DATA AND METHODOLOGY
The longitudinal analysis uses triangulated secondary, company-specific data. Sources of the data include the company's Uniform Franchise Offering Circular (UFOC), corporate 10K Reports, Entrepreneur "Franchise 500" (1987-1994) listings and direct contact with public franchisor shareholder relations offices. One hundred forty companies were identified as having their primary source of income in franchising, company operations of a business-format franchise, and where ownership was at least partially public. Of the 140 companies, data could only be collected on 83 companies for which at least two consistent sources of information existed. Time and resource constraints precluded further data collection on the remaining companies1. For each public franchisor, monthly total returns (dividends distributed and capital gains) were collected for each month between January 1987 and December 1994 from the CRSP tapes and Compustat. Similar returns variables were collected for a market index, in this case the S&P500 Index. These returns were used to estimate the systematic risk (beta) of such firms as well as the abnormal monthly returns (Jensen's Alpha). Independent variables captured include royalty rate, franchise fee, contractually obligated advertising fee, number of company-owned and franchised outlets, whether the company was venture capital backed, years in business and years in franchising and the rate of growth of outlets over the period 1987 through 1994.
Franchisor Index Performance
Exhibit 1 plots the performance of the Standard and Poors 500 Index versus an equally-weighted index of 83 publicly listed franchisors between January 1st 1987 and December 31st 1994. Both indices were set at a base of 100 on January 1st 1987 to provide a comparable starting point2. It is interesting to note that the franchisor index very much tracks the largest market index, with two significant breaks, one in early 1992 and a second in late 1992. These discontinuities can be traced back to the addition to the index of a number of franchisor firms with very strong early post-IPO performance in the later part of 1992. In order to gauge the return performance, it is essential to correct for the level of risk inherent in these franchisor companies. The distribution of systematic risk3 in the sample firms is presented in Exhibit 2. The mean beta is equal to 1.152 and is not statistically different from 1.00 at the usual levels of confidence. The median beta is 1.008, with a variance of 0.770.
Overall, franchisor companies seem to have offered returns very much in line with the market as a whole and with very similar average risk levels. They do not exhibit abnormal returns with respect to standard models of the risk-return process, as indicated in Exhibit 3 which graphs the distribution of Jensen's alpha, the stocks' average abnormal performance per month. This evidence does not support the view that franchisors constitute a very specific group in terms of the risk-performance relationship: quite to the contrary, they seem to offer the same general distribution of characteristics than the market as a whole.
System Growth, Franchise Format and Contractual Terms of Agreement
A basic hypothesis of this paper is that franchising offers a very powerful system that helps deliver fast growth to companies with promising concepts. The first hypothesis to be tested in this respect is that growth can be accelerated by removing the financing constraints inherent in corporate-based development and instead calling upon a franchisee-owned growth. Similarly, franchise contracts which tend to be more generous towards the franchisees, i.e. those characterized by lower upfront franchise fees, lower royalties, lower advertising fees and longer contractual terms, should also facilitate the growth in number of outlets. To test these relationships, regressions are run to explain the rate of growth in total outlets over the period 1987-1994 (RGROWTH), using as explanatory variables the percent of outlets which are company owned4 (PERCTOWN), the total number of outlets in the system5 (TOTOUTS), the franchise fees (FFEE), royalty payments (ROYALTY), advertising fees (ADFEE) and contractual terms (TERM) of the franchise agreement.
A first check is run in Exhibit 4 for the possibility of multicolinerarity in the explanatory variables, an issue that could hinder the interpretation of the regression coefficients. The Pearson correlation coefficients between pairs of variables remain acceptable in all instances.
The regression results are summarized in Exhibit 5. A number of interesting insights emerge. First of all, the percentage of outlets which are directly owned by the franchisor appears to be positively correlated with the rate of growth at the 5% level of confidence but the coefficient is positive, implying that growth in the number of outlets is actually positively correlated with company-financed outlets. Put differently, the fastest rates of outlet growth appear to be associated with franchisor-financed stores, thus not supporting the hypothesis that growth would be best served by calling upon external capital. Second, the coefficients on franchise fees and royalty are as expected (that is negative, meaning that higher fees tend to be associated with slower growth), as well as the coefficient on advertising fees (positive, to the effect that higher advertising fees, and hopefully expenses, are conducive to higher growth). Finally, the coefficient on total number of outlets is positively related to growth, something to be expected given that the variable is measured ex-post.
