Determining Franchise Performance: A Transaction Cost Approach

Determining Franchise Performance:

A Transaction Cost Approach

An event study evaluating the value of franchises through the lens of mergers

and acquisition

Master Thesis

MSc BA Strategic Innovation Management

University of Groningen, Faculty of Economics & Business

Duisenberg Building, Nettelbosje 2

9747 AE Groningen, The Netherlands

Supervisor: K. J. McCarthy

Second Supervisor: E. P. M. Croonen

Determining Franchise Performance: A

Transaction Cost Approach

An event study evaluating the value of franchises through the lens of mergers and acquisition

Total word count: 13240 Abstract:

Franchising is an intriguing organizational form that has garnered the attention of many scholars in the past. Most research at the system level has focused on expansion through governance decisions (the addition of a franchised or company-owned outlet) and chain performance. However, conventional expansion strategies such as mergers and acquisition have been mostly neglected. After decades of research the question remains whether franchised companies financially outperform non-franchised companies. Evidence pertaining towards that end is inconclusive and mixed at best. Using the transaction cost logic, this paper aims to determine the efficacy of using merger and acquisitions (M&A) as an alternative expansion strategy and determines difference in performance of franchised chains versus wholly owned chains. Specifically, this is accomplished by performing an event study of mergers and acquisitions in the food retail industry. The results suggest that regardless of the governance form of the acquirer, acquisitions of franchised chains financially outperform non-franchised chains. Acquisition performance is further increased if the acquirer is also a franchise.

Keywords: Franchising, Mergers and Acquisitions, Event Study, Plural Form, Transaction Costs

I. INTRODUCTION

Franchising remains one of the most prominent methods of strategic and organizational expansion. Its international prominence is encapsulated by the unique method by which organizational expansion is created. At its core, franchising is characterized by licensing. The method by which one party allows another access to sell branded products (Fulop and Forward, 1997). In its entirety, franchising is a lot more complex than a rudimentary licensing agreement. Reason being, that the relationship between a franchisor and a franchisee is defined by a certain degree of integration and coordination that surpass the characterization of a licensing deal (Williamson, 1975). Not only does the franchisor grant permission to sell its branded products and access to distribution channels, but also the ability to be part of a fully developed business system

(Kaufmann & Eroglu, 1999). This method is particularly attractive to the franchisor as it allows faster market penetration and geographic dispersion. The franchisee on the other hand enters into franchising agreements, as it poses a lower risk alternative to conventional entrepreneurship (Zeller et al., 1980).

Not only is franchising unique in its organizational approach but also unprecedented in its success. In 2015, revenues form franchised companies reached $889 billion, in the United States alone (International Franchising Association, 2015). Another testament to the ever increasing popularity of franchising is that the growth rate of total employees, economic output and established franchises has been consistently rising in the last 10 years (IHS report 2016).

3 growth and organizational expansion

(Shane, 1996). Firm growth rates are paramount in understanding the underlying factors of firm efficiency, survival, and the determinants of profitability. Conventionally, it was postulated that the ability of a firm to grow depends on the expansion of the focal firm’s assets and the aptitude of the firm’s managers to employ them (Penrose, 1959). Franchising, however, offers a diverging perspective as its growth is not characterize by the growth through assets, but rather through the increase in contractual partners (Norton, 1988). An additional strategy for further geographic dispersion, franchisors grant certain franchisees the rights to a master franchise. Master franchisees have been given a specific geographic location in which the have the right to take over all franchise activities (Hoffmann and Preble, 1991).

This poses an interesting strategic deviation from conventionally organized business. Geographic expansion in companies is achieved through strategic alliances, joint ventures M&As. While franchising offers an inventive strategic alternative to the capital intensive options of joint ventures or M&As, the question remains whether these expansion strategies offer a valuable alternative to franchising in franchised companies. That is to say, is the option of direct investment in other organizations a valuable alternative to building the own franchise system organically?

Another interesting question in the field of franchising is whether franchising has an effect of financial performance. Whereas, there is a variety of evidence underling the effectiveness of franchising for firm growth, the question of performance differences between franchise and

non-franchise companies remains inconclusive. The reason lies in the theoretical foundation of most franchise related research. The main theoretical lens through which franchises are investigated is agency theory. From an agency perspective, firms use franchising in order to reduce costs associated with identifying and motivating qualified managers (Lafontaine 1992). Using agency theory, however, financial performance is assumed as (1) its continuing existence is an a priori testament to its value (Rubin, 1978) and (2) if it were not performance enhancing it would not have continued to exist in the market (Fama & Jensen, 1983; Williamson, 1991). The second theoretical foundation of franchising is the resource scarcity theory. Resource scarcity theory argues that due to capital and resource constraints the franchisor is restricted in its growth, thus licensing out its business format to another party. Moreover, the franchisee chooses to buy the franchise rather than to establish an independent business due to the reduced risk involved (Kaufmann & Dant, 1996; Oxenfeldt & Kelly, 1969). However, the resource scarcity theory primarily focusses on firm growth and survival, neglecting the financial performance outcomes. Hence, as both theoretical foundations disregard financial performance, the overall evidence in the literature is limited.

However, the authors draw attention to the lack of generalizability of their results, as the findings are constrained to a single industry in one country.

Regarding the lack of conclusive and overarching evidence in examining the performance differences, Combs, Ketchen, and Hoover (2004) urged that “for researchers interested in strategic management, understanding the determinants of financial performance is central”.

With that sentiment in mind, the main goal of this paper is twofold. Firstly, to investigate the inherent value of franchises and to determine whether the advantages that lead to increased growth and higher survival lead to superior financial performance. To counteract the shortcomings of the dominant theories in the franchising literature, the transaction cost theory is applied. Secondly, given the success with which franchises expand geographically, the question remains whether they still rely on conventional methods of expansion, specifically M&As. To answer these two questions, an event study of mergers and acquisition is conducted. This has two reasons. One, though the lens of mergers and acquisitions, the inherent value of each acquisition target is evaluated by the market. Therefore, if in fact franchises perform better than independent business, that difference should be reflected in the wealth gains of the acquirer through the evaluation of the market. Two, an event study will reveal whether franchises in fact use M&As as an alternative to establishing new company outlets. Additionally, an event study will determine the success with which franchises use conventional means of expansion in comparison to non-franchise companies.

