Analysis of franchise performance through use of a typology: an Australian investigation.

Abstract

There have been increasing attempts to examine whether the franchising industry is homogenous or heterogeneous. This paper describes a study in which the characteristics of 93 Australian franchise firms were examined using hierarchical cluster analysis. Distinct characteristics were found to exist within four franchise groups which reflected the heterogeneous characters of the franchise firms. Subsequently, the four franchise groups were confirmed against key performance measures using discriminant analysis. The research findings suggest that the franchising industry is heterogeneous and its heterogeneous characteristics can be used to predict the levels of the group performance. Practical implications for franchisors and franchisees are discussed.

Keywords: Franchising; business performance; organisational characteristics; franchise profiling; cluster analysis

Studies of franchises have often been conducted with an underlying assumption that they operate similarly, and are relatively homogeneous in their structures and characteristics (for example, Doherty and Quinn, 1999; Frazer et al, 2000). Most studies of proportional franchises have found general relationships without first distinguishing between franchise types (for example, Combs and Castrogiovanni, 1994; Shane, 1998). However, there is an increasing recognition that franchises progress through various stages of development, and that these stages influence the way they operate and perform (Quinn and Rohrbaugh, 1983). Despite this growing understanding of franchise diversity, research on franchising often fails to recognise the heterogeneous nature of franchises. Franchises differ greatly in the ways they do business. For example, franchises vary considerably in terms of the number of franchises generated, the experience of the franchisors, the entry fee for each franchise outlet, the ratio of franchised versus managed stores, and the degree of expansion within both local and more widespread localities. Notwithstanding this complexity of franchise qualities, the data set in franchising research is largely treated as homogenous (for example, Agarwal and Ramaswami, 1993, Alon and McKee, 1999; Falbe and Welsh, 1998; Frazer et al, 2000).

The heterogeneous profile of franchising may also be related to the levels of firms' maturity within the franchise industry. The process of franchise firms' maturity can be explored in terms of the organisational development model in which changes in organisational structure follow a predictable pattern (Dodge et al, 1994). This continuous interaction between organisation and environment leads to various stages of organisational development (Lillis et al, 1976; Smith et al, 1985).

It has been argued that the profile of franchise development stages reflects the life cycle model of the franchise industry (Lillis et al, 1976). As in the case of franchising, the franchisor continues to evolve the business strategies according to their competitive environment. By understanding the discrete stages of franchise development, franchisors may accurately predict the problems and opportunities that they may encounter in each development stage (Cameron and Whetten, 1981), and thereby apply different strategic priorities to encourage successful growth and development. If the franchisor fails to adapt, the firm may stagnate or decline over time (Lillis et al, 1976; Smith et al, 1985). Empirical research of franchising may also suffer from insufficient recognition of the developmental influences on franchise outcomes (Castrogiovanni and Justis, 2002; Julian and Castrogiovanni, 1995). Thus, the capacity to clarify the nature of franchise heterogeneity should be central to franchising research. An understanding of the characteristics of each franchise stage might therefore assist franchisors in increasing the likelihood of enhanced franchise performance.

This paper seeks to identify and explore the typology of franchising heterogeneity, using the results of a study of Australian franchise firms. If the heterogeneous typology is positive, the links between the franchising heterogeneous profiles and franchise performance will be further established. Part I of this paper explores the heterogeneous profiles of franchising while Part II focuses on linking the established profiles to franchise performance.

Part I

Franchise Typology

There seems to be disagreement as to whether franchise firms can be grouped by their analogous characteristics or should be treated on the basis of general wide sample. Studies of heterogeneous franchising argue that franchise firms may be grouped according to their archetypal features such as their structure and strategies (Callum and Graham, 1999; Carney and Gedajlovic, 1991; Castrogiovanni et al, 1993; Floyd and Fenwick, 1999; Lillis et al, 1976; Oxenfelt and Kelly, 1968-69). These studies base the heterogeneous profile of franchises primarily on the creation theory of franchising: the resource scarcity theory and the administrative efficiency theory. Resource scarcity theory predicts that firms of small and medium sizes will franchise to obtain capital or human resources and to transfer the business risks to the franchisees (Oxenfelt and Kelly, 1968-69). However, the theory suggests that once the firms are established, their requirement for capital or human resources is reduced, thereby decreasing the strategic value of franchising (Caves and Murphy, 1976). Firms in the initial penetration stage may engage heavily in franchising to raise capital resource and to recruit a hopefully efficient agent as a means of achieving organisational growth. As the systems increasingly mature, firms franchise fewer outlets or employ a buy back policy of the profitable outlets (Hunt, 1974). Thus, the resource scarcity theory explores the use of management priorities and the strategic values to understand the process of maturity of franchise firms.

Alternatively, the administrative efficiency theory views franchising as a response to agency problems with exchange relationships. This theory predicts that firms recognise the benefits of franchisee motivation and risk-sharing for all stages of franchise development (Lillis et al, 1976). A range of explanations has evolved to support this perspective. For example, it is suggested that in standard commercial form, managers (the agents) tend to shirk their duty to the firm (the principal) because their compensation is fixed, resulting in high monitoring costs to firms (Rubin, 1978). Conversely, firms engaged in franchising reduce the monitoring costs because the contract between the franchisor (the principal) and the franchisees (the agents) is designed to keep their incentives closely aligned (Brickley and Dark, 1987; Mathewson and Winter, 1985; Rubin, 1978). On this basis, the franchisees will act in the franchisor's best interest as their takings are linked to those of the performance of the franchise outlets (Lafontaine, 1992). Research confirms that franchisors are likely to franchise an outlet in geographically dispersed areas to reduce the costs of administrative monitoring (Brickley and Dark, 1987; Norton, 1988a; Norton, 1988b). The administrative efficiency theory infers that the growth and expansion strategy of franchise firms would differ at each stage of franchise development. As with the resource scarcity theory, the administrative efficiency theory suggests that franchise firms go through different stages by employing their best fit strategies to suit their business environment.

The continuous debate between the resource scarcity theory and the administrative efficiency theory seems to revolve around the heterogeneous nature of franchising. While there is vigorous discussion of a heterogeneous typology of franchising, other studies appear to have assumed and applied an homogeneous view (for example, Krueger, 1991; Lafontaine and Shaw, 1998; Withane, 1991). For example, studies on franchise performance have often been equalled the performance of the franchise firms with small independent businesses (Howard and Hine, 1997). Many studies assume that franchises predominantly operate as small entities with sole proprietors and less than 10 outlets (for example, Bronson and Morgan, 1998; Carman and Klien, 1986). The findings from these studies suggest that there was an increased likelihood of business success for franchise-based operations and that their results have then been generalised to the whole franchising sector. However, the underlying assumption of high business performance may not be homogeneous. The franchise system ranges from sole proprietors to large multinational corporations, and newly-started businesses to well-established enterprises. Therefore, franchising studies should take the characteristics of the data sample into consideration.

