Franchising structure changes and shareholder value: Evidence from store buybacks and refranchising

Regarding the advantages of franchising, agency theory and TCA highlight efficiency-related benefits that accrue to franchisors from the unique skills, competencies, outlet specific know-how, and local market knowledge of the franchisees (e.g., Heide, 1994; Windsperger & Dant, 2006). Together, these benefits add to the internal knowledge base of franchisors (Sorenson & Sørenson, 2001) and can help them to innovate (Bradach, 1997). Additionally, franchising provides strategic options to managers, where they can terminate existing partnerships and align with new franchisees based on changing market conditions (Balakrishnan & Wernerfelt, 1986)—often at lower transactional costs than if they attempted to manage these changes within the boundaries of their firms. Such strategic options can be a source of competitive advantage, as they allow franchising firms to better manage distributional challenges and nimbly respond to market needs (Palmatier et al., 2020). Together, these observations suggest that refranchising can enhance future cash flows of franchisors, adding to their shareholder value.

Despite these benefits, firms can also face significant headwinds with greater reliance on franchisees. Specifically, agency theory alerts that franchisors face the risks of franchisees unfairly exploiting their assets such as brands and business format expertise due to information asymmetries and incentive misalignment (Bergen et al., 1992; Combs et al., 2011). To protect against the possibility of unfair exploitation of franchisor market-based assets (and the resulting reduction of future cash flows) due to horizontal and/or vertical free riding by franchisees (Mathewson & Winter, 1985), franchisors need to incur the coordination costs of monitoring franchisees on an ongoing basis (e.g., Agrawal & Lal, 1995). Aligned with this, TCA indicates that to manage the behavioral uncertainties associated with franchisees, franchisors would need to undertake coordination and communication efforts that can increase their transaction costs (e.g., Rindfleisch & Heide, 1997), reducing their future cash flows and lowering shareholder value.

One solution to these problems can be to realize the coordination benefits of fiat from an increased reliance on hierarchical governance (Williamson, 1985) through buybacks of downstream retail units. Greater reliance on company-operated stores allows franchisors to increase bargaining power to manage franchisees more cost effectively (Bradach, 1997; Michael, 2000). It also enables franchisors to have direct interaction with customers at a larger number of locations. This should lead to superior customer knowledge for firms, allowing them to offer better customer experiences to positively affect financial performance. However, it is also worth noting that store buybacks can be expensive, placing downward pressures on future cash flows of firms. Additionally, agency theory argues that franchisors can have agency issues with internal agents, i.e., managers of company owned stores, with these agents not performing their roles adequately (e.g., Brickley & Dark, 1987; Norton, 1988; Rubin, 1978). This can necessitate firms to place efforts in monitoring employees and providing higher performance-based incentives to them, which can dampen their prospective cash flows.

In summary, there are arguments both in favor and against whether firms should have higher or lower levels of franchising in the retail chain. We contend that, ultimately, the right level of franchising for firms would be where distribution channel governance structures are appropriately aligned with agency issues and transactional costs dimensions. The “Darwinian economics” rationale advanced by Anderson (1988) holds that competitive market forces compel firms to select strategies that approximate optimal behavior. Indeed, previous research provides evidence that financial markets are supportive of the Darwinian rationale by showing that investors reward firms for both brand acquisitions and brand disposals based on the context (Wiles et al., 2012). With respect to retail channels, firms would similarly benefit from adjusting their franchising levels at discrete intervals through refranchising or buybacks based on their unique conditions. Since investors understand that distribution channels are market-based assets with financial value (Srivastava et al., 1998) and that firms adjust channel structures infrequently, they would likely reward firms for both these franchising level changes. Together, we posit:

Following previous research, we contend that factors across firm and industry levels would present boundary conditions (e.g., Geyskens et al., 2002) for the impact of refranchising and buybacks on stock returns. With respect to firm-level factors, we first derive from agency theory to focus on royalty rate and advertising intensity of franchising firms as firm level moderators. Next, using the TCA lens, we outline the role of ROA and trade credit provided in shaping the stock returns to franchising structure change announcements made by firms.

