., 2000;Kundu, 1998a, 1998b;Dikova and Brouthers, 2016;Madanoglu et al., 2017). Other studies have established that a firm's size affects the relationship between the institutional aspects of the host countries and the level of commitment abroad (Melo et al., 2019). We also controlled for the franchisors' sectors according to the ABF classification that categorizes business activities into 11 sectors: food; construction; technology and electronics; recreation; travel and hotel; cleaning and maintenance; fashion; health and personal care; automotive; business services; and education.
In Model 2, we controlled for the host countries' classification as an advanced economy or an emerging and developing economy by assigning a dummy variable the value of 1 if the company was from an emerging country. By measuring the distance between the home and host countries and controlling for the host countries' classification, we sought to determine whether the similarity between the home and host countries' environments affected the choice of international governance mode (Cuervo-Cazurra, 2012; Cuervo-Cazurra and Genc, 2008; Surdu et al., 2018;Wright et al., 2005). Table 2 shows the means, standard deviations, and correlation coefficients for all of the variables under study. We conducted a binomial logistic regression analysis in order to test the hypotheses presented. We used this statistical model because of the dichotomous nature of the dependent variable and the mix of continuous and categorical independent variables (Hair et al., 1995). We followed other studies that used logistic regressions to investigate choices about international entry and governance modes (Ang and Note: High control mode (1 = high-control, and 0 = low-control); System-specific: Brand (1 = apply advertisement fee, and 0 = do not apply); Financial: Royalties (1 = apply royalties, and 0 = do not apply); Host country classification (1 = emerging and developing economies, and 0 = advanced economies). Coefficients are statistically significant at the *p < 0.05; **p < 0.01 levels. Michailova, 2007;Chiao et al., 2010;Kundu, 1998a, 1998b;Dikova and Van Witteloostuijn, 2007;Erramilli et al., 2002;Fladmoe-Lindquist and Jacque, 1995;Meyer et al., 2009). In the binomial logistic regression, the regression coefficients estimate the impact of the independent (or control) variables on the probability that the international franchishor will choose a high control mode (which carries the value of 1). The model can be expressed as

Research design
where Y is the dependent variable, and Z is the linear combination of the independent and control variables. That is, where β 0 is the intercept, β 1 …β n are the regression coefficients and X 1 …X n are the independent and control variables. To facilitate the comparisons, we normalized all of the non-binary variables by subtracting the mean and dividing by the standard deviation. 1 Another econometric issue is that the data might be correlated with several operations of the same franchisor. To address this issue, we used robust standard errors clustered at the firm and country-of-operation level (Cameron and Trivedi, 2005). We conducted the analysis by creating two models. Model 1 estimates the effects of the first set of predictor variables (environmental uncertainty, behavioral uncertainty, system-specific assets, and financial assets) on the likelihood of a firm choosing a high control mode. Model 2 estimates the effects of the second set of predictor variables (home and host countries' environmental distance, behavioral uncertainty, system-specific assets, and financial assets) on the likelihood of choosing a high control mode. By comparing the two models, we can determine whether the similarity between the home and host countries has an impact on the choice of international governance mode. If there is an effect, the result might suggest that firms from emerging and developing countries behave differently than those from advanced economies. In contrast, if there is no effect, the result might suggest that Brazilian franchisors follow the same patterns reported in the traditional governance mode literature. Table 3 reports the results for Models 1 and 2.