The heterogeneous typology studies argued that franchise firms with similar characteristic profiles may be grouped together so that their strategies, inputs, and outcomes are easily identified and evaluated. In earlier investigations, researchers grouped franchise firms based on their age and size (for example, Lillis et al, 1976; Oxenfelt and Kelly, 1968-69). More recently, Carney and Gedajlovic (1991), using a configuration procedure to group 13 parameters relating to franchise ownership, derived five complex franchise categories: the rapid grower, the expensive conservative, the converters, the matures and the unsuccessful (see Table 1). Their findings identified heterogeneous effects of the franchising industry so they argued that since franchise firms can be clearly classified into groups, a homogeneous approach to the data set was inappropriate. As a result, researchers should recognise the distinctive profiles of each franchise group when analysing franchising data. Castrogiovanni, Bennett and Combs (1995) replicated the Carney and Gedajlovic (1991) study, using similar parameters, but found partial support for the theory. The study findings matched two other groups: 'the rapid grower and the converter' (see Table 1). Their limited support argues that franchisors have to balance resource and incentive consideration to respond to strategic choices in the early stage of franchise development. Although Castrogiovanni, Bennett and Combs (1995) did not dispute the heterogeneous profile of franchising, the confirmation of the grouping profiles has been problematic.

The lack of group consensus among the heterogeneous typology may be due to weaknesses in the study methodologies. For example, Lillis et al (1976) and Oxenfelt and Kelly (1968-69) constructed a franchise life cycle profile based on two dimensions: company age and size. Given that franchise organisations generally operate in complex environments, this simple classification approach may not be sufficient to explain franchise groups, particularly as the relationship between firm age and size may not be linear. In addition, the methodology of the configuration procedures conducted by Carney and Gedajlovic (1991) were based on factor analysis which is an inductive procedure. The number of identifying factors is subjective, thereby possibly lacking empirical validation (Hair et al, 1998). While the use of factor analysis for a configuration procedure is appropriate and the selected parameters adequately covered the aspects of franchise operation, a lack of empiricism was claimed, as the results were difficult to replicate (Meyer et al, 1993). Thus, the issue of methodological weakness in franchise development studies highlights the need for further empirical research that validates previous findings.

Although divergent views on the key characteristics of classifying franchise groups or development remain, the correlation of a heterogeneous typology seems justified and generally accepted. It has been argued that franchise organisations can be profiled in terms of their stages of evolution (Lillis et al, 1976; Oxenfelt and Kelly, 1968-69). Using a more complex configuration, it is also evident that franchise firms evolve over time and tend to adjust their business strategy accordingly (Carney and Gedajlovic, 1991). Hence franchise firms are likely to be heterogeneous in their profiles and these profiles may be clearly distinguished and justified.

[H.sub.1]: Franchisors are heterogeneous and can be classified into groups on the basis of their profiles.

Method

Participants

A sampling frame comprising 351 Australian franchisors was devised for the study. Participants were drawn from various franchise systems ranging from large retail enterprises to small educational providers. The wide variation in size and structure was actively sought to provide the sampling variety necessary to check for heterogeneity. Multinational franchise firms were omitted from the survey to reduce any possible cultural effects caused by international practices and influences. The franchisor sample was drawn from the Franchise Yearbook and Directory, with a randomised selection process used to identify possible participants. A response rate of 27 per cent was obtained (n = 93), after two follow-ups to non-respondents. Eighty three per cent of the respondents were male and 17 per cent were female. The majority of the franchisors had been in the franchise business for 3-15 years (74 per cent) and were between 41 to 60 years of age (68 per cent). One-third of the respondents held higher qualifications (MBA, graduate and post-graduate qualifications). The proportion of those with no formal education was relatively small (3 per cent).

Overall, the contribution was obtained from franchise respondents Australia-wide. The proportion of the franchise systems and their locations appeared to adequately represent Australian franchise organisations. Respondents from Victoria represented the highest proportion among all the states (29 per cent), followed by those from New South Wales (25 per cent), Queensland (20 per cent), Western Australia (16 per cent), South Australia (6 per cent) and Australia Capital Territory (3 per cent). These figures were proportional to the number of the franchisors reported in each State in the Franchising Survey conducted by the Franchise Taskforce (McCosker and Frazer, 1998).

Instrument

A self-administered questionnaire was constructed for use in the study after initial pilot testing of the instrument with local Western Australian participants. The survey requested information relating to participant backgrounds and the company characteristics. Data on six dimensions representing 10 variables, relating to size (total number of outlets and number of franchise outlets), degree of dispersion (percentage of outlets located in home province), growth orientation (outlets established per year since inception and franchise outlets established per year since first franchise), pricing (average capital requirements per franchise), vertical integration (percentage of outlets franchised), and timing (year since inception, year since established of first franchise, time since first inception and first franchise) were collected (Carney and Gedajlovic, 1991). Other information was also collected, which will be reported in the later part of this study.

Analysis

Prior to the analysis, univariate outliers were investigated through an initial data screening and multivariate outliers were examined through the use of a p < 0.001 criterion for Mahalanobis Distance ([D.sup.2]). Two cases of univariate outliers detected on the company size from the sample were eliminated. The elimination of the outliers reduced the sample size to 91. The six franchise dimensions were used to map a likely distinctive profile of franchise organisations.

Agglomerative hierarchical cluster analysis was used to test the first research hypothesis. Cluster analysis was performed to determine if the franchisors could be effectively segmented into unique groups according to their organisational characteristics. Ward's (1963) method with the squared Euclidean distance was preferred as a clustering technique because it is an effective tool for identifying distinct groupings for a relatively small sample size (n < 200) (Everitt, 1993). Ward's (1963) method calculates the means of all variables within each cluster, then calculates the squared Euclidean distances to the cluster mean of each case, and finally sums across all cases (Everitt, 1979). The Ward method with the squared Euclidean distance has been claimed to be the most reliable method of the hierarchical cluster routine for uncovering group structures in the data set (Lassar and Kerr, 1996, Punji and David, 1983). As a result, this technique has been increasingly employed in the area of marketing and organisational research (for example, Lassar and Kerr 1996; Miles and Covin 2000; Miller and Besser 2000).

The results of the hierarchical clustering procedure suggested a four-cluster solution. The optimum number of clusters was determined by an examination of the agglomeration schedule and the dendrogram results as recommended by Aldenderfer and Blashfield (1984). The agglomeration schedule revealed a large increase in the agglomeration coefficient at the 87th stage, indicating a four-cluster solution was the most appropriate. Accordingly, the dendrogram clearly displayed four partitions of the graphic illustration. The confirmed results, then, suggest that there were four distinct groups among the franchisors examined in this study.

Multivariate Analysis of Variance (Manova) was used to further test the clustering solution (Aldenderfer and Blashfield, 1984). Results of the analysis confirmed a significant effect, demonstrating that the four clusters differed in their profiles ([F.sub.3, 87] = 18.21, p < 0.001). Follow-up Analyses of Variances (Anova) with post hoc procedure (Dunnett T3 multiple comparison tests) were also used to test the significance of each of these clusters. Table 2 displays the cluster structure on means, standard deviations and statistical tests of four cluster groups.

The results from Ward's (1963) clustering procedure provided support for Hypothesis 1: Franchise organisations were found to be heterogeneous and could be grouped on the basis of their characteristic profiles. The four distinct groups identified from the analysis were labelled as beginners, developers, growers, and matures. The parameters that described these profiles involved size, timing, and strategy orientations, thus, we labelled the cluster profiles as the 'franchise stages' (Lillis et al, 1976; Oxenfelt and Kelly, 1968-69). The detailed profiles of these franchise stages are explained later in the result section.