Our main unit of analysis is an announcement by a publicly traded firm owning franchising chain(s) to (a) refranchise previously company-owned and operated retail units or (b) buy back retail units from existing franchisees to operate them as company-owned units. To test our hypotheses, we use archival methods and bring together information from multiple data sources. First, we use the Bond’s Franchise Guide, Entrepreneur Magazine’s Franchise 500 Ranking, and FRANdata website to generate a list of 350 business-format franchising chains, which are either publicly-traded or owned by the publicly-traded firms. All three sources—Bond’s Franchise Guide (e.g., Jindal, 2011; Lafontaine & Shaw, 2005), Entrepreneur’s Franchise 500 Rankings (e.g., Lafontaine & Shaw, 2005; Shane et al., 2006), and FRANdata (e.g., Hsu et al., 2017)—are well-established sources extensively used in research and provide reliable annual information on U.S. franchisors. Next, we check for the completeness of franchising structure data as well as financials and ownership data for each firm in the sample because franchising chains are occasionally bought and sold by their parent firms. We ascertain the availability financial data needed for our main variables (as detailed below) in the COMPUSTAT database. If any of the required data was not available, we removed those firms/chains from the list. This step resulted in 102 franchising firms with the franchising structure and financial data in 2001–2020 needed for our analysis.

Second, to compile a sample of refranchising/buyback announcements for the franchising firms specified above, we perform a broad keyword search using FACTIVA and Nexis Uni, newswire services, annual reports, and corporate websites, for each year from 2001 to 2020. Corporate news often reaches markets via multiple channels, which warrants an examination of a variety of sources to ensure comprehensiveness in data collection and accuracy in detecting the dates of the first information release (Fotheringham & Wiles, 2022). The keywords utilized in our search were ‘refranchising,’ ‘buyback,’ ‘repurchase,’ ‘buyout,’ ‘chain growth,’ ‘conversion,’ ‘contract renewal,’ ‘contract termination,’ ‘ownership redirection,’ ‘vertical integration,’ ‘proportion of franchised outlets.’ Where there were ambiguities about the precise announcement date, we remove those announcements from the dataset. This step results in total 343 events (123 buybacks and 220 refranchising announcements).

Next, to minimize “noise” from potential confounding effects, we check for any contemporaneous announcements. Specifically, based on accepted practices in marketing research (e.g., Wiles et al., 2012), we control for financial (earning announcements, stock splits, stock buybacks) and strategy-related announcements (mergers and acquisitions, partnerships, joint ventures, lawsuits, executive management changes, and new product launches). When such announcements occurred within a two-trading-day window around the focal announcements, those contaminated events were removed from the sample. We also check if any of the collected announcements were a part of previously announced program to restructure firms’ franchising systems. During data collection, we collect details of why firms decided to pursue refranchising and buybacks and how they would be implemented (if such information is provided in the announcement). None of the collected announcements were identified as parts of previously announced restructuring programs. Further, our search provides confidence that none of the events were leaked to the public and investors before the event dates utilized in our study.

After accounting for confounding events, the final sample in our analysis includes 205 announcements (125 refranchising and 80 buybacks) made by 41 firms for 45 chains (see Figure WA.1 in the Web Appendix A). The sample is similar in size and composition to the samples used in other studies utilized publicly listed firms in the franchising context (i.e., Combs & Ketchen, 1999; Madanoglu et al., 2011; Hsu et al., 2017; Srinivasan, 2006). Web Appendix B (Tables WB 1 & WB 2) provides a yearly breakdown of firms announcing refranchising or buying back business units during the period of observations, along with the number of announcements made (refranchising vs. buybacks) by the firms every year.

Firm Royalty Rate (Royalty_i): Reflects the ongoing royalty paid by franchisees to the franchisor and is measured as a percentage of franchisee sales (Michael, 2002).