To that end this paper aims to make several contributions. This study attempts to fill a gap in the transaction cost literature. Transaction costs economics (TCE) is a theory that is primarily concerned with the antecedents to the acquisition rather than acquisitions itself. TCE investigates the circumstance surrounding the decision to internalize (make) or acquire (buy) the specific strategic resource. Transaction cost economics is mostly concerned with the make-or-buy dichotomy, yet disregards the analysis of which target is most fitting after the ‘buy’ decision has been made. Hence, this study addresses the theoretical gap in the transaction cost theory in determining if a reduction in transaction costs through the acquisition of franchises increases acquisition performance. Additionally, this paper modestly tries to fill the empirical gap of distinguishing whether franchises perform differently than its independent counterparts. And Lastly, this study will determine the frequency and effectiveness with which franchises will expand through M&As.

II. THEORETICAL BACKGROUND Fundamental of Transaction Costs Economics (TCE)

5 purchasing strategy (Stump and Heide

1996), and distribution channel management (Anderson and Weitz 1992; Heide and John 1988), vertical integration decisions (Anderson 1985; John and Weitz 1988) and foreign market entry strategy (e.g., Anderson and Coughlan 1987; Klein, Frazier, and Roth 1990).

Coase argues that transaction costs are incurred when “… it is necessary to

discover who it is that one wishes to deal with, to inform people that one wishes to deal and on what terms, to conduct negotiations leading up to a bargain, to draw up the contract, to under- take the inspection needed to make sure that the terms of the contract are being observed, and so on” (Coase, 1937). Thus, transaction

costs are comprised of search and information costs, bargaining costs and policing or monitoring costs. In order for an economic exchange to take place, two parties need to invest time and resources and evaluate the necessary information in searching for the correct party. In case the search is successful, the two parties need to specify their arrangement in terms of a contract. Furthermore, each party has to inform each other of the exact exchange opportunities at hand, which again, will require resources. Agreeable terms for an economic exchange between two parties can usually only occur after an exhaustive and mutually costly bargaining period. The final category of transaction costs, are the costs incurred by ensuring that both parties hold up their end of the bargain and see to their agreed upon obligations (Dahlmann, 1979).

According to Williamson (1975), transaction costs are determined by the interplay between two key assumptions about human behavior (opportunism and bounded rationality) and two key

transaction-specific dimensions, namely asset specificity and uncertainty. The assumption that decision makers have limits to their rationality due to cognitive constraints is called bounded rationality. Bounded rationality infers that even though individuals intend to act rationally, their ability to do so is circumvented by limited information processing and communication abilities (Simon, 1957).

Thus, bounded rationality increases transaction costs as these cognitive constraints become problematic if the individual cannot evaluate behavioral or environmental uncertainties involved in or surrounding the exchange (Rindfleisch and Heide, 1997).

Opportunism is the behavioral assumption that decision makers tend to act according to their self-interest. Therefore, the assumption of opportunism renders it impossible to determine trustworthiness a

priori to a market exchange (Barney, 1990).

Ergo, transaction costs are increased with the possibility of opportunism, as the search for information costs are inflated, more resources have to be spent discovering an adequate exchange partner, bargaining costs rise as the one party may act in their own self interest and policing and monitoring costs increase as either party cannot be sure that the counterpart will meet the terms of the contract (Rindfleisch and Heide, 1997).

Transaction costs in Franchising

mentioned previously, the franchise is confronted with a strategic make-or-buy decision. That is, should a specific organizational unit or outlet be company-owned or franchised. The decision between the two alternative form is explained by the certain hazards that influence the amount of transaction costs. These factors are shirking and perquisite consumption by employees on the one hand and free-riding based on brand name value on the other (Brickley and Dark, 1987). Wherein, shirking is primarily associated with employees in the company-owned outlets and free-riding occurs in franchise arrangements. The susceptibility towards certain hazards is determined by the unique incentive structure of the respective governance modes.

Operators of company-owned stores are usually compensated with a fixed salary (Krueger 1990). The issue with fixed compensations, is that individual performance is not rewarded or punished by the parent company. Therefore, the managers of company owned outlets are susceptible to shirking as their compensation does not depend on performance, or to opportunism due to the awareness that superior performance is not rewarded (Dnes, 1996). Due to the increased risk of shirking in company-owned outlets the franchisor is forced to implement control mechanisms to monitor employee behavior (Ouchi 1979). These monitoring costs directly contribute to transaction costs, as an administrative apparatus has to be created, managed and implemented, requiring extensive resources particularly by the franchisor (Windsperger and Hendrikse, 2004).

Shirking hazards are unlikely to occur in franchised outlets, as the interests between the franchisor and the franchisee are

aligned. The main reason that shirking is improbable in franchising is that the low-powered incentives (through the fixed salary) in the company owned outlets are replaced by by high-powered incentives in franchised outlets. As franchisees are responsible for their own compensation there is an inherent incentive to increase perform. However, there is a trade-off between high- and low-powered incentives, namely the perverse effect of increased free-riding risk (Windsperger and Hendrikse, 2004). Free-riding is the result of the inherent conflict between the franchisor seeking control, quality standards and brand reputation and the franchisee pursuing entrepreneurial autonomy (Dant and Gundlach, 1999). Quality debasement and cutting costs are a common form of free-riding related opportunism as the entire franchise system bears the costs due to the detriment of reputation and brand value, instead of the individual franchisee (Kidwell et al., 2007). In sum, it can be concluded that the probability that a unit will be company-owned depends on the level of free-riding hazards and the intensity of shirking-related costs (Bercovitz, 2004).

7 asset specificity puts the franchisee in a

position in which the probability of opportunistic behavior is more pronounced. To counteract the dominant position of the franchisee, the franchisor requests an initial payment of up-front fees in addition to royalties throughout the length of the contract. By requiring the franchisee to pay the initial fee, ex-post opportunism is prevented (Klein, 1980). Williamson (1985) argues that in the franchise relationship the initial fee serves as a hostage. Hostages in a franchise contract refer to ‘unfair’ clauses with the expressed intent to impose a penalty upon the franchisee in cases of contract breach (Klein and Saft, 1985). From a transaction costs approach, the inclusion of safeguards in the contract significantly reduces transaction costs (Williamson, 1985).