Hypotheses related to the theories of transaction and agency costs
First, we investigate which model fits well with the emerging markets franchisors' decisions on the governance modes. In Model 1, we tested the influence of the host countries' environmental uncertainty in governance mode choice. The results of the Model 1 support Hypothesis 1a. Consistent with the traditional international franchising governance mode literature based on transaction cost theory and agency theory (Baena, 2012;Fladmoe-Lindquist and Jacque, 1995;Jell-Ojobor and Windsperger, 2014), environmental uncertainty had a significant (p < 0.01) negative relationship with the likelihood of choosing a high control mode. Thus, when the environmental uncertainty increases, the probability of a firm choosing a high control mode decreases.
Moreover, considering all of the variables on average, the likelihood of an outstanding franchisor choosing a high control mode was 8.67%. The probability of choosing a high control mode increased to 29.0% if the franchisors operated in countries with less environmental uncertainty (all other variables remained constant). In other words, considering that higher scores in the composite measure correspond to greater environmental uncertainty, in countries with 1 standard deviation below the average franchisors are more likely to choose a high control mode. For instance, they are more likely to select a low control mode in India and South Africa than Ireland and France.
The results of Model 2 for the distance between home and host countries were not statistically significant. As a matter of robustness, we also tested the distance between the home and host countries measured with Kogut and Singh's index, but the results remain not statistically significant. Thus, Hypothesis 1b is not supported. Findings of Models 1 and 2 imply that Brazilian franchisors consider the environmental uncertainty of host countries a primary factor in their choice of the governance mode, regardless of the home country's conditions and its relative distance to the host countries. Thus, the traditional transaction cost theory and agency theory arguments for the choice of international governance mode choice also apply to the franchisors from Brazil that internationalize their operations. Similarities between the home and host countries are not relevant to Brazilian franchisors' decisions about their international governance mode.
Moreover, the control for the host countries' classification was also significant in Model 2 (p < 0.01) and indicated a negative relationship between operating in emerging and developing host countries and the likelihood of choosing a high control mode. In other words, Brazilian franchisors do not choose high control modes in other emerging and developing countries because of their similar home country conditions. With all variables on average, the probability of choosing high control declined from 37.0% when the host country was an advanced economy to 4.13% when the host country was an emerging or developing economy.
Behavioral uncertainty, measured by the geographic distance, had a significant (p < 0.05) and negative relationship with the likelihood of choosing a high control mode. Thus, in countries that are further from the home country, franchisors are more likely to choose a low control mode. As the geographic distance from the home market increases, so does behavioral uncertainty. This result supports Hypothesis 2. The probability of choosing a high control mode increased from 8.67%, with all variables on average, to 16.93% when Brazilian franchisors operated in countries closer to home (1 standard deviation less than the average distance and all other variables remain constant). For example, they are more likely to select a low control mode in Guatemala and Angola than Uruguay and Argentina.

Hypotheses related to the resource-based and organizational capabilities theories
The significant (p < 0.1) and positive relationship between the companies' financial capacity and the likelihood of choosing a high control mode provide support for Hypothesis 3. Given their revenues in the domestic market, franchisors have more financial capacity to fund their international expansion, increasing the probability of a higher control mode. Thus, our results support the literature claiming that franchisors with limited financial resources to internationalize their operations will favor lower control modes and count on their local partners' financial assets (Combs et al., 2004;Windsperger, 2014, 2017).
Regarding Hypothesis 4, the results for brand name and business know-how were not statistically significant, although their coefficients were positive, as expected. Thus, these results do not support the hypothesis that system-specific assets, specifically, brand name and business know-how, increase the probability of choosing a high control mode in foreign markets.
The results of the franchisors' size as a control variable revealed a significant (p < 0.1) and positive relationship with the likelihood of a high control mode, meaning that franchisors with more units in the domestic market favor high control modes abroad. The firm's age had a negative and significant (p < 0.01) relationship with the likelihood of choosing a high control mode. In other words, franchisors with more years in business are more likely to adopt low control modes with local partners. Previous studies found mixed results regarding the effect of the franchisors' experience on the modes of entry and governance. Some reported that franchisors with more experience should have the resources to choose high control modes and avoid intermediary-related costs Kundu, 1998a, 1998b). Others did not find significant relationships (Erramilli et al., 2002). Our findings are aligned with Burton et al. (2000), who reported that more experienced franchisors employed intermediaries in their governance modes, and with Fladmoe-Lindquist and Jacque (1995), who documented a relationship between more experienced franchisors and the likelihood of using franchising contracts. Franchisors' experience and improved ability to select and monitor competent intermediaries despite the agency costs might explain the results. Note: The dependent variable high control mode equals 1 if firm i chooses to enter the host country market with a high control mode, and zero otherwise. All of the non-binary independent variables are standardized by subtracting the mean and dividing by the standard deviation. Heteroskedasticity-robust standard errors (clustered at the firm level and operationalized by the country level) are shown in parentheses. Coefficients are statistically significant at the *p < 0.1; **p < 0.05; ***p < 0.01 levels.