Firm Advertising Intensity (AdvIntens _i): Reflects a firm’s advertising focus and operationalized as ratio of advertising expenses to firm sales (Combs & Ketchen, 2003).

Firm Return on Assets (ROA_i): Reflects firm’s asset utilization efficiency and operationalized as earnings before extraordinary items divided by total assets (Homburg et al., 2014).

Firm Trade Credit Provided (TradeCredit_i): Reflects firm’s investments in downstream channel partners; operationalized as ratio of trade receivables to firm sales (Astvansh & Jindal, 2022).

Industry Dynamism (IndDynamism_j): Reflects demand variation in an industry. We measure it as the standard error of the regression slope coefficient of the trend in industry sales divided by average sales in the industry over the past 5-years, with industry at the 4-digit SIC level (Dess & Beard, 1984).

Industry Munificence (IndMunificence_j): Captures growth of demand in an industry. We operationalize it as the regression slope coefficient of the trend in industry sales by the average sales in the industry over the past 5-years, with industry at the 4-digit SIC level (Dess & Beard, 1984).

We estimate stock market reaction to refranchising or buyback announcements by calculating the cumulative average abnormal returns (CAAR) for our events of interest (Web Appendix C), Next, we implement cross-sectional analyses in two steps. First, it is possible that information not observable by investors drives the decision of a firm to refranchise or buyback (Kai & Prabhala, 2007). This may lead to selection bias in our sample as we only include those firms engaged in restructuring their franchising systems in the analysis. To safeguard against possible selection bias due to any potential systematic differences between the firms that franchise and undertake refranchising/buybacks decisions versus those that do not, we estimate the probability of a firm decision to refranchise/buyback retail units and calculate the inverse Mills ratio (IMR) (Web Appendix D). Next, we include the IMR as an additional firm-level control variable in the regression model shown below (Eq. 1) to evaluate our hypotheses for firm i in industry j. All the variables are as described earlier and in Table 2.

Although the event study methodology suffers from limited endogeneity concerns, we take some additional steps to ensure that the potential of endogeneity is further attenuated. Specifically, it is possible that shareholder returns to firm’s franchising structure decisions are endogenous with other firm characteristics, i.e., firm royalty rate, advertising intensity, ROA, and trade credit provided. Endogenous variables may be correlated with the error term, which violates the OLS assumptions, resulting in regression estimates that are unreliable (Wooldridge, 2002). To address the potential endogeneity issues, we apply the two-stage least squares (2SLS) methodology in estimating Eq. 1. The instrumental variables in 2SLS should be highly correlated with the endogenous variables (meet the relevance criteria) but have no direct effect on the dependent variable (the exclusion criteria). We follow the established practice in marketing (e.g., Germann et al., 2015) to instrument royalty rate, advertising intensity, ROA, and trade credit provided with the industry averages of these measures in the given, year excluding the focal firm, with industry defined at 4-digit SIC level. The proposed instruments are deemed appropriate as they meet the relevance criterion. This is because the focal firms face similar market conditions as their industry peers and it is reasonable to assume that their individual characteristics are correlated with the industry averages. Next, the instruments meet the exclusion criterion because the industry averages are unlikely to systematically impact individual firm’s financial performance and more specifically stock abnormal returns to refranchising/buyback announcements. We add industry average number of employees (minus the firm) and geographic dispersion (i.e., number of states in which the firm operates) as additional instruments, to meet the overidentification restriction criteria (Wooldridge, 2002). Hausman’s tests of endogeneity confirm that the 2SLS estimation approach is more appropriate and performs better than traditional OLS estimation. The postestimation analyses, specifically the first-stage regression results and the overidentification restrictions tests, confirm that the instruments are valid and the models perform consistently (Web Appendix E).