Transaction costs in M&As

In relations to mergers and acquisitions, transaction costs economics has become the predominant underlying rationale determining the firm’s decision to engage in M&A activity (Haleblian at al. 2009). However, the main limitation in the transaction cost literature is that acquisitions are only ever investigated as an alternative to another market exchange option. To elaborate, acquisitions are solely investigated through the lens of the make-or-buy decision with an emphasis on determining when a company should acquirer another as opposed to hybrid governance structures such as strategic alliances or joint ventures. Markides and Williamson (1996) investigate operational efficiencies of acquiring companies compared to strategic alliances. Dunning (1993) suggests that acquisitions are inferior to greenfield investment as an entry

mode into a foreign country due to lower transactions costs. Hennart and Park (1993) on the other hand argue that acquisitions may be the superior entry mode if the focal firm is diversified as they have sophisticated management system in place that reduce transaction costs to a point that greenfield investment and its alternatives no longer offer sufficient benefits to direct acquisition.

To conclude, acquisition have been extensively researched using the theoretical lens of transaction costs. Yet, the focus of these studies is either to determine whether an acquisition should take place in comparison to its hierarchical or hybrid mode counterparts. The transaction cost literature seems to solely emphasize the decision between make-or-buy choices. Clearly absent in the literature is the impact of transaction costs once the ‘buy’ decision has been made. The central question is, do companies discriminate their choice of target based on varying levels of transaction costs and is there an effect on performance?

III. HYPOTHESES Franchising

simplifies environmental complexities through its business format. Due to the level of standardization between outlets and uniform guidelines, policies, contracts and overall operation method, environmental uncertainties are minimized (Windsperger and Hendrikse, 2004). Due to the reduction in environmental uncertainties, information processing is simplified. In turn, if information processing is more manageable, the impact of bounded rationality on the search process should decrease transaction costs.

The main argument that the acquisition of franchises poses less transaction costs in comparison to non-franchised is derived from the overall organizational form of franchises. As discussed earlier, franchises are organized in a hybrid mode, in which part of the units are owned by the company and some are franchised (Shane, 1986). Each of the governance structure is confronted by its respective internal transaction costs. Company-owned units are more prone to shirking due to low-power incentives and franchised units are more likely to be free-riding (Combs & Ketchen, 2014). As a response to this fact, franchisors increase direct ownership in case that free-riding hazards are prevalent. On the contrary, company-owned outlets will be franchised in situation in which shirking-related monitoring costs are high. Therefore, as a franchise system can decrease monitoring costs through franchising, transaction costs are lower comparatively to wholly-owned companies. As an example, Krueger (1991) studied the performance differences between company outlets that are operated by a franchisee and those operated and owned by the company. The author found that outlets that are managed by franchisees distinctly have lower payroll tax due to the decrease in

monitoring costs. He concludes that due to lower payroll costs as a result of a reduction in monitoring franchised outlets experience better performance.

Overall, a franchise system provides the perfect organizational form to counteract opportunism which other business formats cannot accomplish because franchise system can leverage its franchised and company-owned outlets to compensate each others caveats. In other words, a franchise system is uniquely structured to counteract the increase of transaction costs, whereas other non-franchise companies have to bear them.

Due to the lower transaction costs involved in acquiring a franchise, as a result of decreased search and information costs and mitigation of opportunism through its inherent organizational form it is hypothesized that:

Hypothesis 1: Acquisition performance is increased in acquisitions of franchises compared to non-franchised companies.

Franchising Experience

Additionally, if the acquirer is a franchise transaction costs are further decreased. Even through transaction costs are comprised as search-, bargaining- and policing costs, Dahlmann (1989) argues that this taxonomy is unnecessarily elaborate. He argues that all three dimensions can be functionally reduced to a single one as they all represent resource losses due to the lack of information. In other words, the level of overall uncertainty regulates the transactions costs of a given decision.

9 of uncertainty involved in a variety of

market exchanges. Organizational fit is comprised by a number of factors, one of which is the similarity in management styles between the acquirer and the target (Bhagat and McQuaid, 1982. Management styles encompass factors such as the the decision-making approach, risk-taking behavior, communication patterns and control mechanisms (Covin and Slevin, 1988). Through the process of the acquisition, the two conflicting management styles can lead to difficulties in the post-acquisition process (Covin and Slevin, 1988; Kerr, 1982; Leontiades, 1982; Miller, 1987). Significant differences between management styles can lead to ‘cultural ambiguity’ (Buono, Bowditch, and Lewis, 1985). That is, a state of uncertainty as to which management style will dominate. Either outcome includes the imposition of one management style upon the other. However, usually the acquired firm adopts to the acquirer. Hirsch and Andrews (1983) postulate that the imposition of another management style results in the loss of identity among the acquired firm’s management leading of poorer acquisition performance. The similarity between management styles and a shared philosophy significantly reduces the uncertainty involved in the acquisition and the post-acquisition integration (Buono, Bowditch, and Lewis, 1985). Further, differences in management styles are found in the evaluation and reward systems (Galbraith, 1977).

Another aspect of organizational fit is the combination potential between the acquirer and the target. The combination potential is captured by the relatedness of the two firms, partially in regards to the industry but also to the overall similarity between the organizational forms of the

bidder and the target (Morck et al. 1990). A high combination potential can lead to the achievement of synergistic complementarities through “economies of sameness” and “economies of fitness”. Economies of sameness are synergies between two companies that are achieved through similar operations, whereas economies of fitness are achieved through the seamless combination of these operations (Hitt et al, 1993).

To sum up, if both the acquirer and the target share the organizational form of franchising, it is expected that the positive relationship between franchising and acquisition performance is increased. The search and information costs are reduced to superior knowledge of the business form and general organizational relatedness. Furthermore, transaction costs are decreased through the reduction in uncertainty due to the similarity in management styles and overall organizational fit. Additionally, acquisitions of related bidders and target should experience superior performance are the synergistic realizations through combination potential. Therefore, the moderation effect is hypothesized as follows.

The hypothesized relationships are summarized in the following conceptual model:

IV. METHODS Research Setting

The following section briefly discusses the databases that were consulted in order to retrieve the necessary data for this study. Three distinct databases were needed in order to collect data on mergers and acquisitions, the corresponding financial data of the acquiring firms and franchising data.

