ANALYSIS OF SURVIVAL RATES AMONG FRANCHISE AND INDEPENDENT SMALL BUSINESS STARTUPS'

by Timothy Bates
ABSTRACT

Analysis of small firm formations over the 1984-1987 period reveals that franchise startups exhibit both higher rates of firm discontinuance and lower mean profitability than cohort independent business startups. The young franchising firms are heavily concentrated in retaihng, and it is the retail franchises that most often fit the high risk, low return profile. Popular franchising niches in retailing appear to be saturated, leaving diminished prospects for newcomers.

A recent ad in Business Week succinctly states the conventional wisdom on the risk involved in entering selfemployment by purchasing a franchise: "A franchisee has a four times greater chance to succeed than an entrepreneur who launches a new independent business." The franchise, we are told, is a safe bet. However, findings of this study indicate that young franchise startups exhibit both higher rates of firm discontinuance and lower mean profitability than cohort independent business startups. When owner and firm traits are controlled for statistically in logistic regression, the franchise characteristic is found to be negatively related to firm survival prospects. These findings are
Dr. Bates is professor of labor and urban affairs at Wayne State University in Detroit, Michigan. His research interests include small business formation, expansion, and survival patterns, as well as the impacts government economic development programs have upon small business viability. 'Research reported in this study was conducted on-site at The Center for Economic Studies, U.S. Bureau of the Census. Findings addressed are those of the author and do not necessarily reflect the views of the U.S. Bureau of the Census.

based upon analyses of approximately 20,000 young small businesses, utilizing nationwide data compiled by the U.S. Bureau of the Census. The findings of this study are consistent with the hypothesis that popular franchising niches have become saturated, leaving diminished prospects for newcomers. Net income in 1987 among firms formed during the 1984 to 1987 period, for example, averaged -$4,102 for retailing franchises and $14,572 for cohort independent business startups. For these same retailing firms that were operating in 1987, 45.1 percent of the young franchises had gone out of business by 1991, versus 23.4 percent of the independent young retail firms. Relative to aU small business startups, franchises are heavily concentrated in retailing, and it is the retail franchises that most often fit the high risk, low return profile. Note that firms sold to a new owner, merged, or otherwise altered are not counted as discontinued in this study if

26

Journal of Small Business Management

they continued to operate. Franchise discontinuance rates reported in this study differ dramatically from those cited by various franchisors and franchisor associations. According to research sponsored by the International Franchise Association, for example, 96.9 percent of the franchised units opened nationwide within the past five years are still in operation (Arthur Anderson and Co. 1992). Significantly, this survival rate information was compiled by surveying franchisors the corporations that sell franchises rather than the actual franchisee owners of the operations whose survival is at issue. In contrast, the approach of this study is to rely upon franchisee owners to selfreport information about the continuity of operations in the small businesses that they own and operate. Most evidence on franchise survival rates comes from sources other than scholarly journals, so that peer review monitoring of research quality is not present. The February 1994 issue of McGalls magazine claims that 50.7 percent of Decorating Den's franchisees terminated operations during the three-year period ending in December 1992, as compared to the Anderson number quoted above. Thus, claims about franchise survival rates have often tended to extremes. The purpose of this study is to raise the debate to a higher plane by using nationwide data from the Census Bureau on business survival and to conduct appropriate statistical analyses on these data.
CHARACTERISTICS OF FRANCHISE AND INDEPENDENT YOUNG FIRMS

The presence of a large new nationwide small business database compiled by the Census Bureau the Characteristics of Business Owners (CBO) database makes it possible for issues of franchise survival to be analyzed comprehensively. The CBO database is