In addition, we take further steps to lower endogeneity concerns by following recommendations of Cameron and Miller (2015) for generating cluster-robust inferences for data sets with few clusters by clustering at the firm level. We also model year fixed effects. The decision to cluster at the firm level and include year fixed effects in the estimations is driven by following considerations. First, it can be argued that within any given year, clustering is due to shocks that are the same across all the observations in the year and can be effectively addressed by controlling for year fixed effects in the estimations. Second, clustering at the firm level produces a sufficient number of clusters, allowing for reliable cluster-robust inferences that account for cross-correlation and dependence across multiple observations for the same firm. We also control for the “eating places” industry through an indicator variable and include multiple industry factors, as main predictors (i.e., industry dynamism and munificence) and control (HHI). Finally, following previous research investigating two strategically opposite actions taken by firms (Wiles et al., 2012), we estimate Eq. (1) separately for refranchising and buybacks.

We estimate the main models using the approach detailed above. To identify the most appropriate event window for cross-sectional analyses, we follow the common practice (e.g., Homburg et al., 2014; Geyskens et al., 2002; Sorescu et al., 2017) of using the event window that most completely captures the cumulative average abnormal returns (CAARs) (i.e., the event window that consistently reports the most significant t-test and z-test statistics) across various tests (i.e., parametric and non-parametric tests). The event window (0; + 1) demonstrated the most stable and consistently significant results across different model specifications, i.e., market-adjusted model vs. market model vs. Fama–French model vs. Fama–French-Carhart four-factor model, and across different periods of estimation. Therefore, we use CAARs over the event window (0; + 1), assessed with the market-adjusted model with the equally weighted index, as a dependent variable in the cross-sectional analyses. Our reliance on the market-adjusted model with equally weighted index to test the cross-sectional hypotheses is driven by prescriptions provided by marketing scholars for conducting event studies (Sorescu et al., 2017). To examine the specific drivers that shape the stock market response to the refranchising vs. buyback strategies, we estimate the model outlined in Eq. 1 separately for these events. We also check and observe that multicollinearity is not a major concern affecting our results. All the variance inflation factors are below 10 (Meyers et al., 2006), with average VIF_avg = 2.48 for the refranchising model and VIF_avg = 1.6 for the buybacks model.

Results show that the refranchising model, with the main predictors and control variables as outlined in Table 2, explains 40.7% of variance in CAARs in stock markets following the announcements and is significant at p < 0.001 level (see Table 8, Column (a)).

With respect to the individual predictors, we find that a firm’s royalty rate has a negative and marginally significant impact on firm’s abnormal stock returns caused by refranchising announcements, (β = -0.04; p-val < 0.1), providing partial support for H2a. However, we do not observe a significant effect of advertising intensity on stock market returns to refranchising announcements (β = 0.10; p-val < 0.50). As such, we do not observe support for H3a. In support of H4a, ROA has a significant negative effect (β = -0.17; p-val < 0.01) on stock returns following refranchising events. Finally, we observe that firm trade credit provided has a positive and significant effect (β = 0.07; p-val < 0.01) in shaping shareholder returns from refranchising decisions, supporting H5a.

At the industry level, industry munificence is seen to have a negative and significant impact (β = -0.35; p-val < 0.01), while industry dynamism has no effect on CAARs (β = 0.63; p-val < 0.17). As such, we find support for H7a, but not for H6a. Finally, with respect to controls, we find cash flows of firms to significantly increase the stock returns generated by refranchising. This may be because higher cash flows allow firms to attenuate some of negative agency and TCA related costs involved with franchising. We do not find firm size or industry concentration to have a significant effect, reflecting generalizability of our findings is not limited by the firm size or level of industry competition.

Next, we estimate the cross-sectional model for the buyback subsample (see Table 8, Column (b)). The results show that the model is significant at p < 0.001 level and explains 34.1% of variance in CAARs from the buyback announcements. However, the factors that drive abnormal returns in the buyback model differ strikingly from those in the refranchising model. More specifically, at the firm level, we only observe royalty rate to have a positive and significant effect (β = 0.02; p-val < 0.01) on abnormal stock returns from buyback announcements, confirming H2b. All other hypothesized firm-level factors are observed to have no significant effect, providing no support for H3b, H4b, and H5b. With respect to industry level factors, industry munificence has a positive and marginally significant effect on firm abnormal stock returns from buyback announcements (β = 0.24; p-val < 0.10), thus providing partial support for H7b. Yet, industry dynamism has no significant effect, rejecting H6b. Finally, with respect to controls, we do not find firm size and cash flows to impact firm stock returns generated from buybacks. This indicates that extra resources available to bigger firms and those with higher cash flows are not sufficient to attenuate some of the negative effects of buying back previously franchised units and bringing them within the boundaries of the firm. We also do not find industry concentration to have an effect, reflecting generalizability of our findings is not contingent on the level of competition in the industry.