Robustness tests
To conduct robustness tests of Model 1, we used other variables related to the host country's environmental uncertainty as an independent variable. The results for Models 3 and 4 appear in Table 4. In the first test, we changed the composite measure of the business environment by using the 2019 OECD Country Risk Classification, which measures the country's credit risk and the likelihood that a country will service its external debt (Model 3). In the second test in Model 4, we changed the composite measure of the business environment by using the GDP per capita based on purchasing power parity (constant in 2011 international dollars). Several papers have used country risk and economic development (GDP) as proxies for environmental uncertainty Brouthers and Hennart, 2007;Kundu, 1998a, 1998b;Morschett et al., 2010). Given that both of these factors are associated with measures of the business environment, we substituted them in Model 1 to check if the effects remained the same.
Results for Model 3 showed that country risk had a significant (p < 0.01) and negative relationship with the likelihood of choosing a high control mode. The riskier the host country, the greater the probability of Brazilian franchisors choosing low control modes. Furthermore, the results for Model 4 indicated that the lower the GDP per capita of the host countries, the greater the probability of low control modes. These results are in line with our previous results.
Our findings support the behavioral uncertainty and financial capacity hypotheses in all models. Results for business know-how remained not statistically significant. However, in Model 3, the results for brand name were positive and statistically significant, consistent with Hypothesis 4.

Extensions
In this section, we extend the analysis based on our earlier findings. Previous results demonstrated that the environmental uncertainty of host countries has an impact on the Brazilian franchisors' choice of international governance mode but the distance between the home and host countries does not influence their decision. Thus, considering Model 1 as our basic model, we modified it to include the interactive effects between environmental uncertainty and the impact of brand name (Model 1.1), business know-how (Model 1.2), financial capacity (Model 1.3), and behavioral uncertainty (Model 1.4) on the choice of governance mode.
Jell-Ojobor and Windsperger (2014Windsperger ( , 2017 suggested negative interactive effects between environmental uncertainty and the impact of financial assets and behavioral uncertainty on the franchisor's tendency to use high control modes. Furthermore, our previous results showed that the impact of system-specific assets such as brand name and business know-how have no significant impact on the choice of international governance modes of Brazilian franchisors. As some scholars argue, the impact of a firm's specific resources on this choice is not uniform, and the firm's advantages may vary depending on the country. Institutional dimensions might Note: The dependent variable high control mode equals 1 if firm i chooses to enter the host country market with a high control mode, and zero otherwise. All of the non-binary independent variables are standardized by subtracting the mean and dividing by the standard deviation. Heteroskedasticity-robust standard errors (clustered at the firm level and operationalized by the country level) are shown in parentheses. Coefficients are statistically significant at the *p < 0.1; **p < 0.05; ***p < 0.01 levels.
also influence the ability of companies to exploit specific resource-based advantages (Brouthers et al., 2008;Hennart and Slangen, 2015). For this reason, we also tested the interactive effects between environmental uncertainty and the impact of brand name and business know-how on the franchisor's tendency to use high control modes. The results presented in Table 5 revealed two significant interactive effects.
Interactive effects between environmental uncertainty and business know-how. The environmental uncertainty of host countries had a significant (p < 0.1) interactive effect on the relationship between the business know-how of Brazilian franchisors and the choice of the governance mode abroad (Model 1.2). While the franchisors' business know-how by itself did not influence the choice of international governance mode (Model 1 in Table 3), it became significant in conjunction with environmental uncertainty. Thus, more uncertain environments increased the impact of business know-how on the franchisor's probability of choosing higher control modes.
Interactive effects between environmental and behavioral uncertainty. The results for the interaction between environmental uncertainty and behavioral uncertainty contradicted the expected result (Model 1.4). While as the literature suggested, the isolated effects of greater environmental uncertainty and greater behavioral uncertainty had a negative impact on the probability of high control modes (Combs et al., 2004;Fladmoe-Lindquist and Jacque, 1995;Grewal et al., 2011;Jell-Ojobor and Windsperger, 2014), the interactive effect of environmental uncertainty mitigated the negative impact of behavioral uncertainty on the probability of choosing high control modes. A possible explanation for this result is related to the high costs that result from the interaction between behavioral uncertainty and environmental uncertainty. More uncertain environments might increase behavioral uncertainty to the point that contractual agreements such as franchising are not able to mitigate the costs. The interaction may increase franchisee-related agency costs as well, making high control modes or dual distribution preferable (Erramilli et al., 2002;Lévesque et al., 2018;Michael, 2000).