described in detail in Bates (1990a), Nucci (1992), and in Appendix B: Nature of the Database. Of the roughly 90,000 small businesses surveyed to create the CBO database, over 72 percent responded. All of the reported statistics in this study are weighted to adjust for both survey non-response and the Census Bureau's non-random sampling in the creation of the CBO. The firms described in this study are representative of young firms that grossed at least $5,000 in total revenues in 1987 and filed a small business income tax return. This study covers only firms formed over the 1984-87 period, and the unit of analysis is firms, not persons. Thus, the universe of firms covered in this study is 4,005,562 small businesses, 3.36 percent of which are young franchise operations. Choice of the 1984-1987 time frame of firm formation was dictated by both the specific structure of the CBO database (described at length in Appendix B: Nature of the Database), and the need to generate a large enough franchise sample to permit analysis of subgroups such as franchises in retailing. The actual sample of CBO firms analyzed in this study includes unweighted firm counts of 1,276 franchises and 19,278 independent business startups (6.21 percent franchises). The higher franchise proportion appearing in the raw sample data reflects the fact that the CBO sampling frame over-sampled firms having paid employees, since franchises are more likely than independent firms to utilize paid employees. Because of the nonrandom nature of the CBO's sampling frame, only weighted statistics reflecting all young firms as defined above are reported in this study. The sample of franchise firms analyzed here is not identical with the universe of franchise units opened between 1984 and 1987 for several reasons. First, some franchise units are owned by the franchisor. Second, some new franchise 27

April 1995

units are owned by multi-unit firms that began operations before 1984. Third, multi-unit new firms included in this study might own franchise operations at several locations, but such a firm is only counted once because the unit of analysis here is the firm. Thus, the failure rate for all new franchise units operating in 1987 that opened up between 1984 and 1987 may differ from the closure rate reported in this study. Existing studies identify traits that are positively correlated with firm longevity. Larger-scale, older small businesses have higher survival rates over time than smaller, younger firms (Evans 1987, Bates and Nucci 1989). Among very young firms, owner traits associated with greater likelihood of survival

include owners working fuU-time in the firm, highly educated owners, and large owner financial capital investment in the firm at startup (Bates 1990b). Ikble 1 indicates that franchisees are generally better endowed with traits linked to survival than non-franchise young firms. The young franchisees report $513,961 in mean 1987 sales revenues, five times larger than the corresponding figure of $102,410 reported by the independent businesses (Tkble 1). Capitalization at startup is similarly much greater with a mean value of $86,493 for the franchise firms, almost three times greater than the non-franchise firm capitalization of $29,822. Only in the area of owner educational background do the franchise firms appear to be weaker than the inde-

Table 1 YOUNG FIRMS OPERATING IN 1987: A COMPARISON OF OWNER AND FIRM TRAITS FOR FRANCHISE AND INDEPENDENT BUSINESS STARTUPS (Firms formed from 1984-1987 Only) A. Firm Traits Franchises Nonfranchise firms Univariate F 1987 sales (mean) $513,961 $102,410 153.5* Number of employees (mean) 5.2 1.2 26.9* 1987 net income (mean) $2,649 $18,850 13.2* Percent of firms still operating 65.3% 72.0% 14.2* iniate1991 B. Owner Traits 1. Owner education: less than high school high school graduate college: 1-3 years college graduate post graduate 2. Firm capitalization at startup: total financial capital (mean) equity (mean) debt (mean) 3. Other owner traits: owner gender (male = 1) owner age (mean) owner annual hours of labor input (mean) owner management experience (yes = 1) unweighted sample size weighted sample size

10.3% 35.7% 27.3% 17.6% 9.1% $86,493 $38,355 $48,138 73.3% 38.9 2,013 39.3%

11.7% 28.8% 23.3% 20.0% 16.2% $29,822 $13,591 $16,231 74.1% 40.6 1,870 30.1%

1.1 15.1* 4.4* 2.3 23.7* 81.6* 46.7* 53.1* 0.2 13.9* 9.8* 25.4*

1,276 19,278 130,180 3,875,382 Source: Characteristics of Business Owners database. *Trait differences between the above groups are statistically significant at the .05 significance level.