Part II Cluster Evaluation

To confirm the stability of the derived cluster profiles, the identified franchise stages needed to be further evaluated as a major limitation of the clustering procedure is a lack of significant testing (Hair et al, 1998). Multiple discriminant analysis was used to evaluate the franchise stages. This analysis was then used in conjunction with cluster analysis to validate the identified groups and to assess the cluster stability (Dillon and Goldstein, 1984; Hair et al, 1998). While cluster analysis classifies the unknown structure in the data, discriminant analysis statistically tests the internal validity of the derived franchise stages. In addition, discriminant analysis provides the classification ratio, which acts as an indicator of the goodness of fit of the original cluster assignments (Klecka, 1980). Therefore, the four identified franchise profiles: the beginners, the developers, the growers and the matures were evaluated against performance criteria. Financial performance, growth performance, market performance, and franchise reputation were selected as a key variable for measuring the derived cluster stability. These criteria were also used to justify the heterogeneity theory as it is claimed that company performance may differ as a result of the variant responses and strategies which are employed within the competitive environment (Cameron and Whetten, 1981; Lillis et al, 1976). These responses are also claimed to evolve with the degree of franchise system maturity (Oxenfelt and Kelly, 1968-69).

Franchise Stages and Franchise Performance

An effective criterion in measuring franchise performance should reflect both short-term and long-term factors (Zaheer et al, 1999). Short-term factors primarily relate to the financial benefits, which have a significant and immediate effect on franchise organisations. Long-term factors relate to survival, growth and reputation, which ensure the franchise operates effectively over a period of years (Reimann, 1982). It is reasonable to assume that short-term performance will naturally predict long-term franchise outcomes. For example, the implementation of successful franchise strategies which creates short-term performance and improves outlet profits will induce growth and enhance market shares of franchise organisations. Further, the short-term performance of maintaining and improving the franchise brand name and trademark also enhances the franchise reputation in the long run. However, in some cases, short-term and long-term performances are not always aligned. Franchise organisations that focus too much on short-term effects can sometimes jeopardise organisational long-term effectiveness (Connors, 1999; Jaworski, 1988). In fact, a short-term strategy that appears ineffective may actually be a part of an effective long-term strategy (Cameron and Whetten, 1983; Cannolly et al, 1980; Venkatraman and Ramanujam, 1986). Therefore, franchise organisations must carefully measure and assess the consequences of their efforts to achieve the performance of both short and long-term time frames.

Despite a lack of franchising studies that empirically examine the associated links between franchise heterogeneity and franchise short-term and long-term performance, other organisational studies have revealed a significant relationship between developmental stages and a set of dimensional performance outcomes. Cameron and Whetten (1981) used 18 simulated organisations to demonstrate how the criteria for effectiveness and the degree of performance changed as organisations went through different stages. Quinn and Rohrbaugh (1983) evaluated four multiple dimensions of effectiveness across four organisational life cycle stages, focusing particularly on entrepreneurship, collectivity, formalisation and control, structure elaboration and adaptation. Some related franchising studies also found that the more mature the franchisors, the higher the chance of survival rate and performance (Bates, 1998; Castrogiovanni et al, 1993; Lafontaine and Shaw, 1998). It is also likely that franchise organisations that are at a specific stage of development would reflect similar levels of resources and expertise when compared to others in their groups. These franchise firms would also employ similar strategies to other representatives of their cohort (Carney and Gedajlovic, 1991). It is likely that different levels of short-term and long-term franchise performance will be found at different stages of franchise development (Lillis et al, 1976; Smith et al, 1985). Hence we argue that the heterogeneous profiles of franchising which reflect similar profiles in the same cohorts would initiate different levels of performance. Linking the multiple dimension performance criteria to the heterogeneous profile of franchising, franchise stages would initiate different levels of franchise financial performance, growth and market performance, and franchise reputation. This leads to our hypothesis:

[H.sub.2]: Franchise stages reflect different levels of franchise performance.

Instrument-Performance Measure

A set of performance indicators companying sales growth, profit before tax and return on investment was used to measure franchise financial performance. Reputation was measured in terms of an assessment of the positive image of the franchise organisation for potential franchisees. These sets of performance indicators were selected as it is likely to influence the firms' sustainability and growth. In response to the financial, market position, and growth measures, participants were asked to make subjective evaluations of their performance in comparison with other, similar, competitors. Subjective evaluation allows the respondents to estimate their own performance when compared to their competitors who were estimated to be at the same age, size, and stage of development (Chandler and Hanks, 1993). The subjective evaluation of performance is preferred because franchise objective information on performance is difficult to obtain or is not publicly available. This method is supported as an effective substitute method of measuring performance (for example, Chandler and Hanks, 1993; Dess and Robinson, 1984; Reimann, 1982; Sapienza et al, 1988). These comparisons with competing businesses were based on a seven-point Likert scale ranging from 1, 'much worse than the others' to 7, 'outstanding performance compared to the other'. Reputation was measured in relation to the impact of the franchise brand name on potential franchisees and the public. A seven-point scale was used, anchored at 1, with 'strongly disagree', to 7, 'strongly agree'. The aggregated means of individual items were averaged to develop composite scores for financial performance, growth, market performance and reputation. The reliability coefficients for each of the four assessments were at acceptable levels ([alpha]'s [greater than or equal to]0.69).

Analysis

The second hypothesis proposed that franchise firms in different stages demonstrate different levels of success. Discriminant analysis was used to test hypothesis 2. Table 3 reports the standardised scores of the discriminant functions, the function correlations, along with the group centroids on the four franchise stages. Out of the three discriminant functions, two discriminant functions were found to be statistically significant.

The first function produced a significant effect, ([chi square]= 30.31, d.f. = 12, p < 0.01), accounting for 56 per cent of the between-group variability. After removal of the first function, the second function still remained statistically significant, ([chi square] = 12.76, dr. = 6, p < 0.05), contributing another 41 per cent to the explained variance. The two functions, in total, accounted for 97 per cent of the between-group variability.

The results revealed support for hypothesis 2. The two significant functions were interpreted by examining the discriminant standardised coefficients, the function correlations and the function group centroids. The first discriminant function revealed a highly significant correlation on the market share variable (r = 0.79, p < 0.05), indicating that the first function was separating the franchise groups based on their market share performance. The second function was significantly associated with financial performance (r = 0.68, p < 0.05), franchise reputation (r = 0.63, p < 0.05) and growth performance (r = 0.59, p < 0.05), reflecting the statistical differences in the three performance dimensions in this function. Overall results provide support for Hypothesis 2, suggesting that the four defined franchise groups differed in their level of franchise performance.

Further differentiation of the four franchise groups is possible by examining the differences between group means on each discriminating variable. Table 4 reports the results of the one-way Analysis of Variance using the Dunnett T3. Multiple Comparison Post Hoc procedure. All discriminating variables were statistically different among the four identified franchise groups.

The results of the first discriminant function revealed that the growers had the highest network growth rate among the groups (mean = 5.11, p < 0.05), followed by the beginners (mean = 4.58). As the scores of the group centroids for the developers (-0.16) and the matures (-2.17) were negative, the growth rates for the developers and the matures were actually decreasing. The matures had the lowest growth, followed by the developers. For the second discriminant function, the growers and the developers were positively related to the three performance discriminant variables. The growers were performing the best in terms of their financial performance (mean = 3.20), had the significantly largest market share compared to the matures (mean = 5.25) and were the most recognised brand name or trademark (mean = 5.14) among the dominant groups, followed by the developers. In contrast, the matures were found to have the lowest financial performance, were losing the most market share and were building the least reputation relative to the beginners. Overall, the growers were found to be the best performers among the groups.