Beyond the basic franchisor rate of outlet growth, it is also important to gain a better understanding of the factors that ultimately determine value creation at the franchisor level, proxied here by the risk-adjusted rate of return on the franchisor stock. A number of relationships can be hypothesized. First of all, growth itself, and its associated investments in fixed assets and working capital, stresses the financial resources of a company, leading to relatively lower reported earnings. The financial constraints of growth should be particularly marked for these companies that elect to own most of the outlets in the system as opposed to those that tap the external capital markets by franchising a higher proportion of their outlets. Second, higher franchise fees, royalties, and advertising payments from franchisees should increase the returns to franchisor capital. The total number of outlets in the system can serve as a proxy for the franchisor's relative bargaining power vis-à-vis suppliers (positive effect anticipated on performance) and franchisees (negative effect anticipated). The combined effect of these changes in bargaining power with total franchise size is unknown. The terms of the franchise contract, including franchise fees, royalty payments, and advertising contributions are expected to positively affect franchisor performance by creating larger income and growth opportunities. The Pearson correlation coefficients for the explanatory variables are presented in Exhibit 6, while the model coefficients are detailed in Exhibit 7.
A number of interesting figures can be highlighted in Exhibit 7. First of all, the explanatory power of the model as presented is extremely high (R2=0.71 and R2adj=0.21), especially when considering the fact that the dependent variable is risk-adjusted abnormal performance, as measured by Jensen's Alpha, and not simply total stock performance6 Second, even though none of the variable coefficients are significantly different from zero at the usual levels of confidence, they appear to be wearing the expected signs. The percentage of outlets owned by the franchisor seems to be positively correlated with performance, indicating possibly inside information at the franchisor level (if the prospects are really good, the incentives to shares the profits with franchisees are limited). Royalty and advertising fees are positively associated with franchisor stock returns, while the franchisee fee is negatively related (larger franchise fees associated with lower franchisor performance). Again, this finding is consistent with the explanation above: franchisor with the brightest prospects may signal their quality by contractually favoring royalties and ad fees instead of the fixed franchise fee. In other words, a high franchise fee creates a conflict of interest between franchisor and franchisee, whereas a performance-based payment system (especially the use of royalties on sales) may realign the interest of both franchisor and franchisees. The plot of the regression residuals indicates no systematic bias.
Franchisor Systematic Risk
An interesting question left to be analyzed is that of the factors contributing to franchisor risk, and in particular the systematic (or market-related) component of that risk. Beta risk is a combination of business and financial risk. With limited financial information at this stage, it is nevertheless interesting to investigate the effect of various components of business risk upon the resulting systematic risk. For example, high system growth can be hypothesized to create the need for further financing and thus of recourse to the market, with its rewards and dangers. This is particularly relevant when most of the growth in new outlets is financed by the franchisor itself and not franchisees. On the other hand, a large number of outlets, and their associated earning power, would dampen the dependence on the market. Franchise fees and royalties should similarly provide alternative sources of capital, reducing market sensitivity. These relationships are investigated in Exhibit 8.
Even though the overall explanatory power of the model is quite limited, and none of the regression coefficients are significant at the usual levels of confidence, it is interesting to note that high system growth in negatively correlated with beta. This finding may be linked to the previous observation that growth may be associated with internal financing, resulting in a lower degree of market exposure. As expected, high royalty payments seem to be associate with lower betas, but franchise and advertisement fees are positively associated with systematic risk. Again, this may be a reflection of the better alignment of franchise payments (expressed a percentage of sales), such as royalties and ad fees, with actual performance, than with the fixed franchise fee.
1The study could clearly be enhanced by adding the
remaining 57 public franchisors to the sampel but no systematic
bias is anticipated from the current transaction.
2As is common in this literature, monthly portfolio rebalancing is assumed in both indices.
3Systematic risks were measured with respect to the same S&P500 market index used in Exhibit 2 using 60 monthly returns. When less than 60 monthly returns were available but more than 24, betas were calculated on the number of months at hand. When less than 24 months of historical returns were available for regression purposes, systematic risks were reported as not available. This procedure is in line with standards adopted by Value Line and Morningstar for reporting betas.
4The rest of the franchised outlets are then franchisee-owned.
5As of December 31st. 1994.
6Similar results have been obtained when utilizing average monthly returns, unadjusted for risk, as the dependent
variable, even though the R2 were in general slightly smaller in these cases.
© 1997 Babson College All Rights Reserved
Last Updated 06/08/98