28

Journal of Small Business Management

pendents: 9.1 percent of the former and 16.2 percent of the latter had pursued graduate studies beyond the bachelor's degree. All of the above mean differences are statistically significant. Yet, despite the obvious strengths of the young franchise firms, they are dramatically less profitable than independent firms of the same age (Tkble 1), and they exhibit a lower survival rate 65.3 percent (versus 72.0 percent for nonfranchise firms) over the 1987-late 1991 time period; these differences are statistically significant at conventional levels. Aspiring entrepreneurs choosing to become franchisees certainly expect to improve their odds of survival over the turbulent early years of business startup. This expectation of lower risk for franchise as opposed to independent business startups is certainly portrayed in existing industry and academic studies (see, for example, Castrogiovanni, Justis, and Julian 1993). Overall, many of the traits typifying young franchises appear to be consistent with enhanced viability and improved survival prospects. Relative to similarly aged independent firms, they are much larger, better capitalized, and generate higher sales per employee and per dollar of financial capital. Their only seeming disadvantage as a group is their lower incidence of very highly educated owners. Could this lower incidence of graduateeducated owners account for the fact that the franchise group exhibited a lower rate of survival than independent business startups? Impacts of owner education, firm size and capitalization, the franchise trait, and other factors upon survival and discontinuance are further investigated in Tkble 2. Over the period from 1987 to late 1991, over 28 percent of the young firms described in Tkble 1 went out of business. Based upon the findings of past econometric studies explaining small business

longevity, greater owner investments of human and financial capital are expected to be related positively to the survival chances of young small businesses (Bates 1990a, Bates 1994; HoltzEakin, Joulfaian, and Rosen 1994). In Tkble 2's logistic regression analysis of firm survival p a t t e r n s over the 1987-1991 period, financial capital is measured by the log of the sum of debt and equity capital invested in the firm at the point of startup. Labor input quantity is measured by owner hours spent working in the business, as well as marital status and number of paid employees. Married persons living with their spouses are expected to benefit from the availability of family labor, which potentially increases labor input quantity. Quality of owner human capital is measured by two variables: level of formal education and managerial experience prior to small business entry. Applicable demographic traits include owner age, racial grouping, and gender. Greater owner age, a broad proxy for work experience, is expected to benefit firms until diminishing effort associated with old age sets in. In an important sense, all of the above factors can be thought of as control variables, since the major point of Table 2's logistic regression analysis is to observe the impact of the franchise trait, other things being equal, upon firm survival. In Ikble 2's analysis of young firms that were operating in 1987, positive coefficient values are associated with firms still operating in late 1991, and vice versa. The franchise characteristic, other factors constant, is a highly significant determinant of firm survival: franchisees are much more likely to go out of business than cohort independent firms. Other Tkble 2 findings reinforce results of past studies explaining firm survival patterns. The surviving firms active in late 1991 are disproportionately those headed by highly educated owners who worked full-

April 1995

29

Table 2 LOGISTIC REGRESSION: EXPLAINING FIRM SURVIVAL OVER THE 1987-LATE 1991 PERIOD (Firms formed from 1984-1987 only) Regression Coefficient Standard Error Variabie Mean

-1.474* .212 Constant Education: .171* .112 less than high schooi .063 high school graduate -.037 .046 .284 college graduate .400* .052 .205 post-graduate .654* .059 .159 Management experience/ no management experience .038 .042 .312 Owner age .079* .010 40.555 -.001 * .0001 1,766.1 Owner age^ Owner gender -.034 .039 .738 Owner labor input .025* .002 18.764 Number of employees .088* .010 1.373 Married/unmarried -.030 .042 .779 .086* .004 6.846 Log financial capital Leverage .014* .003 2.516 Franchise/nonfranchise -.523* .094 .033 .307 -.218* .046 Firm entered 1986 -.819* .043 .373 Firm entered 1987 Minority-owned firm/ nonminority-owned firm .101 .060 .092 Note: precise variable definitions for the above explanatory variables are detailed in Appendix A. Sample size =19,463 -2 log likelihood = 20,610.3 Chi-square = 2,335.3 *Statistically significant at the 5 percent significance level. time in the business. The surviving firms, further, were the larger firms in the sense that they began operations with greater owner financial capital investments, and labor input measured by number of employees was also higher among the survivors (Tkble 2). The very youngest firms those started in 1987 were most likely to discontinue operations, which is consistent with past findings (Evans 1987, Bates 1990b, Jovanovic 1982). Other than the franchise survival relationship finding, the results of Tkble 2's logistic regression analysis are highly consistent with past empirical studies of small business discontinuance patterns. WHY ARE YOUNG FRANCHISES DOING POORLY? In s t a n d a r d microeconomic theory, negative profitability among firms operating in a particular market niche is a signal t h a t too many firms are active in t h a t sector; firm exit will presumably cure this overcrowding. Although this study examines young firms only, it is t r u e t h a t franchises of aU ages a r e very heavily concentrated in certain retail markets, a n d this retailing concentration also typifies t h e franchises d e scribed in Tkble 1. A comparison of Tkble l's firms by major industry group highlights t h e particular industry distribution of franchises: Young Franchises Only 42.8% 10.7% 16.2% 30.3% All Young Small Firms 17.3% 42.0%
7 70/