A further step in examining the validity of the discriminant function is through a construction of a classification matrix. Classification matrices provide an assessment of the discriminating power of the function by revealing how well the function classifies the units (Klecka, 1980). The classification analysis indicated approximately 61.5 per cent of the beginners cases, 68.6 per cent of the developers cases, 21.4 per cent of the growers cases, and 33.3 per cent of the matures cases could be classified correctly by the discriminant functions, resulting in an overall hit ratio of 57.1 per cent. It is suggested that the classification accuracy should be greater than that expected by chance alone. To establish a minimum acceptable hit ratio, the proportional chance criterion may be used where the sample sizes between the groups are uneven (Hair et al, 1998). The proportion chance criterion for the sample of the study is 35.6 per cent. Considering the hit ratio of 57.1 per cent, the classification accuracy for the four franchise groups meets this criterion, suggesting that the model has an acceptable level of explanatory power.

Results and Discussion

The findings from the clustering procedure provide support for hypotheesis 1: Franchise organisations were found to be heterogeneous and could be grouped on the basis of their characteristic profiles. The four distinct profiles were labelled as the beginners, the developers, the growers, and the matures. The second hypothesis which proposed that franchisors at different stages of development demonstrate different levels of performance outcomes is also supported (see Figure 1). The four franchise profiles are explained in detail as follows:

[FIGURE 1 OMITTED]

Type I The Beginners

Forty-three per cent of the samples were categorised as the beginners. Profiles of the beginners parallel the penetration stage of Lillis et al (1976). The beginners are young and small. This franchise group had the lowest number of franchise established outlets (mean = 1.70 outlets) and the lowest franchisee entry requirement fee (A$41,000; A$1 is about S$1.25 or US$0.74). The low price strategy may be caused by less established market values or less brand name recognition as perceived by the potential franchisees (Fombrun and Shanley, 1990). The strategy of rapid franchising encourages a low fee structure to build up their franchise system (Floyd and Fenwick, 1999; Oxenfelt and Kelly, 1968-69). In other words, the beginners were still establishing their franchising identity, with the low establishment fees employed as a strategy for growth.

Additional franchises are a major source of capital for newly established franchise firms. The percentage of total franchised outlets for the beginners was the highest across the four stages (80 per cent). The analyses suggest that once the firms begin to franchise, expansion through company owned units decrease. This may imply that firms at the beginning stage have expanded exclusively through franchising, or they have converted some of their company-owned outlets to franchise outlets to increase their capital resources. The findings confirm past studies which identify resource attainment as an important factor in the early stage of development (Combs and Ketchen, 1999; Dodge et al, 1994). However, the results contradict previous work by Lillis et al (1976), which placed less importance on capital gains at the penetration stage. Lillis et al argued that although the resource attainment was an important reason for firms engaging in franchising, the benefits from effective monitoring of franchisee behaviours were far superior to resource attainment. However, the results of Lillis et al (1976) are of limited value since their franchise stages were conceptually classified based on only two dimensions: firm age and size.

The beginners tended to limit their outlet expansion to the inception area, with the majority of franchise outlets (64 per cent) located in the place of inception. It could be argued that these firms favour local expansion due to the higher business concept recognition in their home province, less established infrastructure of newly franchised firms, or lack of market knowledge in the new market territory (Dodge et al, 1994). It would also be easier to monitor the activities of the franchises. These findings contradict previous studies which argued that firms geographically franchise to reduce the administrative costs of monitoring (Brickley and Dark, 1987; Norton, 1988b). However, the contradictory results may be attributed to the fact that this prior research considered franchising to be a homogeneous industry.

The beginner profiles also related significantly to their performance outcomes. Firm size was positively associated with growth orientation (r = 0.77, p < 0.01; r = 0.80, p < 0.01), vertical integration (r = 0.37, p < 0.05), firm age (r = 0.33, p < 0.05) and reputation (r = 0.39, p < 0.05), indicating that the beginners increased in size as they expanded into more franchise outlets and gained higher brand name recognition after the establishment of the first franchise. Negative associations were found between size and dispersion (r = -0.34, p < 0.05); reputation and dispersion (r = -0.42, p < 0.01); and growth orientation and age (r = -0.37, p < 0.05). The beginners increased their size by expanding outside their home province. This was normally achieved as higher brand name recognition inter-state was accomplished. However, growth in the franchising section is a challenging strategy for beginners. Between the period of first inception and first franchise (six years), the beginners were mostly focused on the growth of their own branches.

The results also displayed significant relationships between the market share performance, financial performance and network growth rate. Compared with other franchise groups, the beginners had the highest market share but a relatively low network growth rate, low financial performance, and low reputation. As the firms had just advanced into franchising, their brand recognition was perceived by potential franchisees as being relatively low. The growth orientation of the firms was also unable to match their network growth rate. The beginners chose a rapid growth orientation, however, the implemented strategy was not quite successful, with a reverse effect between growth orientation and network growth rate. As a result, the beginners were the second poorest performers in network growth and financial performance compared with the other franchise groups.

Type 2 The Developers

Thirty-eight per cent of the participating franchisors were categorised as deve-lopers. On average, the developers operated 73 retail outlets and charged moderately high fees (A$164,000), which suggests a better franchise format establishment and an increasing franchise reputation (Baucus et al, 1993). However, these firms were likely to reduce their fees as their size increased (r = -0.36, p < 0.05), reflecting their need to attract more candidates and, ultimately, more resources to develop the franchise systems (Carney and Gedajlovic, 1991). Still in the developing stage, few had achieved a nation-wide expansion, with only 39 per cent of their franchise outlets located out of the home province.

The growth strategy of the developers had a geographical emphasis. Firm size was negatively related to dispersion (r = -0.52, p < 0.01) and average capital requirement (r = -0.36, p < 0.05) and dispersion was negatively related to firm age (r = -0.39, p < 0.05) and growth orientation (r = -0.42, p < 0.05) reflecting their policy of early expansion on a competitive pricing out of their home province (r = -0.50, p < 0.05). The developers adopted early geographical expansion reflecting the need to build up their system formats (Julian and Castrogiovanni, 1995). In this stage the results show that geographical expansion included both company and franchise outlets, possibly resulting from an increasing build up of infrastructure, or from transferring knowledge between the company-owned outlets and the franchise outlets (Dodge et al, 1994). However, the degree of geographical expansion was only slightly higher when compared to the first stage (D = 3 per cent).

There was also a negative relationship between the developers' profiles and their market performance. Firm size was negatively related to market performance (r = -0.44, p < 0.05), suggesting that the developers were unable to increase their market share in parallel with an increase in size. In addition, the growth policy could not enhance the market performance (r = -0.39, p < 0.05), thus, the return on outlet expansion did not correspond to an increasing rate of market share. This scenario may suggest that the developers preferred to convert their existing outlets to franchised outlets as a means to growth. It is likely that developers employ a conservative approach to expansion.

Even though the growth orientation of the developers was unable to match their network growth rate, they were the second highest growth performers, and ranked second in the financial performance and reputation. The network growth rate of a company's franchising section and company outlets were also significantly related. Franchise outlet growth increased significantly in parallel with the growth rate of the company branches, with the growth rate in franchise outlets being slightly higher. The results confirmed the findings of Castrogiovanni and Justis (2002) suggesting that growth in franchising activity exerted greater influence on network growth. However, the developers seemed to have lower market share, ranking second last in market-share performance.