To increase confidence in the results, we implement several robustness checks. We re-estimate the short-term abnormal returns with alternative benchmarks—the market model with equally weighted index (Brown & Warner, 1985) and Fama–French model (Fama & French, 1993), with alternative estimation periods of 300 days ending 30 days before the announcement day and 260 days ending 10 days before the announcement day (see Fama–French Model results in Web Appendix G). The CAARS on the day of the announcement and over (0; + 1) event windows remain positive and statistically significant across the subsamples and in the combined dataset (including refranchising/buyback announcements), providing added support for H1a&b.

Next, we conducted the Chow test for equality between the coefficients in the refranchising vs. buyback models. The Chow test allows us to examine whether the parameters for the refranchising subsample are equal to those for the buybacks subsample (Chow, 1960). The null hypothesis for the test is that there is no break point and the pooled data (refranchising and buybacks) can be represented with a single regression line. The results rejected the null hypothesis (F (7, 36) = 2.80, p < 0.01), suggesting that the two groups (refranchising vs. buybacks) have different slopes and intercepts and cannot be pooled together. As such, the factors explaining stock market abnormal returns from refranchising announcements are different from those explaining from buyback announcements. We also re-estimate the cross-sectional models with the subsamples winsorized and trimmed at 1% levels and observe largely consistent results (see Web Appendix H). Next, as an additional robustness check, we drop the IMR variable and re-estimate the main models. All the hypothesized relationships hold in terms of the direction and significance (Web Appendix I). Together, these checks increase confidence in our overall findings.

Finally, we check if firms experience any long-term returns to refranchising/buyback announcements, using the long-horizon event study methodology (Kothari & Warner, 2007). For long-term returns, extant literature suggests two alternative approaches, the buy-and-hold abnormal returns approach and the calendar-time portfolio returns with the Fama–French benchmark and a 1-year horizon (Kolari & Pynnönen, 2010). Each of the methods has strengths and limitations (Kothari & Warner, 2007; Srinivasan & Hanssens, 2009). To ensure the robustness of the results, we utilize both approaches. None of the specifications yield statistically significant results, suggesting that firms do not accrue any long-term returns from refranchising and buybacks. These results support the argument that corporate news like refranchising and buybacks provide strong economic information to financial markets about firm strategies. Since such events are costly and less reversible, they provide credible signals about a firm’s commitment towards a strategic direction, effectively reducing the level of uncertainty investors might have about the firm and enabling accurate assessment of its future growth prospects. This assessment gets incorporated into firm stock market price in the short-term and does not require a long-term assessment in the financial markets. It, therefore, appears that in the context of refranchising/buyback announcements, stock markets remain efficient, rendering the short-term event study methodology appropriate for our analysis.

Our study makes multiple contributions to marketing theory. Although extant research offers rich insights into channel structures (e.g., Lafontaine & Kaufmann, 1994), channel governance (e.g., Bergen et al., 1992), channel additions (Geyskens et al., 2002; Homburg et al., 2014), channel deletions (e.g., Kumar, 2021), and channel management (e.g., Palmatier et al., 2020), evidence for financial implications of distribution channel related strategic decisions is relatively limited (e.g., Gielens & Geyskens, 2012). Particular to franchising, some studies have compared the differences in financial performance between franchising and non-franchising firms (e.g., Madanoglu et al., 2011), and others have considered the financial effects of dual distribution structures, i.e., having a mix of franchise-owned and company-owned retail units (e.g., Srinivasan, 2006). Yet, the impact of changes in levels of franchising (and the direction of these changes), while keeping the size of the distribution chain constant, on shareholder value of firms remains to be examined in detail. Indeed, to the best of our knowledge, only one study has considered the impact of changes in franchising proportion on stock market returns (Hsu et al., 2017), and we complement and go beyond their findings in substantive and meaningful ways. We discuss our contributions in more detail next.