Percent in retail Percent in services* Percent in FIRE** Percent in other industries

33.0%

Total 100% 100% * dominant lines are business, personal, repair, professional, and recreational services. ** finance, insurance, and real estate.

30

Journal of Small Business Management

If franchises have saturated particular markets, the above numbers suggest that retailing, rather than services, is the leading candidate. Although tests for market saturation are not definitive, this phenomenon, if present, is expected to coexist with both reduced profitability and heightened firm closure rates, relative to firms operating in all lines of small business. All of the retailing firms initially described within Table l's broader groupings are isolated and examined in Tkble 3, which reveals low profitability and survival rates for: (1) retail franchises as opposed to retail independent startups, and (2) retail franchises, versus all other young franchise firms. Considering franchising only, applica-

ble profit and survival figures for retail and nonretaU young firms are as shown below: Young Retail All Other Franchises Young Only Franchises 1987 net income (mean) -$4,102 $7,762 Percent of firms still operating in late 1991 54.9% 73.1% The above differences are statistically highly significant: young retail firms are m u c h less profitable a n d m u c h more likely to go o u t of business t h a n franchises operating in nonretailing fields. This p a t t e r n is consistent with t h e hypothesis t h a t saturation exists among franchises in t h e retailing industry.

Table 3 YOUNG RETAIL FIRMS OPERATING IN 1987: A COMPARISON OF OWNER AND FIRM TRAITS FOR FRANCHISE AND INDEPENDENT BUSINESS STARTUPS (Firms formed from 1984-1987 Only) A. Firm Traits 1987 sales (mean) Number of employees (mean) 1987 net income (mean) Percent of firms still operating in late 1991 B. Owner Traits 1. Owner education: less than high school high school graduate college: 1-3 years college graduate post graduate 2. Firm capitalization at startup: total financial capital (mean) equity (mean) debt (mean) Franchises $993,152 9.3 $-4,102 54.9% Nonfranchise firms $145,956 2.1 $14,572 76.6% Univariate F 149.5* 169.7* 20.3* 82.7*

3.5% 40.0% 30.9% 20.3% 5.3% $118,643 $50,103 $68,539

13.7% 31.1% 26.3% 20.0% 8.9% $43,593 $18,637 $24,955

30.3* 11,7* 4.9* 0,2 5,5* 81.8* 56.1* 57,5*

3. Other owner traits: owner gender (male = 1) 81.6% 69.1% 23,9* 40.9 2,2 owner age (mean) 41.8 2,030 2,024 ,01 owner annual hours of labor input (mean) owner management 62.5% 33.0% 125.5* experience (yes =1) 645 3,798 unweighted sample size weighted sampie size 55,717 670,441 Source: Characteristics of Business Owners database. *Trait differences between the above groups are statistically significant at the .05 levei.