Type 3 The Growers

Only 15 per cent of the samples were categorised as the growers. The growers had spent a substantial time in the franchise business (mean = 19 years). Perhaps, the growers had more fully developed systems and increasing franchise brand name recognition, and thus could afford an expeditious growth. On average the growers had opened seven outlets per year since their inception and 6.55 outlets per year since they began to franchise. The increased rate of the new outlets and the number of franchised outlets was significantly higher than the beginners' new outlets (1.7 outlets) and the beginners' franchised outlets (2.05 outlets). The growers were the largest franchise network (n = 134), however, the average capital requirement of the growers was significantly lower than the matures ([DELTA]A$694,000, p < 0.001) (see Stage 4). With a successful market formula, proper infrastructure, effective monitoring systems, and nation-wide brand name recognition, the expansion strategy of the growers tended to go beyond the boundaries of their local region (Julian and Castrogiovanni, 1995). The geographical dispersion of the growers was significantly higher than the other groups, with only 37 per cent of franchise outlets located in their home province. This policy of nation-wide expansion also makes the growers the most geographically dispersed.

The growers may use their franchise entry fee as a material determinant of the network growth rate. Average capital requirement of the growers was strongly and negatively associated with franchise size (r = -0.50, p < 0.05), company outlet growth (r = -0.54, p < 0.05) and franchise unit expansion (r = -0.55, p < 0.05). Although the entry fee was relatively high (A$306,430), the growers were likely to reduce their franchise fees when they franchised more outlets. The propensity to demand a relative high initial fees may result from a more established franchise brand name, as interpreted by the potential franchisees (Lafontaine and Kaufmann, 1994).

The profile of the growers closely matched that of the "rapid growers" identified by Carney and Gedajlovic (1991). The growers had significantly increased both of their company owned and franchised outlets. However, the growth rate of the company branches was only slightly higher than that of their franchising agencies, reflecting less reliance on franchising in the growing firms (Caves and Murphy, 1976). This may imply that the rapid growth strategy of the growers was successfully implemented. The growers reported the highest financial performance, strongest network growth rate and highest reputation when compared with the other groups. With all four performance outcomes significantly related, the growers ensured strong financial performance as they successfully achieved their network growth rate and market share performance. A significant relationship between the financial performance and the reputation of the growers also indicates they are most likely to gain their financial benefit from a wide recognition of the franchise concept.

Type 4 The Matures

The results relating to the matures are somewhat tenuous, since the sub-sample was very small (n = 3). However, some trends can be tentatively identified. The matures attained a conservative value of retail expansion by selling their new franchises at an extremely high price (AS 1,000,000). Perhaps the attentive growth policy of the mature firms reflects their prolonged policy of reputation and trademark rather than resource attainment. The matures are costly and select the franchisee candidates who can make a substantial investment and can contribute to the image of the chain. If that kind of investor is not available, the matures prefer to own and operate their retail outlets (Carney and Gedajlovic, 1991). This is reflected in the high percentage of company-owned outlets (57 per cent). This lower emphasis on capital resources of the matures concurs with the late maturity stage of Lillis et al (1976).

Geographical dispersion of the matures was also less, with approximately 50 per cent of franchise outlets operating in their home province. The results also showed a significant reduction of outlets located outside the place of inception, when compared with the growers. These findings parallel past research, indicating an insignificant relationship between firms' age and firms' geographic expansion strategy (Julian and Castrogiovanni, 1995). The matures may emphasise a local growth policy to ensure efficient monitoring by either opening up new franchise outlets close to their headquarters or refusing to renew a franchise contract on expired distant outlets and replacing them with owned outlets.

The matures are the worst performers among the four stages. The analysis revealed that the matures were losing their market share and earning fewer economic benefits. They also reported negative network growth rate and lower popularity of their franchise concepts as interpreted by the franchisee candidates. This may be due to the policy of placing less reliance on franchising by employing extremely high franchise entry fees, and also devoting their attention to the operations of the company branches. Perhaps the popularity of the business concept was also declining.

Theoretical Implication

The findings of this research offer detailed insight into the explanation of the franchising phenomenon. The results of the heterogeneous franchising reveal that neither resource scarcity theory nor agency theory sufficiently explained the creation of franchising. The combined explanation from both theories is likely to be more effective (Combs and Ketchen, 1999; Lafontaine and Kaufmann, 1994). The resource scarcity theory argues that firms franchise as a means of obtaining capital and easing managerial constraints upon their network growth, thereby transferring risk from the firm to franchisees (Oxenfelt and Kelly, 1968-69). This is evident--the age of franchises was found to determine the franchise entry fee (Baucus et al, 1993), with young, small firms significantly increasing the number of franchise outlets (Caves and Murphy, 1976; Oxenfelt and Kelly, 1968-69). On the other hand, the results do not confirm some aspects of the resource scarcity theory. Contradicting the past literature, the type of franchise system did not determine the franchise entrant fee (Baucus et al, 1993), and the relationship between farms' size and number of franchise outlets was not linearly related (Combs and Ketchen, 1999). The results suggest that while franchising may be used as a strategic tool for organisations to increase their capital or intellectual resources, the strategic evolution of franchising firms may be caused by both capital gain and efforts to improve performance outcomes. In addition, agency theory predicts that as a response to the problems between the employer and employee relationship, franchising is used to minimise the monitoring cost of the firms (Rubin, 1978). The franchise heterogeneous profiles can be partially explained from this perspective. Consistent with the findings of Julian and Castrogiovanni (1995), the larger franchisors sought expansion in more geographic areas than the smaller franchisors. The results also highlighted a relationship between the franchise age and the degree of geographical dispersion: older franchise firms sought expansion in more geographic areas than younger franchise firms. The findings may imply that larger and younger firms engaged in franchise businesses in the hope of reducing their agency deficiency and also to ease the administrative costs of monitoring.

Clearly, the configurations of the franchise organisational profiles seem appropriate in predicting the heterogeneous characteristics of the franchise industry. Moreover, it is likely that the contradictory findings between the resource scarcity (for example, Caves and Murphy, 1976; Oxenfelt and Kelly, 1968-69; Shane and Spell, 1998) and the agent theory explanation (for example, Brickley and Dark, 1987; Lafontaine and Kaufmann, 1994; Rubin, 1978) from the past studies were the consequence of the heterogeneous data set. Researchers need to be more attentive to constructing the sampling framework and to analysing the franchising data in order to engage in vigorous inquiry into the franchising phenomenon. This implies that results from the franchising investigation will be more robust when samples are obtained from franchise systems with similar characteristics and profiles or from franchises belonging to the same system, (for example, samples of the franchisees from the same franchisor). Further, in the cross-sectional study of the franchise firms, factors that relate to the sample heterogeneity of franchising such as company size, age, degree of dispersion, cost of franchising, and vertical strategy need to be controlled in the model of analysis to control for heterogeneous factors. These findings also imply that the franchising industry cannot be treated as homogenous, even though the implications surrounding the creation of the franchising industry (contractual agreement, performance-based incentive, agency problems, strategic alliances) are distinctive. Franchise firms are clearly connected with other specific industries (for example, food and beverage, automobile, accommodation, and education). Thus, the implications in franchising exploration should also include the effects from the related industries.

Implications for Practice

The classification of the franchise firms into four groups may offer a valid explanation for and prediction of the movement in the franchising industry. The predicted trends of franchise stages are likely to benefit the franchisors in the sense that these trends can be used as guidelines to achieve sound economic performance. By comparing their firms' profiles against each franchise stage, the franchisors could estimate and predict their firms' position within the franchising industry. For instance, franchise firms that fall within the groups of the beginner or the developer will be able to reap the benefits of high financial performance, high reputation and high growth. However, in order to achieve peak performance, franchise firms should aim toward grower status by focusing on geographical expansion, opening up more companies, setting a moderately high entry fee for franchise outlets and/or reducing the duration of the transition between full company ownership to part franchise ownership.