Thomson Reuters SDC Database.

The Thomson Reuters SDC is a financial database that provides data on global mergers and acquisition activity. The database includes comprehensive content on the M&A profiles, specifying leveraged buyouts, payment methods, reverse takeovers, stake purchases demergers and repurchases. Furthermore, SDC provides a detailed overview of the acquiring and the target company including extensive profiles of each company (country, industry, parent company information), the acquired equity percentage, the deal value, the deal status and other vital financial data. Overall, the Thomson Reuters SDC database covers over one million deals from 1970 to the

present day (Thomson Reuters, 2016). For this study, the SDC database was used to extract data specifically on the 5411 SIC code.

Thomson Reuter’s DataStream.

Thomson Reuters DataStream is a global macroeconomic and financial data platform that covers stock market-, bond-, and equity indices, commodities, economic indicators, and exchange-, and interest rates. For the purpose of this study, DataStream was used to collect stock market and financial data on the the acquiring company. More specifically, data was collected from 1987 through 2016 for all companies in the sample.

LexisNexis. LexisNexis is the

worlds largest database of legal and public records through legal, news and business sources (LexisNexis, 2016). Furthermore, it provides detailed records of companies, comprised of annual reports, AICPA annual reports, stock reports and comprehensive company profiles. This database was the primary source of the identification of franchise businesses.

Data and Sample

All companies in the sample are in the food retail industry. Therefore, in order to compile a dataset, only mergers and acquisition from the SDC database with the SIC code 5411 were gathered. The SIC code 5411 corresponds supermarkets, food- and grocer stores primarily engaged in the retail or sale of food items. The reason for the selection of the food retail industry is that the phenomenon of franchising is most likely to occur in this segment. The selection of an industry in which franchising is most prevalent ensures a sample that is most likely to produce

11 statistical effects that are representative of

the predicted relationship between mergers and acquisitions and franchising.

Furthermore, the sample consist of M&As where both the acquirer and the target firm are located in Germany, the Netherlands, Great Britain, Spain, Italy, Belgium, Sweden, Norway, Denmark, the United States of America and Canada. Furthermore, deals have only been included in the dataset if the occur between 01/01/1987 and 01/12/2016. The data collection was cut off before 1987 as financial data necessary to conduct analysis is scarcely available before that time period. Based on these criteria, the initial sample consisted of 4513 M&As.

However, as the subsequent data collection requires a SEDOL code, all acquiring companies with a missing SEDOL code were dropped from the dataset. This reduced the dataset to 1751 acquisitions.

Following, the identification of all relevant acquisitions in the dataset, the return index of all acquiring firms was identified. This was accomplished using Thomson Reuters DataStream platform. The return index is the theoretical value of a share assuming that all dividends are re-invested (Ince and Porter, 2006). DataStream was unable to calculate the return index of 141 companies, further reducing the sample to 1610. The 1610 acquisitions fall into twelve separate stock market indexes, namely S&P500, FTSE All-Share, DAX, S&P/TSX Composite Index, SBF120, OMXAFGX, OMXH, IBEX 35, AEX, BEL 20, OBX and OMXC.

Next, the return index for respective stock market indices was calculated using DataStream. The subsequent analysis requires each acquisition to have return index data for the acquiring company and the corresponding stock market index on the

event date. However, DataStream was unable to find the return index for SBF120 before 2003, FTSE before 1991 and BEL 20 before 1990. Therefore, acquisitions in which the firm that conducted the acquisition is listed in one of the previously mentioned markets that has an event date before the return index was able to be calculated were excluded form the sample. This further reduced the sample to 1489 M&As.

Lastly, in order to determine the eventual performance of the acquisition, the acquiring firm needs to have enough stock price data surrounding the event date. Therefore, all companies that do not meet this condition were removed from the dataset, resulting in a final sample of 1325 acquisitions.

Measures

Dependent Variable. In this study

the acquisition performance is measured using an event study. An event study is an instrument with which the consequences of an event can be measured by investigating the change in firm value by the market, captured by the stock prices (Hayward and Hambrick, 1997). Event studies are a common tool to approximate performance and have been used especially in fields such as finance (Collins, Rozeff and Salatka, 1982; Cross et al., 1988) and the strategic management field (McKinley, 1997; Peterson, 1989; Binder, 1998; Kothari and Warner, 2007). Furthermore, it is one of the predominant methods to determine acquisition performance in the Mergers and Acquisition literature (see Zollo and Singh, 2004; Hayward, 2002; Cording, Christman and King, 2008).

events will be reflected immediately in security prices.

An event study measures performance by comparing the expected performance of the acquiring firm with its actual performance. The expected performance are the predicted returns that the acquiring firm would have had in the absence of the focal event, whereas actual performance measures the returns conditional on the event taking place (MacKinley, 1997). The difference between the predicted and the actual returns is represented by a measure called abnormal returns. Therefore, abnormal returns are the gains and losses of the acquiring firm on a specific date. However, in order to capture effects of information leakage prior to the event and control for slow moving information to reach the market after the event, abnormal return need to be accumulated for a predefine period. The summation of the abnormal return for that predefined time window provides the cumulative abnormal return (CAR) for the event. Hence, the CAR represents the overall performance of the acquisition in the specified time window. Consequently, in the case of a positive CAR, the acquisition created value, whereas a negative one is the result of value destruction. It is important to note that is remains controversial on what constitutes a correct announcement window (MacKinley, 1997). However, a three day event period is commonly used in the literature. Furthermore, as short-term acquisition performance is an indicator for long-term acquisition performance, the choice of a short event period is justified (Zollo, 2008)

Independent Variable. The main

rationale behind the independent variable is to discriminate the acquisition performance of target companies that either are franchises or are not. Therefore, a dummy

variable was created to test this relationship. The dummy variable is 1 if the target company is a franchise. More specifically, a target company is classified a franchise if the main business model is the replication of chain-stores through the granting of franchise licenses to a third party (Rubin, 1978). There are two types of franchising. Firstly, business format franchising in which the entire system, including the products, the trade name and franchise relationships are licensed to a franchisee and secondly product and distribution franchising the focus is mainly on the products, without the permission to use the brand name (Webber, 2012). In this study, both types of franchising are considered. Conversely, companies that do not grant the rights to their business format through licensing to third parties are classified with a 0 for the dummy variable.