April 1995

31

Analyzing young retail firms only, Tkble 4 replicates the logistic regression analysis of firm survival presented in Tkble 2. The resulting analysis of survival over the 1987-1991 period for young retail small businesses reveals that the franchise characteristic is the strongest predictor of firm discontinuance: the retail franchise firms are much more likely to go out of business, other factors constant, than independent young retail businesses. Note that the franchise variable regression coefficient jumps from -.523 (Tkble 2) to -1.298 {Table 4) when the focus is narrowed from small businesses generally to retail operations specifically. These findings, like Table 3's summary statistics, are consistent with the hypothesis that overcrowding in retail franchise niches is generating lower

profitability and higher discontinuance rates for these specific franchises, as opposed to cohort young small businesses. Other traits attributed to franchisees may contribute to their observed high failure rates and low profitability. Entrepreneurs' risk preferences may influence their decision to enter self-employment by buying a franchise. Williams (1994) argues that the more risk-averse person wiU prefer franchises because services offered by the franchisor an established product, for example presumably facilitate spreading the risk between the franchisor and the franchisee. Risk aversion is not obviously captured by any of the variables available in the CBO database, and is therefore not captured in the Tkble 2 and 4 regression analyses. If, indeed.

Table 4 LOGISTIC REGRESSION: EXPLAINING RETAIL FIRM SURVIVAL OVER THE 1987-LATE 1991 PERIOD (Firms formed from 1984-1987 only) Regression Coefficient Constant Education: less than high schooi high schooi graduate coilege graduate post-graduate Management experience/ no management experience Owner age Owner age= Owner gender Owner labor input Number of employees Married/unmarried Log financiai capitai Leverage Franchise/nonfranchise Firm entered 1986 Firm entered 1987 Minority-owned firm/ nonminority-owned firm -3,049* -,444* -,145 -.056 .982* -.330* .163* -.002* .429* ,012* ,072* ,138 ,116* ,003 -1,298* -,203* -,648* Standard Error ,553 ,152 ,112 ,125 ,216 ,110 .026 ,0003 .095 .004 .014 .107 ,013 ,007 ,151 ,119 ,107 ,110 ,320 ,205 ,089 ,358 40.971 1,798,9 .702 20,163 2.745 .795 8.510 3.637 .082 ,281 ,392 Variabie iUlean

,076 ,138 ,114 Note: precise variable definitions for the above expianatory variables are detailed in Appendix A. Sample size = 4,213 -2 log likelihood = 3,351,8 Chi-square = 480.9 *Statisticany significant at the 5 percent significance ievei.

32

Journal of Small Business Management

franchisees are particularly risk-averse as a group, and risk aversion tends to cause higher firm discontinuance rates and lower profitability, this unobserved phenomenon may partially explain the patterns of laggard franchisee performance observed in this study. The risk aversion hypothesis is suspect, however, because it logically implies poorer franchisee performance across the board; it does not explain why franchises in retailing would be a particularly laggard group. Nonetheless, research on franchise behavior is properly viewed as being at an

early stage. Future research is likely to extend our understanding of the underlying causes of the lower survival rates that typify the young franchise small businesses. One possible cause of poor franchise performance is the costs specific to this organizational form upfront fees to the franchisor, royalty fees, marketing fees, and the like. Do these costs exceed the benefits accruing to young firms choosing the franchise route? Are low margin lines of retailing particularly burdened by such fees? The potential research agenda is just beginning to emerge.

APPENDIX A: LOGISTIC REGRESSION - VARIABLE DEFINITIONS

The dependent variable in the logistic regression exercises of Tables 2 and 4 is whether or not the business that was operating in 1987 is still functioning in late 1991. Businesses still operating are considered active firms; those that have closed down are considered discontinued. Independent variables are defined below: Less than high school For owners not completing high school, this variable = 1; (otherwise = 0). For owners completing four years of high school, this variHigh school grad able = 1; (otherwise = 0). For owners completing at least one year of college but not Some college attaining a bachelor's degree, this variable = 1; (otherwise = 0). This variable is excluded from the logit equation. For owners awarded a bachelor's degree, this variable = College grad 1; (otherwise = 0). For owners that attend graduate school beyond the bachePost-graduate lor's degree, this variable = 1; (otherwise = 0). For those working in a managerial capacity prior to ownManagement experience ing the business they owned in 1987, management experience = 1; (otherwise = 0). Owner age A continuous variable measured in years. Owner age squared. Owner age^ Owner gender For male owners, gender = 1; (otherwise = 0). For African American, Asian, Native American and HisMinority-owned firm panic owners, minority-owned firm = 1; (otherwise = 0). Number of hours during the 1987 calendar year spent by Owner labor input the owner working in the relevant small business, divided by 100.