Discussion
The empirical analysis reveals three significant findings. First, the environmental distance between the home and host countries does not impact the Brazilian franchisors' choice of international governance modes. The environmental uncertainty of the host countries is a factor that the Brazilian franchisors in this sample consider for their choice of the governance mode, despite the home country's conditions and its relative environmental distance to the host countries.
The greater the similarity between Brazil and the environments of other emerging and developing economies, which was supposed to be an advantage for Brazilian franchisors to deal with the transaction costs, does not increase the likelihood of high control modes. Note: The dependent variable high control mode equals 1 if firm i chooses to enter the host country market with a high control mode, and zero otherwise. All of the non-binary independent variables are standardized by subtracting the mean and dividing by the standard deviation. Heteroskedasticity-robust standard errors (clustered at the firm level and operationalized by the country level) are shown in parentheses. Coefficients are statistically significant at the *p < 0.1; **p < 0.05; ***p < 0.01 levels.
Instead, these franchisors are more likely to choose low control modes when the host countries are emerging or developing countries. These findings contradict the idea that being used to high levels of uncertainty affects the behavior of firms headquartered in developing and emerging countries towards these host countries (Cuervo-Cazurra, 2012;Cuervo-Cazurra and Genc, 2008). On the contrary, Brazilian franchisors consider the host countries' characteristics, such as better environmental conditions (Lanfranchi et al., 2020;Melo et al., 2015), when making these decisions, as one can see by the high number of Brazilian franchisors operating in the United States (Aguiar et al., 2017;Cantoni et al., 2019). Second, three main factors influence Brazilian franchisors' decisions regarding international governance modes-the environmental uncertainty of the host countries and their behavioral uncertainty, derived from the transaction cost and agency theories, and the franchisors' financial assets, derived from the resource-based and organizational capability theories. Greater environmental and behavioral uncertainties increase the tendency of franchisors to choose low control modes abroad. In host countries with greater environmental uncertainty, franchisors will be more inclined to allocate decision rights to local partners.
The same reasoning applies to distant countries with greater behavioral uncertainty. By choosing lower control modes, franchisors shift the costs of moral hazards and poor choices to the local partner. These results support de Almeida et al. (2018), that identified a preference of three Brazilian franchisors for master franchising agreements in emerging host markets: Igui, a swimming pool manufacturer operating in Mexico; Chilli Beans, that produces sunglasses and accessories and have a master franchisee in Colombia; and Young Free (fictitious name), a textile company operating in the United Arab Emirates. Brazilian firms prefer master franchising arrangements in risky locations.
On the other hand, better financial capability to fund the expansion increases the likelihood of franchisors choosing high control modes. These findings are aligned with previous empirical research that analyzed the choice of governance mode in other contexts (Castrogiovanni et al., 2006;Fladmoe-Lindquist and Jacque, 1995;Hussain et al., 2018;Jell-Ojobor and Windsperger, 2017).
Third, the environmental uncertainty of the host countries has a significant interactive effect on the Brazilian franchisors' business know-how and influences their choice of international governance mode. When the environmental uncertainty of host markets is greater, Brazilian franchisors' business know-how becomes more relevant in deciding to choose a high control governance mode. Thus, the ability to exploit their specific resource-based advantages increases in more uncertain environments (Brouthers et al., 2008;Hennart and Slangen, 2015). Additionally, environmental uncertainty has a significant interactive effect with behavioral uncertainty, mitigating the negative impact of the latter on the probability of choosing high control modes. In other words, the interaction may intensify the franchisee-related agency costs, increasing the likelihood of high control modes.