Moderating Variable. In order to

ascertain whether experience with franchising influences the relationship between franchising and acquisition performance, it is determined whether the acquiring firm is a franchisor. Similarly, to the measurement of the independent variable, the moderating variable is constructed as a dummy. If the acquiring company is a franchise the dummy variable corresponds to 1. On the other hand, if the the acquirer is not a franchise the dummy variable is 0.

Control Variables

Acquiring Firm Size. Scholars have

13 customer attraction, asset growth, employee

productivity and asset productivity of large firms seem to explain why larger firms experience increased acquisition performance (Healy et al., 1992; Cornett & Tehranian, 1992). On the other hand, conflicting explanations have been brought forth by Moeller et al. (2004) who find that large acquisitions by large companies result in negative returns, whereas small acquisitions by small firms results in positive announcement returns. In this study the natural log of the acquiring firm’s employees is used to represent firm size (Becker-Blease et al., 2010).

Leverage. Leverage is defined as the

acquirer’s total debt divided by the acquirer’s total assets. According to Hitt et al. (1998) leverage is an indicator for a company’s ability to maintain a position of strength and to either hold or even gain a competitive advantage. The authors further argue that leverage therefore can affect the outcome of an acquisition.

Cross-Boarder Deal. This variable

distinguishes between domestic and international deals. The reason for the inclusion of this control variable is that domestic deals seem to outperform cross-boarder deals (Datta and Puia, 1995). The authors claim that the difference in performance seems to stem from the inability of managers to accurately evaluate a fair price for the acquisition which is negatively reflected in shareholder wealth. Furthermore, the unfamiliarity with foreign conditions and conventions can increase uncertainty to a point that it can reduce acquisition performance (Davis, Shore, and

Thompson 1991). A dummy variable was constructed in order to capture the difference between cross-boarder deals (1) and domestic deals (0).

Public Deals. The distinction

between whether the target company is a public or a private firm is included as there is evidence towards the fact that returns for publicly listed companies differ considerably from private companies (Fuller et al., 2002, Officer, 2007). Research suggests that the acquirer is more likely to positively benefit from the acquisition of a private target than a publicly listed one. The superior returns of private targets are likely attributable due to reduced agency costs and the fact that the acquirer simply gets a better price for a private company. To distinguish the two, a dummy variable was created, specifying public (1) and private (0) deals.

Prior Firm Performance. The past operating performance of the acquirer seems to be an indicator of acquisition performance. Haleblian et al., (2009) argue that better performing firms tend to experience a superior post-acquisition performance. For the purpose of this study, the acquirer’s prior financial performance is measured following (Bhagat & Bolton, 2008) by recording the return on asset of the year prior to the acquisition year.

Free Cash Flow. The free cash flow

hypothesis states that managers endowed with free cash flow would rather invest it in net negative value projects rather than pay it out to their shareholders (Jensen, 1986). Mergers and Acquisitions are one of the

ways managers would spend excess cash flow. This was confirmed by Stulz, and Walkling (1991). According to the hypothesis, companies with free cash-flow would therefore invest in value destroying deals. Hence, in this study free cash-flow is included as it has a negative impact on deal performance. Free cash-flow is calculated by subtracting taxes, interest expenses, preferred dividends and common dividends from the acquirer’s operating income, and all divided by equity.

Market-to-Book Value. Lastly, the

Market-to-book value is controlled for. Rau and Vermaelen (1988) suggest that companies with a low market-to-book value are more likely to overestimate their own abilities to manage acquisitions, which will result in a decrease in acquisition performance. Market-to-book value is defined as the acquirer’s market value four weeks prior to the announcement divided by total assets.

The variables used in this paper, are summarized in table 1. Additionally, the sources for the data collection of each respective variable is presented.

Statistical Methods

Event Study. To measure the acquisition

performance of the firms in the food retail industry, an event study was performances. An event study determines the impact of a specific event on the performance of the firm (Fama, 1969). Stock price fluctuations before and after the event date are evaluated to ascertain the returns of the acquiring company. Furthermore, the average industry performance is calculated in order to generate a baseline upon which the acquiring firm is compared. The industry performance is approximated by the combined return index based on all companies in the corresponding stock market of the acquiring company. Based on the difference between the predicted performance, in the absence of an event, compared with the actual returns of a company, abnormal returns can be determined. Lastly, by adding all abnormal return in the event window, the acquisition performance is measured. Following is a detailed explanation the steps of the analysis.

15

1) Event Period

Firstly, an event period and estimation window have to be specified (MacKinley, 1997). An event period are the days surrounding the acquisition event. The reason to include the prior days to the event is that the market might have already reacted to the acquisition based either on rumors or information leakages. Post-event returns must also be considered as some information might be slow reach the market (MacKinley, 1997). There is no consensus in the literature specifying an optimal event period. However, evidence seems to indicate that short-term event periods capture the significance of the specific effect, whereas an enlarged event window might capture a more extensive reaction from the market, yet at the cost of confounding the results by introducing other influence that impact the market reaction (McWilliams & Siegel, 1997).

Therefore, in line with previous research, a 3-day event period is used in this study (Friedman & Singh, 1989; Houston & Ryngaert, 1994; Mikhail, Walther & Willis, 2004, Brown & Warner, 1985). To predict how the normal returns of the acquiring company should be in the absence of the event, an estimation window of 30 days is created.

Figure 2 depicts the above described estimation window of 30 days and event period of one day before and after the announcement day.

The next section describes how the actual and abnormal returns of the acquiring company are computed.

2) Abnormal returns

Abnormal returns are determined by subtracting the expected returns from actual returns. The actual returns are determined by the rate of return on the share price of firm i on the day t. The rate of return is expressed as:

(1) The rate of return on the share price of firm

i on day t is represented by "_#$. "_%$ is the rate of return on a market portfolio of stocks on the same day t. The rate of returns of the market is composed by the following stock markets: S&P500, FTSE All-Share, DAX, S&P/TSX Composite Index, SBF120, OMXAFGX, OMXH, IBEX 35, AEX, BEL 20, OBX and OMXC The intercept term represents the systematic risk of stock i. Lastly, the error term is added, which is expected to be 0 (McWilliams & Siegel, 1997). The intercept coefficient and the slope are calculated accordingly:

(2)

firm, by subtracting predicted returns based on the market performance from the actual returns of the acquirer using the subsequent equations:

(4) Lastly, all the abnormal returns in the specified event window are added in order to obtain the cumulative abnormal returns (McWilliams & Siegel, 1997). The following equation adds the abnormal returns from &_', one day before the event and &₍, one day after the event.