April 1995

33

APPENDIX A (Continued)

Employees Married Log financial capital Leverage Franchise Firm entered 1986 Firm entered 1987

Average number of paid workers reported to the federal government on 1987 quarterly payroll forms. For married owners living with their spouse, married = 1; (otherwise = 0). The log of the sum of debt and equity capital used to start or become owner of the business. The ratio of debt to equity capital invested in the firm at the point of entry. For franchises, franchise = 1; (otherwise = 0). If the business was started or ownership was acquired in 1986, then this variable = 1; (otherwise = 0). If the business was started or ownership was acquired during 1987, then this variable = 1; (otherwise = 0). tive of firms nationwide that grossed at least $5,000 in total revenues in 1987 and filed one of the above income tax returns. In the case of multi-owner firms, all owners (up to the ten largest) were surveyed in the creation of the CBO. Owner characteristics, when reported in this study, reflect the traits of the owner who worked the largest number of hours in the applicable firm during 1987 (the "dominant" owner). Of the roughly 26,000 nonminority male and 26,000 nonminority female firm owners surveyed in the construction of the CBO, response rates were 73.8 percent for males and 75.4 percent for females. Due to the tax return sampling frame, information available on nonrespondents included firm sales, location, and industry; this information was used to weigh the respondent sample so that it is representative of the universe of relevant tax return filers/business owners, and not just the respondents. This study covers only firms formed over the 1984-1987 period; the unit of analysis is firms, not persons. Thus, the universe of firms covered in this study is 4,005,562 small businesses. In cases where persons entered self-employment by purchasing an existing firm, the applicable firm is included if it was purchased in the 1984-1987

APPENDIX B: NATURE OF THE DATABASE

The version of the Characteristics of Business Owners (CBO) database utilized in this study was compiled by the U.S. Bureau of the Census in 1992. A description of the CBO appears in Bates (1990a). The CBO was drawn from the universe of persons who filed a small business federal income tax return in 1987, including: (1) Schedule C, Form 1040, (2) Form 1065, and (3) Form 1120s. According to Census Bureau procedure, a small business is defined as existing in cases where: (1) one of these income tax forms was filed, and (2) gross selfemployment revenues listed on that form for 1987 were at least $500. This study utilizes a more rigorous definition regarding what constitutes a small business: fihng of the requisite tax return is required and 1987 gross selfemployment revenues must be at least $5,000. The CBO over-samples women and minority business owners, as well as owners of firms that utilize paid employees. All of the reported statistics in this study are weighted to acljust for the Census Bureau's nonrandom sampling in the creation of the CBO. The firms described in this study are therefore representa34

Journal of Small Business Management

time period, and in the case of multiple owner firms, the new owner is the dominant owner. In cases where persons bought into ongoing firms that were started before 1984, the firm is excluded in all cases of nondominant owners entering between 1984 and 1987. Variables describing firms and their owners were actually compiled from three different sources when the CBO database was created. Most of the firm and owner information was collected from the survey questionnaire filled out by the smaU business owners. Of the variables examined in this study, the following derive from owner survey responses: (1) net income in 1987; (2) whether or not the firm is still operating in 1991; (3) whether or not the firm is a franchisee or an independent startup; (4) owner years of education and degree status; (5) firm capitalization variables, including leverage and total startup capital; (6) owner age; (7) owner managerial experience; (8) hours of owner labor input; and (9) owner race and ethnicity information. Variables collected from tax return data include: (1) 1987 sales; (2) number of employees in 1987; and (3) industry in which the firm is operating. Additional information on owner gender, race, and ethnicity, finally, was collected from social security records. Note that owners occasionally run more than one small firm. In such cases involving sole proprietorships, the applicable firms are consolidated and treated as one small business. Note, further, that many firm owners have wage income in addition to selfemployment income many are parttimers in the small business world. They are included if they meet the criteria of (1) filing one of the applicable small business income tax returns, and (2) grossing revenues of $5,000 plus in 1987. One deficiency of the CBO database is its exclusion of small business corporations that do not file income tax returns