For firms in the mature groups, there are further options to remain competitive. Franchise firms that fit into the profile of the matures can either reduce their franchise entry fee to attract more franchise candidates or open up more franchise outlets to rebuild their reputation and growth performance. If the mature firms do not redirect their franchise strategy, they may stagnate and their performance may diminish further. An awareness of these performance indicators can assist the franchisors in maintaining a strong awareness of viability.

The heterogeneous profiles of the franchise groups may offer valuable guidance for franchisee candidates. Although this process may not identify all the success characteristics in one franchise system, it offers, more or less, some guidelines to recognise certain characteristics and management orientations of the potential successful franchise firms. The potential franchisees may observe the strategic direction of their intended franchisors and compare their profiles with each franchise category. Usually the franchisors' profiles are made publicly available either on their company websites or from franchise information centres. Using this information, the potential franchisees will have a rough idea of their interested franchisors, whether they are the beginners, the developers, the growers or the matures.

Potential franchisees who are searching for the right franchise business are well advised to look for the growers. Although the capital requirement of the growers is relatively high, the franchisees can at least be assured that the growers perform well and may bring sufficient returns in the future. The Growers also display a better reputation, which is believed to be the key competitive advantage of franchise firms. The second best option is to select the developers because they are the second best in terms of their performance and their price for new franchises is also reasonable. Potential franchisees with a low start-up capital might consider the developers as offering high potential at a modest price.

On the current evidence, the potential franchisees should try to avoid the matures. Franchisors in this group may look attractive because they have been in the system longer, are publicly well-known and demand high prices for their new start-up outlets. However, while the most expensive franchises they are relatively poor performers comparative to their performance in earlier years. Perhaps their systems are more stagnant, as reflected in their tendency to redirect their franchise ownerships. Potential franchisees who are looking for long-term relationships should avoid the matures because they have a tendency not to renew an expired contract or a tendency to buy back franchise outlets in order to convert them into their owned company outlets.

Limitations

Some caution should be taken in generalising the results of this study. First, the term "franchise stages" is given based on the three dimensions of the classification profiles of size, timing, and strategy orientation of the samples. Franchise stages is the term used to identify the group characteristics which include the sequential timeline. The term "stages" in this study may suggest the exponential movement of the franchise groups in term of timing rather than the linear movement. This finding concurs with the exponential movement in patterns of organisational life cycle (Agarwal, 1997; Cameron and Whetten, 1981; Castrogiovanni and Justis, 2002). Nevertheless, further research is suggested to verify the differences between these two patterns of organisational life cycle. Second, the limited Australian franchisors sample resulted in small group sizes for each developmental stage, particularly, as for the matures (n = 3). Although, the franchise stages were confirmed as to their differences on the profiles and performance using discriminant analysis, the small number of franchisors in the derived group may still reduce the statistical power in explaining the heterogeneous nature of the franchising industry. Thus, the generalisability of the results relating to developmental strategies remains tentative. Third, it is recommended that when using discriminant analysis, the data should be split into two sets to validate the difference between groups (Hair et al, 1998). With the small franchisor sample size, it was inadvisable to divide the data into two sets as it would further reduce the statistical power. Thus, the results of the performance differential among franchise stages were treated as valid without further confirmation. However, the cross investigation into the group profiles using clustering procedure does confirm and validate the franchise groups. Lastly, there may be a number of parameters that can be included in performing a configuration procedure (for example, contractual duration). Nevertheless, the numbers and the dimensions of franchise organisational parameters are deemed to be sufficient and appropriate in covering the basic function of franchise operations.

Conclusion

The findings provide strong empirical evidence concerning franchise heterogeneity theory. Further research on franchising should note the heterogeneous nature of franchises, as they differ greatly according to their cohorts. The prudent selection and planning in establishing the data set when conducting franchising studies is the area which needs more careful attention. This study also lays the foundation for the research on the relationship between franchise stages and franchise performance. The study confirms that franchise firms do employ different strategies, in accordance with their profiles, resulting in different levels of performance. This needs to be recognised and further tested in the field across other samples.

References

Agarwal R, 1997. "Survival of Finns over the Product Life Cycle". Southern Economic Journal, Vol 63 No 3, pp 571-593.

Agarwal S and S Ramaswami, 1993. "Marketing Controls and Employee Responses: The Moderating Role of Task Characteristics. Journal of the Academy of Marketing Science 21, No 4, pp 293-306.

Aldenderfer MS and RK Blashfield, 1984. Cluster Analysis. Newbury Park, CA, Sage.

Alon I and D McKee, 1999. "Towards a Macro Environmental Model of International Franchising". Multinational Business Review, Vol 7 No 1, pp 76-82.

Bates T, 1998. "Survival Patterns Among Newcomers to Franchising". Journal of Business Venturing, Vol 13, pp 113-130.

Baucus D, M Baucus and S Human, 1993. "Choosing a Franchise: How Base Fees and Royalties Relate to the Value of the Franchise". Journal of Small Business Management, Vol 31, pp 91-104.

Brickley JA and FH Dark, 1987. "The Choice of Organisational Form: The Case of Franchising". Journal of Financial Economics, Vol 18, pp 401-420.

Bronson JW and CP Morgan, 1998. "The Role of Scale in Franchise Success: Evidence from the Travel Industry". Journal of Small Business Management, Vol 36 No 4, pp 33-42.

Callum F and F Graham, 1999. "Towards a Model of Franchise System Development". International Small Business Journal, Vol 17 No 4, pp 32-48.

Cameron KS and DA Whetten (eds), 1983. Organisational Effectiveness: A Comparison of Multiple Models, Academic Press, International.

--, 1981. "Perceptions of Organisational Effectiveness over Organisational Life Cycles". Administrative Science Quarterly, Vol 26 No 4, pp 525-544.

Cannolly T, EJ Conlon, and SJ Deutsch, 1980. "Organisational Effectiveness: A Multiple-Constituency Approach". Academy of Management Review, Vol 5 No 2, pp 211-217.

Carman JM and TA Klien, 1986. "Power, Property, and Performance in Franchising: A Review and Integration of Theory". Research in Marketing, Vol 8, pp 71-130.

Carney M and E Gedajlovic, 1991. "Vertical Integration in Franchise Systems: Agency Theory and Resource Explanations". Strategic Management Journal, Vol 12 No 8, pp 607-629.

Castrogiovanni GJ, N Bennett and JG Combs, 1995. "Franchisor Types: Reexamination and Clarification". Journal of Small Business Management, Vol 33 No 1, pp 45-55.

Castrogiovanni GJ and RT Justis, 2002. "Strategic and Contextual Influences on Firm Growth: An Empirical Study of Franchisors". Journal of Small Business Management, Vol 40 No 2, pp 98-108.

Castrogiovanni GJ, RT Justis and SD Julian, 1993. "Franchise Failure Rates: An Assessment of Magnitude and Influencing Factors". Journal of Small Business Management, Vol 31 No 2, pp 105-114.

Caves RE and WFI Murphy, 1976. "Franchising: Firms, Markets and Intangible Assets". Southern Economic Journal, Vol 42, pp 572-586.

Chandler GN and SH Hanks, 1993. "Measuring the Performance of Emerging Businesses: A Validation Study". Journal of Business Venturing, Vol 8, pp 391-408.