(5)

Ordinary Least Squares (OLS). To

test the hypotheses a regression analysis is conducted to evaluate the relationship between the independent and the dependent variable (Greene, 2003). More specifically, an OLS was used to predict the relationship in question. The sample consists of 1325 acquisitions in the food retail industry between the years 1987 and 2016. In order to determine whether an OLS regression is adequate for the given dataset, the sample was tested for multicollinearity, normality, and outliers (Hair, 2009).

First, in order to test multicollinearity the variance inflation factor (VIF) was performed. The VIFs for the explanatory variable ranges between 1.01 and 1.87. The threshold to determine whether a sample suffers from multicollinearity is VIF ≥ 5 (Craney and Surles, 2002). As none of the explanatory variables surpass that threshold, it is determined that multicollinearity is not present in this study.

The second assumption that needs to be satisfied in order to accurately interpret the results is the condition of normality

(Greene, 2003). Therefore, the residuals were tested for normality using a Skewness and Kurtosis test. The skewness coefficient was 0.001, indicating that the residuals were normally distributed (Bai and Ng, 2005).

Lastly, outliers, especially in OLS, can distort the results drastically. Therefore, the dependent variable was winsorized at a (0.01) level to remove the outliers.

The OLS was performed in a hierarchical manner in order to better visualize the explanatory power of each model (Gelman and Hill, 2006). Model 1 is the base model only including the controls. Model 2 adds the effect of the independent variable ‘Franchising’ to the regression analysis. Finally, Model 3 includes the interaction effect of ‘Franchising Experience’ with ‘Franchising Target’.

V. RESULTS Descriptives & Correlations

whether the acquirer is a publicly traded company influences the likelihood that the acquisition was across boarders. Next, the variable market-to-book value is positively influenced by firm size and prior firm performance, whereas leverage significantly decreases it. In addition, it is worthy to point out that prior firm performance is negatively associated with the company’s leverage. Lastly, whereas the previously mentioned correlations were significant, they were not strong enough to consider excluding them due to issues of multicollinearity. However, free cash-flow is highly correlated to firm size (r=0.9299). The strong correlation between the two control variables raises questions regarding the possible existence of multicollinearity in the sample. Therefore, a variance inflation factor test was conducted. As no VIF value surpassed the threshold of 5, it was concluded that multicollinearity would not be an issue and that all control variables would be included in the sample.

Focusing on the deal characteristics of the sample, it can be observed that far more deals are performed domestically, than across boarders. In 25% of the the acquisitions, the bidder acquired a target outside the focal firm’s home country. Furthermore, the descriptive statistics show that that 98% of the acquisitions were performed by a public bidder.

Table III shows the characteristics of the independent variable Franchising. It shows the distribution of franchising among the acquired companies in the sample. Of the entire sample (n=1325), 354 acquisitions were of franchises, whereas 971 of the acquired companies were not. In other words, 26.72% of all acquisitions in this sample include the entire or partial purchase of a franchise company. Furthermore, the average acquisition performance is shown in the table. Of the entire sample, the average increase in CAR was 1.12%. To elaborate, on average, in this sample, the acquisition performance is equal to a 1.06%increase in firm value. If, however, the sample is divided into franchised and non-franchised targets the impact of acquiring a franchise is already recognizable. On average, acquirers of franchises realized a rate of return of 2.7% in their cumulative abnormal returns. On the other hand, of the 971 acquisitions of non-franchises, the average return was 0.55%. Interestingly enough, regardless of the organizational form of the target, the cumulative abnormal returns were positive. Nonetheless, the average performance was significantly higher among franchised targets.

In order to accurately determine whether franchising can predict acquisition performance an OLS regression was performed. The results are presented in the next section.

Regression analysis

The OLS regression analysis was conducted in order to assess the predicted relationship between franchising and acquisition performance. The regression analysis was conducted in a hierarchical fashion in order to more accurately observe the influence of the independent and moderating variable.

19 Model 1 includes the control variables in

order to create a base model. In Model 1, one of the control variables significantly influences the acquisition performance of the acquirer. Firm size is found to negatively influence acquisition performance (P<0.1). However, the limited explanatory potential of the base model is represented in the low =0.011. Hypothesis 1 argues that, all else being equal, the acquisition of a franchise will outperform an acquisition of a non-franchise. To test that prediction, the independent variable is added to the base model in Model 2. In support of hypothesis 1, a positive and significant (P<0.01) relationship is found

between franchising and acquisition performance. This indicates that the distinction of acquiring a franchise is significantly different form acquiring a non-franchise in terms of subsequent acquisition performance. More precisely, acquiring a franchise results in an increase of 1.66% in cumulative abnormal returns. Additionally, firm size is negatively related to the acquisition performance (P<0.05). However, due to the low coefficient of -0.0015 and the low significance level it is concluded that the effect of firm size is negligible in the assessment of the determinants for acquisition performance. Overall, Model 2 has an !"_{=0.038, which}

Table III:

Regression Analysis - Hypothesis Testing

demonstrates an absolute increase of 2.7% in the explanatory potential compared to Model 1. This indicates that introducing the independent variable franchising to the analysis results in increased acquisition performance and a higher !"_{. Thus, it is}

concluded that there is sufficient evidence to support hypothesis 1. Finally, in the third model the moderation effect between the organizational fit and the target is included. The moderation effect was included to test hypothesis 2, which argues that when the acquirer is a franchise, the positive effects of acquiring a franchise are further increased. In order to isolate these effects, the third model only includes the results for acquisitions in which the acquisition was performed by a bidder that is a franchise. A strong positive relationship (P<0.01) is found between franchising and acquisition performance. Interestingly, similar to the preceding two models, firm size is negatively related to acquisition performance (P<0.05). In comparison to Model 2 which disregards the organizational form of the acquirer, Model 3 demonstrates a larger creation of value with cumulative abnormal returns of 2.08%. Furthermore, Model 3 has the highest !"_{=0.05, indicating that it explains the most}

variance in the dependent variable. Therefore, it is concluded that there is enough evidence to support hypothesis 2.

