Long-term entrepreneurship patterns: a national study of black and white female entry and...

Over the past two decades there has been an increasing number of women who have started their own businesses and are self-employed. Data from the U.S. Bureau of Labor Statistics indicate that in 1969 there were 1.2 million self-employed women. This figure roughly doubled by 1982, and by 1991 the

figure has reached over 3 million (U.S. Department of Labor 1969, 1982, and 1991). Furthermore, the U.S. Bureau of the Census indicates that over the same time period the number of female entrepreneurs has increased roughly five times faster than that of the number of self-employed men and more than three times as fast as women who receive wages and salaries (U.S. Department of Commerce 1987).

These emerging trends clearly underscore the increasing importance of gaining insight into the nature and dynamics of the careers of female entrepreneurs. While research attempts have been made to these ends, they have considerable limitations. Virtually all female entrepreneurship studies to date have been limited in one or more of the following ways: the use of convenience samples, small sample sizes, samples limited to specific geographic areas, lack of a control group or other bases of comparison, and perhaps most importantly the use of cross-sectional data where respondents are observed at a single point in time rather than over time (Neider 1987; Pellegrino and Reece 1982; Longstreth, Stafford, and Mauldin 1987; Mescon and Stevens 1982; Cuba, Dezenzo, and Anish 1983; Sexton and Kent 1981; Waddell 1983). Such studies lack generalizability. Moreover, they fail to identify the dynamics of female entrepreneurship. They do not permit an assessment of individual patterns of entry into, and staying in or, conversely, exiting from, entrepreneurship over time. Thus, it is impossible to identify to what extent self-employees in a given year are comprised of new and continuing entrepreneurs.

The importance of gaining insight into the dynamics of entrepreneurship in terms of entry and stayer/exit status has been echoed by many. For example, a recent U.S. government report on small business states, "Most data on establishment change in the economy are presented in terms of net change," that is, presentation of overall changes in self-employment rates, while ignoring the relative contribution of new entrants and stayers. The report states further "...the gross flow of establishment births and deaths over any period, produce a much more active and dynamic economy than normally described. Attention to the relatively large size of the gross flows may provide some insight on appropriate public policy in support of small business" (U.S. Small Business Administration 1985). Only when self-employment rates are considered in terms of gross flows can the self-employment dynamics of entry into and exit from entrepreneurship be addressed.

The value of adopting a dynamic gross flow approach is clearly relevant when examining entrepreneurship patterns of the overall population, as when examining that of specific population strata. To begin to gain insight into the difference between black and white women's entrepreneurship rate patterns over time, the present study adopts a dynamic gross flow approach. Toward this end, this study uses longitudinal data from the National Longitudinal Survey of Labor Market Experience (NLSLME) which follows a nationally representative sample of women from 1967 through 1989.

To date virtually no study has exclusively focused on black female entrepreneurs, let alone directly compared their experience with that of their white counterparts. Of the few minority entrepreneurship studies undertaken, most have combined men and women into the same sample or examined only men. Therefore, in addition to the array of limitations of entrepreneurship studies in general (use of convenience samples, small sample sizes, etc.), identification of the nature and dynamics of black female entrepreneurs is further handicapped. Some existing minority entrepreneurship studies have attempted to identify characteristics of minority entrepreneurs in terms of demographics (Hisrich and Brush 1986, Gomolka 1977) and psychological characteristics (DeCarlo and Lyons 1979). Others have considered the relative success of minority small businesses. For example, Scott examined the success of nongovernmental assisted minority- and nonminority-owned businesses in terms of profitability, indebtedness, and liquidity and found virtually no difference (Scott 1983). More recently Dadzie and Cho investigated a sample of 178 minority business CEOs of whom less than 50 percent were black (Dadzie and Cho 1989). They found that outside assistance obtained in the early growth stages and the presence of successful entrepreneurs as role models were among the factors contributing to the success of small minority-owned businesses. Although such studies suggest reasons why small minority-owned businesses may survive, they fail to account for potentially important differences in ownership by gender or minority group. Furthermore, they ignore the larger picture of identifying black entry and survival rates alone and in comparison to other benchmarks, such as white entry and survival rates. Obtaining such knowledge should logically precede formulating reasons and remedies for entry and success gaps.