Combs J and G Castrogiovanni, 1994. "Franchisor Strategy: A Proposed Model and Empirical Test of Franchise versus Company Ownership. Journal of Small Business Management, Vol 32, April, pp 37-48.

Combs JG and DJJ Ketchen, 1999. "Can Capital Sarcity Help Agency Theory Explain Franchising? Revisiting the Capital Scarcity Hypothesis". Academy of Management Journal, Vol 42 No 2, pp 196-207.

Connors J, 1999. "Maximising Effectiveness through Enterprise Performance Management Solutions". Business Credit 101, No 1, pp 64-65.

Dess GG and RB Robinson, 1984. "Measuring Organisational Performance in the Absence of Objective Measures: The Case of the Privately Held Firm and Conglomerate Business Unit". Strategic Management Journal, Vol 5 No 3, pp 265-272.

Dillon WR and M Goldstein, 1984. "Cluster Analysis", in Multivariate Analysis: Methods and Applications. John Wiley & Sons, New York, pp 157-208.

Dodge RH, S Fullerton, JE Robbins, 1994. "Stage of the Organisational Life Cycle and Competition as Mediators of Problem Perception for Small Businesses". Strategic Management Journal, Vol 15, pp 121-134.

Doherty AM and B Quinn, 1999. "International Retail Franchising: An Agency Theory Perspective". International Journal of Retail & Distribution Management, Vol 27 No 6, pp 224-236.

Everitt BS, 1993. "Cluster Analysis". Halsted, New York.

--, 1979. "Unresolved Problems in Cluster Analysis". Biometrics, Vol 35, pp 169-181.

Falbe C and D Welsh, 1998. "NAFTA and Franchising: A Comparison of Franchisor Perceptions of Characteristics Associated with Franchisee Success and Failure in Canada, Mexico, and the United States". Journal of Business Venturing, Vol 13, pp 151-171.

Floyd C and G Fenwick, 1999. "Towards a Model of Franchise System Development". International Small Business Journal, Vol 17 No 4, pp 32-48.

Fombrun C and M Shanley, 1990. "What's in a Name? Reputation Building and Corporate Strategy". Academy of Management Journal, Vol 33 No 2, pp 233-258.

Frazer L, C McCosker and J Paynter, 2000. "Comparing the Growth, Development and Maturing of Franchising in Australia and New Zealand". Small Enterprise Research, Vol 8 No 1, pp 14-25.

Hair JJ Jr, RE Anderson, RL Tatham, WC Black, 1998. Multivariate Data Analysis. Prentice-Hall International, Inc.

Howard D and D Hine, 1997. "The Population of Organisations Life Cycle (POLC): Implications for Small Business Assistance Programs". International Small Business Journal, Vol 15 No 3, pp 30-41.

Hunt SD, 1974. "The Trend toward Company Operated Units in Franchise Chains". Journal of Retailing, pp 110-119.

Jaworski B J, 1988. "Toward a Theory of Marketing Control: Environmental Context, Control Types and Consequences". Journal of Marketing, Vol 52, July, pp 23-39.

Julian SD and GJ Castrogiovanni, 1995. "Franchisor Geographic Expansion". Journal of Small Business Management, Vol 33 No 2, pp 1-11.

Klecka WR, 1980. "Discriminant Analysis" in Quantitative Applications in the Social Sciences, Vol 07-019. Sage, Newbury Park.

Krueger AB, 1991. "Ownership, Agency and Wages: An Examination of Franchising in the Fast Food Industry". The Quarterly Journal of Economics, Vol 106, February, pp 75-101.

Lafontaine F, 1992. "Agency Theory and Franchising: Some Empirical Results". Rand Journal of Economics, Vol 23 No 2, pp 263-283.

Lafontaine F and P Kaufmann, 1994. "The Evolution of Ownership Patterns in Franchise Systems". Journal of Retailing, Vol 70 No 2, pp 97-113.

Lafontaine F and K Shaw, 1998. "Franchising Growth and Franchisor Entry and Exit in the US Market: Myth and Reality". Journal of Business Venturing, Vol 13, pp 95-112.

Lassar WM and JL Kerr, 1996. "Strategy and Control in Supplier-Distributor Relationships: An Agency Perspective". Strategic Management Journal, Vol 17, pp 613-632.

Lillis CM, CL Narayana and JL Gilman, 1976. "Competitive Advantage Variation Over the Life Cycle of a Franchise". Journal of Marketing, Vol 40, October, pp 77-80.

Mathewson FG and RA Winter, 1985. "The Economics of Franchise Contracts". The Journal of Law and Economics, Vol 28, pp 503-526.

McCosker C and L Frazer, 1998. "Franchising Australia 1998: A Survey of Franchising Practices and Performance". University of Southern Queensland, p 43.

Meyer AD, AS Tsui and CR Hinings, 1993. "Configurational Approaches to Organisational Analysis". Academy of Management Journal, Vol 36, pp 1175-1195.

Miles G, CC Snow, MP Sharfman, 1993. "Industry Variety and Performance. Strategic Management Journal, Vol 14 No 3, pp 163-177.

Miles MP and JG Covin, 2000. "The Relationship between Environment Dynamism and Small Firm Structure, Strategy, and Performance". Journal of Marketing Theory and Practice, Spring, pp 63-74.

Miller NJ and TL Besser, 2000. "The Importance of Community Values in Small Business Strategy Formation: Evidence from Rural Iowa". Journal of Small Business Management, January, pp 68-85.

Norton SW, 1988a. "Franchising Brand Name Capital and the Entrepreneurial Capacity Problem". Strategic Management Journal, Vol 9, Summer, pp 105-114.

--, 1988b. "An Empirical Look at Franchising as an Organisational Form". Journal of Business, Vol 61 No 2, pp 197-218.

Oxenfelt AR and AO Kelly, 1968-1969. "Will Successful Systems Ultimately Become Wholly-Owned Chain?" Journal of Retailing, Vol 44 No 49, pp 69-83.

Punji G and SW David, 1983. "Cluster Analysis in Marketing Research: Review and Suggestions for Application". Journal of Marketing Research, Vol 20, May, pp 134-148.

Quinn RE and J Rohrbaugh, 1983. "A Spatial Model of Effectiveness Criteria: Towards a Competing Values Approach to Organisational Analysis". Management Science, Vol 29 No 3, pp 363-377.

Reimann BC, 1982. "Organisational Competence as a Predictor of Long Run Survival and Growth". Academy of Management Journal, Vol 25 No 2, pp 323-334.

Rubin PH, 1978. "The Theory of the Finn and the Structure of the Franchise Contract". The Journal of Law & Economics, Vol 12, pp 223-233.

Sapienza HJ, KG Smith, MJ Gannon, 1988. "Using Subjective Evaluations of Organisational Performance in Small Business Research". Entrepreneurship Theory and Practice, Vol 12 No 3, pp 45-52.

Shane S and C Spell, 1998. "Factors for New Franchise Success". Sloan Management Review, Spring, pp 43-50.

Shane SA, 1998. "Making New Franchise Systems Work". Strategic Management Journal, Vol 19, pp 697-707.

Smith KG, TR Mitchell and CE Summer, 1985. "Top Level Management Priorities in Different Stages of the Organisational Life Cycle". Academy of Management Journal, Vol 28 No 4, pp 799-820.

Venkatraman N and V Ramanujam, 1986. "Measurement of Business Performance in Strategy Research: A Comparison of Approaches". Academy of Management Review, Vol 11 No 4, pp 801-814.