First, our reliance on the efficient contracting perspective (i.e., agency theory and transaction cost analysis (TCA)), along with the use of the event study methodology enabled us to articulate and evaluate the causal effects of changes in franchising structure (in terms of refranchising and buybacks) on shareholder value of franchising firms.³ In particular, we theoretically illustrate the downsides and benefits of refranchising and buybacks, two seemingly opposite strategies, for franchising firms. Our analysis confirms that financial markets reward firms for both refranchising and buybacks decisions. There has been a considerable debate regarding the optimal proportion of franchise ownership for firms, with some studies asserting that firms are likely to favor franchising initially but then move towards company ownership of retail units, whereas others arguing the opposite (Dant & Kaufmann, 2003). In squarely responding to this debate, we underscore that, from a shareholder value perspective, neither of the two opposing arguments stand rejected. Instead, financial markets align with the “Darwinian economics” rationale (Anderson, 1988), supporting the need for firms to react to competitive market forces in a manner that best leverages their situation. Thus, firms may select opposing strategies (in terms of hierarchical or market-based governance) to optimize their performance.

Second, by investigating the impact of opposing strategies in the context of franchising, we add to the limited work in marketing strategy validating the shareholder value of apparently conflicting strategic moves by firms based on contingent factors (e.g., Wiles et al., 2012). It also allows us to go beyond Hsu et al. (2017), who examined annual changes rather than discrete and specific events involving changes in franchise ownership structure and did not offer visibility into the separate effects of refranchising and buybacks on firm stock returns. Furthermore, our framework and empirical methodology provides a template for understanding other firm decisions where agency theory and TCA prescriptions are at play in the determination of the organization of economic activity within or outside the boundaries of the firm. For instance, decisions involving in-house sourcing vs. outsourcing (for a range of business functions such as production, advertising, marketing research, salesforce etc.), licensing vs. owning facilities in international markets, and offshoring vs. in-shoring are few examples of opposing strategies that have both benefits and downsides for firms. Our approach of utilizing an efficient contracting perspective (e.g., Bergen et al., 1992; Rindfleisch & Heide, 1997), in combination with the event study methodology, can be extended to evaluate the shareholder value implications of these strategic choices made by firms.

Third, we build on the assumptions of agency theory and TCA to identify and articulate boundary conditions imposed by firm and industry factors on the shareholder value impact of franchising decisions (refranchising and buybacks). Both theories alert to the potential of moral hazard and opportunism in the franchisee-franchisor relationship and the costs involved in mitigating these exchange hazards. They also highlight the impact of environmental uncertainty on the hierarchical and vertical governance structures utilized by firms. We draw from these views to present a nuanced picture of the forces shaping the boundary conditions in our study.

Specifically, based on agency theory we find that a firm’s royalty rate has an attenuating/enhancing effect on the incremental shareholder value derived by firms from refranchising/buybacks. This supports the argument that as firms with higher royalty rates face greater risks of free-riding and moral hazard by franchisees, the gains accruing to them from refranchising get attenuated and firms are better placed to buy back some of their existing franchised units. Along similar lines, we had built on agency theory to argue franchising firms with high advertising intensity would also derive lower/higher stock returns from refranchising/buybacks. Our predictions were driven by observations that higher advertising intensity reflects advertising as a central element of a firm’s marketing strategy and the firm’s investment in building customer-based resources. In such, cases, heightened agency issues associated with franchisees would reduce the gains accruing to the firm. However, our results did not provide confirmation for these hypotheses. It is possible that the local market knowledge of franchisees allows firms to better target their advertising efforts (an aspect not observable in our data), generating consumer surplus which may be overcoming the agency costs associated with working with them. Similarly, we had posited that firms would be able to appropriate the higher value generated by advertising for themselves in the case of buybacks, instead of having to share this with franchisees. As such, firms with higher advertising intensity were predicted to benefit more from buybacks. However, results did not confirm this prediction as well, reflecting that the consumer surplus generated from advertising, without the benefits of local know-how afforded by franchisees, may not be sufficient to cover the acquisition costs and other agency issues associated with hierarchical governance.