as S corporations (Form 1120s not used). Excluding these corporations from an analysis of young franchisee survival introduces a potential selection bias. Assuming that the excluded small business corporations tend to be larger scale operations than the average firm covered by the CBO database, an analysis based solely upon CBO data would underrepresent the larger franchise small businesses. If the larger-scale franchises have higher survival rates than their smaller counterparts, then an analysis relying solely upon CBO data would understate the actual survival rate of franchise operations. In sample observations where item nonresponse was a problem, my rule has been to try to keep the applicable firm in the sample. Item nonresponse was pronounced for only one of the variables analyzed in this study 1987 net income. The Table 1 sample size of 20,554 firms falls to 19,463 firms in Tkble 2 because of item nonresponse. Firms from Ikble 1 are included in the Tkble 2 regression analysis only if they provided information on all of the variables included in this statistical analysis. Interested researchers may access the CBO database at the Center for Economic Studies, U.S. Bureau of the Census in Suitland, Maryland.
REFERENCES

Arthur Anderson and Co. (1992), Franchising in the Economy: 1989-1992. Washington, D.C.: International Franchise Education Association. Bates, Timothy (1990a), "New Data Bases in the Social Sciences: The Characteristics of Business Owners Data Base," Journal of Human Resources 25 (F^), 752-756. (1990b), "Entrepreneur Human Capital Inputs and Small Business Longevity," The Review of Economics and Statistics 72 (November), 551-59. 35

April 1995

(1994), "Social Resources Generated by Group Support Networks May Not Be Beneficial to Asian Immigrant-Owned Small Businesses," Social Forces 72 (March), 671-689. Bates, Timothy, and Alfred Nucci (1989), "An Analysis of Small Business Size and Rate of Discontinuance," Journal of Small Business Management 27 (October), 1-7. Castrogiovanni, Gary, Robert Justis, and Scott Julian (1993), "Franchise Failure Rates: An Assessment of Magnitude and Influencing Factors," Journal of Small Business Management 31 (April), 105-114. Evans, David (1987), "The Relationship Between Firm Size, Growth, and Age," The Journal of Industrial Economics 35 (June), 567-582.

Holtz-Eakin, D., D. Joulfaian, and H. Rosen (1994), "Sticking It Out: Entrepreneurial Survival and Liquidity Constraints," Journal of Political Economy 102 (February), 53-75. Nucci, Alfred (1992), "The Characteristics of Business Owners Data Base," U.S. Bureau of the Census Center for Economic Studies discussion paper. Jovanovic, Boyan (1982), "Selection and Evolution in Industry," Econometrica 50 (May), 649-670. Washer, Louise (1994), "Inside Story," McCaii's (February), 113. Williams, Darreli (1994), "Why Do Entrepreneurs Become Franchisees?: An Empirical Analysis of Organizational Choice," (unpublished).

Copies of articles from this publication are now available from the UMI Article Clearinghouse.
For more intormation about the Clearing-house, please fill out and mail back the coupon below.

Creariiighoiiso
Yes! I would like to kncM' more about UMI Article Clearinghouse. I am interested in electronic ordering through the following systemlsl: Z DWLOC Dialorder Z OnTyme ~ ITT Dialcom Z OCLC ILL Subsystem NameTitle Institution/Company Department Address City Phone ( ) State-Zipi__

!Z Other (please specifyl. H I am interested in sending my order by mail. Z Please send me your current catalog and user instructions for the svstem(sl I checked above.

Mail to: University Microfilms International 300 North Zeeb Road. Box 91 Ann Arbor. Ml 48106

36

Journal of Small Business Management