Withane S, 1991. "Franchising and Franchisee Behaviour: An Examination of Opinions, Personal Characteristics, and Motives of Canadian Franchisee Entrepreneurs". Journal of Small Business Management, January, pp 22-29.

Zaheer S, S Albert and A Zaheer, 1999. "Time Scales and Organisational Theory". Academy of Management Review, Vol 24 No 4, p 725.

Dr Chutarat Inma Murdoch University

Professor Shelda Debowski Organisational and Staff Development Services University of Western Australia

Table 1: Heterogeneity Profiles in the Franchising Study

Study             Groups   Franchise Type           Dimensions
                           or Stages

Oxenfelt and      4        New                      Firm age
Kelly                      Rapidly growing          Firm size
(1968)                     Established &
                           Prosperous
                           Mature & declining

Lillis et al,     4        Penetration              Firm age
1976                       Growth                   Firm size
                           Maturity
                           Late maturity

Carney and        5        Rapid grower             Size
Gedajlovic,                Expensive conservative   Dispersion
1991                       Converters               Growth
                           Matures                  Pricing
                           Unsuccessful             Contractual
                                                    provision
                                                    Vertical
                                                    integration
                                                    Timing

Castrogiavanni,   5        Rapid grower             Size
Bennett and                Expensive conservative   Dispersion
Combs (1995)               Converters               Growth
                           Matures                  Pricing
                           Unsuccessful             Contractual
                                                    provision
                                                    Vertical
                                                    integration
                                                    Timing

Floyd and         3        Hatching                 Year
Fenwick                                             franchising
(1999)                     Nesting                  Size
                                                    Years before
                           Fledgling                franchising

Study             Note

Oxenfelt and      Conceptually
Kelly             constructed
(1968)

Lillis et al,     Empirically
1976              tested

Carney and        Empirically
Gedajlovic,       tested
1991

Castrogiavanni,   Validate the
Bennett and       heterogeneity
Combs (1995)      profiles
                  suggested by
                  (Carney and
                  Gedajlovic,
                  1991

Floyd and         Exploratory
Fenwick           study with 10
(1999)            franchisors

Table 2: Characteristics of Four Franchise Groups Derived from Cluster
Analysis (n=91)

                                   Franchise Groups

                            Cluster 1   Cluster 2    Cluster 3
Variables                   Beginners   Developers    Growers
(Unit of measurement)       (n = 39)     (n = 35)    (n = 14)

Number of retail              22.69       73.09       133.86
outlets (Outlets)            (20.51)     (94.36)      (61.65)

Percentage of                 63.77       60.89        37.43
franchised outlets           (32.98)     (31.57)      (29.79)
located in the place
of inception (per cent)

Number of outlets              1.70        3.92         7.01
established per year          (1.66)      (5.26)       (3.78)
since inception (Outlets)

Franchise outlets              2.05        5.74         6.55
established per year          (1.80)     (10.28)       (3.95)
since inception (Outlets)

Average capital               41.06      163.58       306.43
requirement (A$ 000)         (22.31)     (59.28)      (63.20)

Percentage of total           87.56       82.77        81.22
outlets franchised           (19.99)     (22.11)      (24.46)
(per cent)

Year since inception          17.38       21.51        22.36
(years)                      (14.06)     (16.59)      (13.18)

Year since first              11.54       12.40        19.36
franchise (Years)             (7.85)      (7.87)      (12.66)

Year between inception         5.85        9.20         3.00
and first franchise           (8.14)     (15.84)       (3.01)
(Years)

                            Franchise Groups

                            Cluster 4
Variables                    Matures      Mean
(Unit of measurement)        (n = 3)      (sd)

Number of retail              97.00       61.63
outlets (Outlets)           (105.16)     (76.79)

Percentage of                 52.95       58.25
franchised outlets           (41.42)     (32.96)
located in the place
of inception (per cent)

Number of outlets              3.28        3.42
established per year          (2.95)      (4.15)
since inception (Outlets)

Franchise outlets              2.34        4.17
established per year          (2.82)      (6.90)
since inception (Outlets)

Average capital             1000.00      160.62
requirement (A$ 000)          (0.00)    (187.36)

Percentage of total           73.10       84.27
outlets franchised           (42.54)     (22.14)
(per cent)

Year since inception          30.67       20.18
(years)                      (9.24)      (14.93)

Year since first              19.67       13.34
franchise (Years)            (14.84)      (9.28)

Year between inception        11.00        6.87
and first franchise          (15.70)     (11.62)
(Years)

                            Franchise Groups

Variables                     Anova      Dunnett
(Unit of measurement)        F-tests     P < 0.05

Number of retail             10.44 **    3>1; 2>1
outlets (Outlets)

Percentage of                 2.60 *      1>3
franchised outlets
located in the place
of inception (per cent)

Number of outlets             7.07 ***    3>1
established per year
since inception (Outlets)

Franchise outlets             2.59 *      3>1
established per year
since inception (Outlets)

Average capital             452.46 ***   4> All
requirement (A$ 000)                     others

Percentage of total           0.67        n.s.
outlets franchised
(per cent)

Year since inception          1.15        n.s.
(years)

Year since first              3.26 *       n.s.
franchise (Years)

Year between inception        1.22        n.s.
and first franchise
(Years)

Multivariate test of significance (Wilks): F (4, 86) = 18.72 ***

Note: * p < .05; ** p < .01; *** p < 0.001
Means are reported. Standard deviation is in parentheses

Table 3: Results of Discriminant Analysis Predicting
Performance on Franchise Group

               Function 1                   Function 2

               Standardised   Function      Standardised   Function
Variables      Coefficient    Correlation   Coefficient    Correlation

Finance        0.300          0.556         0.485          0.677 *
Growth         0.133          0.574         0.495          0.591 *
Market Share   0.591          0.785 *       -0.256         0.085
Reputation     -0.524         -0.558        0.633          0.632 *

Wilks  [chi square] = 30.31 **, (d.f = 12)  [chi square] = 12.76 **,
(d.f = 6)

                  Anova       Group Centroids

               Univariate
Variables        F(4, 88)     Cluster       F1             F2

Finance           3.82 *      Beginners     0.245          -0.368
Growth            3.45 *      Developers    -0.157         2.00
Market Share      3.77 *      Growers       0.172          0.629
Reputation        3.63 *      Matures       -2.17          -0.479

Note: p < 0.05; ** p < 0.01.

Means are reported. Standard deviation is in parentheses

Table 4: Differentiating Performance Profiles of Four Franchise Groups
(n = 91 )

Performance     Cluster 1     Cluster 2     Cluster 3   Cluster 4
Dimensions     (Beginners)   (Developers)   (Growers)   (Matures)

Finance           2.87           2.99         3.20        2.08
                 (0.59)         (0.49)       (0.52)      (1.81)

Growth            4.58           4.69         5.11        3.60
                 (0.76)         (1.12)       (1.17)      (0.28)

Market share      5.15           4.83         5.25        3.50
                 (0.80)         (0.92)       (1.20)      (0.50)

Reputation        4.32           4.90         5.14        5.80
                 (1.27)         (1.08)       (0.74)      (0.29)

Wilks Lambda                 F(4, 86) = 2.49 **

Performance       Anova       Dunnett T3
Dimensions       F(4,91)       P < 0.05

Finance          3.82 *      3 > All others

Growth           3.45 *      3 > All others

Market share     3.77 *      3 > 4

Reputation       3.63 *      4 > All others

Wilks Lambda

Note: * p < 0.05; ** p < 0.01. Means are reported. Standard
deviation is in parentheses