A similarly complex set of findings emerge for the two firm-level moderating factors motivated by a TCA lens—ROA and trade credit provided. Consistent with the prediction, results support the argument that firms with high ROA would benefit less from refranchising, given that high ROA reflects lower need for firms to rely on outside partners to alleviate their resource scarcity. Further, as predicted firms that provide higher trade credit to their channel partners, reflecting investments in stronger channel partnerships, appear to derive greater shareholder benefits from refranchising. For buybacks, we do not find support for these factors, possibly due to positive and negative forces balancing each other out.

Finally, our study draws attention towards the boundary conditions created by the industry environment faced by firms. We find that industry munificence weakens the beneficial effect of refranchising on shareholder value, while marginally elevating the stock returns derived from buybacks. This supports our position that when the industry environment offers more growth opportunities, it may be possible for firms to do well operationally without reliance on partners (franchisees). Further, with buybacks, firms will be able to capture more of the overall industry growth for their own cash flow gains, enhancing the shareholder value effects. In dynamic industries, agency theory and TCA led us to argue that greater reliance on franchisors would be beneficial for firms. However, our analysis didn’t confirm these hypotheses, indicating industry dynamism as not a value driver for firms making refranchising and buybacks moves.

There have been numerous calls for researchers to show how marketing strategy decisions contribute to shareholder value (e.g., Srivastava et al., 1998). In adding to the marketing-finance interface literature that has emerged in response to these calls, we assess the impact of franchising decisions on firm stock returns. As we outlined previously, franchising is an important form of distribution strategy, which contributes substantially to the economy both in terms of economic output and employment. Furthermore, by focusing on franchising as a specific context, we are able to provide more customized insights to managers in franchising firms, compared to what generalized studies would be able to offer (Stremersch et al., 2022). Specifically, our research answers several key questions faced by managers of franchising firms considering structural changes to their distribution channels, while keeping the size of their distribution chain size constant. Moreover, we highlight that different boundary conditions govern the firm stock returns derived from refranchising and buybacks, providing nuanced guidance to franchisors.

Our research also suffers from some key limitations that suggest useful areas for further inquiry. In deriving our framework, we outlined key boundary conditions but our inquiry was limited to factors observable through secondary/archival data. As such, we were able to only indirectly capture certain theoretically relevant moderators like management supervision, quality control, and brand reputation through ROA, trade credit provided, and royalty rate respectively. Future studies can directly measure these constructs through primary research, such as managerial surveys. In addition, product-demand and channel-demand growth (as studied by Geyskens et al., 2002) as well as order of entry of firms in the industry may influence investor reactions to refranchising announcements. However, we do not have product and channel-level demand data for our sample. Future work can investigate the role of these boundary conditions, if the required data becomes available. Additionally, our sample was restricted to relatively larger, public-traded firms. However, the franchising industry has a large number of smaller and privately held firms, not as well represented in some of the secondary data sources. We observe firm size to be a weak boundary condition in our analysis, indicating that our results are likely to be generalizable to smaller firms. Further research can apply our framework through primary data collection techniques to provide additional confidence in the usefulness of our findings for managers across a larger spectrum of franchising firms. Finally, we restricted our context to franchising, which offers certain benefits in terms of pointed and directly useful findings for franchising managers (Stremersch et al., 2022). We submit that our theoretically derived framework and event study methodology should be applicable to other contexts as well and scholars can build on our study to evaluate these contexts in future research to enhance the external generalizability of our findings.