Robustness

The main limitation of the event studies is that the exact method to capture the entirety of the event remains questioned (MacKinley, 1997). Different outcomes could be generated depending on the amount of time information is allowed to influence the market value. In order to limit the amount of excess information of the market to distract from the actual

acquisition, the event period was reduced to the day before and after the event. The rationale is that information may leak out ahead of the event. Additionally, the day exceeding the event day must be considered as the announcement may occur after the closes (MacKinley, 1997). However, in the literature many variations of the event study methodology are presented. Specifically, the event windows of one-week (e.g. Ishii, & Xuan, 2014; Golubov, Petmezas, & Travlos, 2012, Ahn, Jiraporn, & Kim, 2010, Antonios, Dimitris, & Huainan, 2007) and one-month (e.g. Fan & Goyal, 2006, Bae, Kang, & Kim, 2002), are frequently used.

21 variable is, however, lower than in the

initial model. Whereas the acquisitions of franchises by other franchises (Model 3) produced an !" _{= 0.05, the was reduced to}

0.035. This concludes that the observed effects of a three-day event window are still found in a larger event window, yet are less pronounced. However, none of the hypothesized effects are found when enlarging the event window to a month. Merely, the control variable firm size showed a negative but significant (P<0.05) relationship in Models 7 and 8. Most importantly though, the effect of

franchising does not determine a difference in acquisition performance any longer.

Curiously, none of the control variables with the exception of firm size demonstrated the predicted effects presented in the literature. The coefficients of the control variables were as expected, in the sense of their general direction, yet remain insignificantly distinguishable from zero. In order to investigate this peculiarity, the OLS regression analysis has been repeated using lagged variations of the control variables free cash-flow, leverage, and the market-to-book value (see Appendix, Table 1). That is, the value of each of these variables for the year prior to the acquisition was used to conduct the regression (Murray, 2006). With the control variable lagged, the overall relationship

Table IV:

Regression Analysis - Robustness Check

between franchising and acquisition performance does not deviate from the results of the initial analysis. A few interesting results emerge, however. Similar to the initial OLS regression, firm size posits a significant and negative relationship to acquisition performance in all three models(P<0.05). Prior firm performance, as found by many scholars

(Haleblian et al., 2009) indeed significantly

and positively influences acquisition performance in the base model and with the inclusion of the independent variable (P<0.01). Moreover, the leverage of the prior year, that is the ratio of total assets to total debt, significantly increases both models 10 and 11. Strangely, the positive effects of both prior firm performance and leverage of the previous year both disappear when including the moderating effect in model 12. Additionally, it is important to note that is higher in all three models in comparison with the initial analysis. The base model 10 shows an !"_=0.021,

corresponding to twice the size of the base model 1. Equally surprising is that model 11 presents an !" _{=0.049, which also}

demonstrates a higher explanatory potential than the second model in the initial analysis. The last model, however, is comparable both in the effect of the relationship between the dependent variable and the moderating effect, but also in regards to the Suspicious of the increased effects, the variable inflation factor test was conducted. Both prior firm performance and leverage (p.y.) have a VIF of 3.37 and 3.36 respectively, indicating a strong correlation between the two. While the cutoff value of the VIF is at 5, and none of the variables surpass that threshold, the increased positive effects of both variables in addition to a higher could be an indicator of multicollinearity. If that is indeed the case,

the effects of the controls in the robustness test must be taken with a grain of salt.

VI. DISCUSSION

The purpose of this study was to investigate whether franchised companies outperform wholly-owned companies. In order to do so, an event study was performed to determine how the market would respond to acquisitions of franchises. The main findings of this study suggest that (1) acquisition performance is increased when a franchise is the target of the acquisition. Results indicate that share return of the acquiring company are increased by 1.66% in the event that a franchise is acquired. Furthermore, the results show that (2) the relationship between acquisition performance and franchising is positively influenced in situations in which the acquirer is a franchise. The shareholder wealth gains of the bidding company are 2.06% higher than for the acquisition of independent businesses.

23 insignificant (Yaghoubi et al., 2016).

Therefore, the fact that the two smaller event windows show similar results and long-term windows are frequently insignificant, leads to the conclusion that the results are robust.

Another anomaly that became apparent in the results of the regression analysis is the fact that all but one control variable demonstrated an insignificant effect on acquisition performance. This is surprising as all variables were identified in the literature as paramount in determining acquisition success. in an attempt to investigate further, lagged version of the variables were used to repeat the analysis. However, the lagging of the variable seemed have resulted in collinearity, rendering any implications of the findings invalid. Therefore, any conclusions will be derived operating under the assumption that none of the control variables, with the exception of firm size, influence acquisition performance.

The presented findings contribute equally to the franchising and the M&A literature stream. In franchising it has long been theorized that superior performance can be reaped by utilizing the franchise format. Evidence of this phenomenon has been limited as the underlying theory behind most studies in the field are more concerned with the process of franchising rather than the outcomes. Nonetheless, there are few attempts in characterizing the differences in performance between franchises and independent companies. For instance, in a literature review, Combs et al. (2004) argue that using the extant theory rationale franchising should positively increase performance. Extant theory argues that both the franchisor and the franchisee specifically choose to collaborate and initiate a franchise agreement in order to improve their ability to compete, and

maximize their earnings. As a logical consequence, performance in franchising should be superior to its alternatives.

Based on this theoretical assumption, empirical examinations inevitably followed. Roh (2002) argued that franchising increases financial performance due to the reduction in of variability in operating cash flow. The author confirms this by finding observing that publicly owned restaurants with a higher share of franchised than company owned outlets experience less variation in their operating cash flows, thereby creating a more favorable risk-return trade-off. In order to directly capture the franchising performance effect, Spinelli et al. (2003) measured a portfolio of 91 franchising stocks on the Standard and Poor’s 500 Index and compared their performance against the Index. The authors found that during the period of 1991 to 1997 franchising companies outperformed the index in terms of total return to the shareholder. The most straightforward measurement of the performance difference, however, was conducted by Madanoglu et al. (2011). The authors compare risk-adjusted financial performance of franchised and non-franchised companies in the restaurant industry. Financial performance in this study was approximated by five financial rations. Their findings show that franchised companies outperform non-franchised companies in all five separate measurements.

franchises result in a higher shareholder wealth gain is testament to the fact that the market evaluates franchises favorably towards the alternatives. Therefore, the evidence put forth by this paper complements previous research and confirms unresolved theoretical predictions.