Conclusions
In this article, we tested the multi-theoretical model of the choice of international franchising governance mode in the Brazilian context. We discussed to what extent it can explain the decisions of firms headquartered in an emerging market. The findings from the empirical analysis revealed that the similarity between the home and host countries' environment does not influence the decisions of Brazilian franchisors. Moreover, the environmental uncertainty of host markets and their behavioral uncertainty have a negative impact on their tendency to choose high control modes in international markets. In contrast, their financial assets have a positive effect on this tendency. Thus, from the analysis of the Brazilian franchisors, it is possible to conclude that the traditional international franchising literature can explain the choices of international governance modes of franchisors from an emerging market. We also observed that more uncertain environments increase the impact of business know-how and mitigate the negative impact of behavioral uncertainty on the probability of choosing high control modes.
Our results contribute to the international franchising literature by verifying if the integrative model of the choice of international franchising governance mode needs to be extended to the emerging market context (Cuervo-Cazurra, 2012;Ramamurti, 2012). We found that, besides explaining the decisions of developed countries franchisors (Jell-Ojobor and Windsperger, 2017), the theoretical model also has explanatory power to understand the choices of franchisors from an emerging market.
It also contributes by providing a statistical test to complement the theoretical evidence, case analysis, and partial results provided by previous research Windsperger, 2014, 2017). Moreover, it follows the insights of Dant and Grünhagen (2014), Jell-Ojobor and Windsperger (2017), and Combs et al. (2004) by investigating the choice of governance mode in different markets, in this case, from the Brazilian franchisors' standpoint. This study also has practical implications in that it provides franchisors seeking international expansion with the variables they should consider when choosing an international governance mode.

Limitations and future developments
One limitation of our study is the lack of data for other variables suggested by Jell-Ojobor and Windsperger (2014) in their multitheoretical model, namely, local market assets, transaction-specific investments, and intangible assets. Future research can improve the model by collecting data to test the effects of these three other variablesfor instance, information about international partners' host market knowledge, investments, and absorptive capacity. Besides, we acknowledge that geographic distance is not the only factor that affects behavioral uncertainty. Other aspects, such as business attitudes, work ethics, qualification level of local staff, psychic distance, and language, are relevant (Burton et al., 2000;Jell-Ojobor and Windsperger, 2017). Also, the time of experience abroad could be included as an additional control variable in future studies. Unfortunately, information about these factors was not available for all the 55 host countries and 108 franchisors in the sample, and access to data about franchisors' international partners is a challenge (Windsperger and Dant, 2006).
Another limitation is related to the impact of the home country's effects. We verified the effects of the similarities between the home and host countries by using the composite measure of the distance between the home country (Brazil) and the host countries, as well as the IMF's classification of countries as emerging and developing economies or advanced economies. In order to determine whether the home country has a direct effect on the choice of international governance mode, we need a sample of different home countries. We also recognize that the classifications of emerging and developing economies do not consider the similarities and differences between countries. Thus, we suggest future studies that investigate other home countries in emerging and developing economies.

Funding
This work was supported in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior -Brasil (CAPES) -Finance Code 001.