Furthermore, this study contributes to the mergers and acquisition literature in the following ways. Firstly, it contributes to the empirical gap of adverse acquisition performance. Failure rate of M&As is in the realm of 70-80%, which begs the question why firms insist on relying on them (Nahavandi &Malekzadeh, 1993). While short-term returns are reaped by the target of the acquisition, the acquiring company often underperforms, experiencing negligible overall wealth gains (Krug and Aguilera, 2005). The phenomenon of a failure prone strategy that has experienced unprecedented growth, as M&As have, captured the attention of many scholars. In the strategic management literature, acquisition performance is mainly analyzed based on pre-merger issues (strategic fit) and post-merger issues (post-acquisition integration) (Bauer and Matzler, 2013). Ahuja and Katila (2011) focus on the ‘strategic fit’ between the bidder and the target. It is argued that knowledge transfer, resource combination and management styles determine the level of strategic fit, which will in turn lead to a higher overall acquisition performance (Dutta, 1991, Ahuja and Katila, 2001). A greater strategic fit is vital to acquisition performance as it increase the potential market power and productivity of the focal firm (Cartwright and Schoenberg, 2006). Additionally, a main emphasis is placed on the post-acquisition integration of the target company in the operation of the focal firm (Datta, 1991). The integration of the

acquired company is paramount as value creation occurs during the post-merger phase (Bauer and Matzler, 2013). The strategic fit between the two companies will eventually determine the speed of the integration between the two companies and determine value creation. As argued in the theoretical background of this paper, the level of uncertainty in these stages is vital in to decision making process and target selection of the acquirer. After all, strategic fit cannot be fully evaluated prior to the acquisition. The findings of this paper show that the reduction of uncertainty through acquiring a franchise results in positive acquisition performance. Transaction costs caused by uncertainty are further diminished through organizational relatedness, shown by the fact that wealth gains are moderated by the fact that both acquirer and bidder share the same business format.

Another contribution to the literature is the lack of a significant relationship between the control variables and acquisition performance. Evidence for the control variables is strongly established in the literature and confirmed in a variety of different studies and research settings

(Haleblian et al., 2009). This could suggest

that the impact of these variables are conditional upon context in which the acquisition takes place. This non-finding serves as an opportunity for further avenues of research.

VII. CONCLUSION

25 rationale of doing so, is that the event

studies of M&As reflect the value of any given company, evaluated by the market. Therefore, the performance differences of franchises and non-franchises are theorized to manifested in these market valuations. A positive and significant relationship was found between acquiring franchise targets and acquisition performance. More specifically, acquisition of franchises generates a 1.66% increase in shareholder wealth. As the market is assumed to be efficient, meaning that all activities and information are immediately reflected in the share price, it can be concluded that the 1.66% percent increase is a direct consequence of the markets evaluation of the franchise performance. Additionally, it was hypothesized that in the event that the acquisition was carried out by a franchise, the relationship between franchising and acquisition performance is positively moderate. The evidence supports this claim as the overall wealth gains were 2.06% higher in comparison to non-franchises. The following sections will outline the managerial and theoretical implications. Furthermore, the limitations are discussed.

Theoretical Implications

Several theoretical implications can be derived from the main results. Firstly, this study provides empirical support for the transaction cost economics approach. Furthermore, it uses TCE in a novel manner, focusing specifically on performance of franchises and acquisitions. Secondly, and most importantly, was argued that (1) franchises perform better due to lower transaction costs and (2) that a reduction transaction costs will positively influence the acquisition performance. The positive results confirm the adequacy and the necessity of using transaction costs in

acquisition decisions. As argued previously, transaction costs are primarily used to provide a solution to a make-or-buy decision (Walker and Webber, 1984). However, the vast literature neglects to describe and predict if transaction costs are further applicable as a decision making mechanism once a firm decides to ‘buy’. The fact that the acquisitions of companies that reduce transaction costs leads to increased acquisition performance, proves the fact that transaction costs are applicable to the target selection process in the ‘buy’ decision.

Practical Implications

process. As franchises are independent entrepreneurs, they will for the most part remain autonomous from the parent company. Therefore, expansion through acquiring franchising offers a strategy with increased benefits of immediate local adaptation without the drawbacks of cultural differences.

Furthermore, even though the results suggest that this strategy is more profitable for franchise companies, the acquisitions of franchises is profitable for non-franchise companies. Therefore, for independent businesses that want to expand geographically or simply want to increase their market share, can similarly use this strategy to create value.

Limitations

This study assumes several limitations that may create avenues for further validation or future research. Firstly, due to the event study methodology, only acquiring companies that operate publicly are included in this data. As the share price returns are the main element of performance in this study, all private firms are discriminated against. Therefore, the findings in this study may be biased towards acquisition carried out by private firms. The initial sample was comprised of 4513 M&As in the food retail industry. Of those observations merely 1325 were included in the final analysis. Hence, the possibility exists that the relationship between franchises and acquisition performance behaves differently for private acquirers. In order to include the remaining firms in the analysis, a different methodology is needed.

Secondly, the analysis was based on the data derived form a single industry, specifically the food retail sector. This

sector was chosen as is possesses the relevant characteristic necessary to conduct this analysis, namely the popularity of the franchising format, the richness of public companies and general predisposition of chain-like and standardized company outlets. However, replicating this study in another industry would increase the validation of the findings immensely.

Another limitation regarding the general impact of the results is that event studies merely capture a snapshot of the value creation of an acquisition. In order to explicate true acquisition performance and to conclusively determine whether franchises create superior value, the long-term effects of these acquisition need to be generated and analyzed. However, in order to do so, future research needs to provide a apply a different methodology as event studies cannot accurately take into account long-term effects.

Fourth, the cross boarder effect was approximated rather rudimentarily. To truly capture the effects of geographic distance in franchising and acquisition performance, a more sophisticated variable taking into account total distance should be constructed.