DATA AND METHODS

This study uses micro-data from the National Longitudinal Survey of Labor Market Experience. These data were collected for the U.S. Bureau of the Census by the Center of Human Resources of the Ohio State University under a contractual agreement with the U.S. Department of Labor established in 1965. A multi-stage sampling procedure was used to construct the proposed study's nationally representative sample of women aged 30-44 in 1967. Of the 5,393 women originally selected for the sample, 5,083 were successfully contacted for the initial survey in 1967. Sample attrition has occurred, though not beyond the level normally expected.

Though the NLSLME was not undertaken with the specific intention of studying entrepreneurship dynamics, the survey's longitudinal information pertaining to class of worker (self-employed, wage-earner), and current labor force participation lends itself to this type of investigation. The above information is available in each of the 14 survey years between 1967 and 1989. The survey years were 1967, 1969, 1971, 1972, 1974, 1976, 1977, 1979, 1981, 1982, 1984, 1986, 1987, and 1989.

To initially explain variations between black and white female self-employment rates, differences in the probability of entering into entrepreneurship and the probability of staying in entrepreneurship are identified. The probability of entering into entrepreneurship is calculated as the proportion of individuals who become entrepreneurs in a given survey year (|T.sub.1~) from the pool of individuals who were not entrepreneurs in the previous survey year (|T.sub.0~). In a similar fashion, the probability of staying in entrepreneurship is calculated as the proportion of individuals who continued as entrepreneurs in a given survey year (|T.sub.1~) from the pool of individuals who were entrepreneurs in the previous survey year (|T.sub.0~). Therefore, only respondents who provide self-employment information for pairs of successive survey years are included in the analysis.

While differences in entry and staying probabilities might appear to explain variations in self-employment rates, by themselves they do not permit identification of the overall difference between self-employment rates attributable to each of them. The reason for this is that the difference between two self-employment rates is only in part attributable to differences in rate schedules: entry and staying probabilities. It is also attributable to differences in population distributions: the proportion of the population that constitutes the pool of potential entries and that which constitutes the pool of potential stayers.(1) That is, the proportion of the population who were not entrepreneurs in the previous year (|T.sub.0~) and the proportion who were entrepreneurs in the previous year (|T.sub.0~). Thus, to permit an identification of the overall difference between self-employment rates attributable to contributing component factors, one must decompose the overall rate differ ence into those attributable to rate schedules and at the same time to population distributions.

The methodology used is a standard decomposition of rate differences, similar to that described by Kitagawa (1955). The overall difference in the self-employment rates between two populations (i.e., |PO.sub.1~ - |PO.sub.2~), is decomposed into the amounts attributable to differences in rate schedules (R), including the components of the probability of entry (|R.sub.e~) and staying (|R.sub.s~), and that which is due to differences in population distributions (|P.sub.s~), including the components of the proportion of the pool of potential entrants (|P.sub.e~) and pool of potential stayers (|P.sub.s~). (The derivation and resulting estimates of these components are available from the authors.)

Table 1
SELF-EMPLOYMENT RATES

        Total Sample(a)          White               Black
White-Black
       Rate Year     N     Rate (%)       N     Rate (%)       N     Rate (%)
    Difference

1969   4,605     3.69       3,253     4.51       1,281     1.41
3.13(*) 1971   4,445     4.70       3,162     5.56       1,217     2.47
  3.09(*) 1972   4,317     4.75       3,084     5.64       1,168     2.31
    3.33(*) 1974   4,161     4.26       2,967     4.85       1,128     2.66
      2.19(*) 1976   4,004     4.52       2,857     5.46       1,083     2.13
        3.33(*) 1977   3,790     4.88       2,709     6.01       1,027
2.04         3.97(*) 1979   3,624     5.08       2,583     6.04         988
  2.43         3.61(*) 1981   3,469     4.96       2,481     6.37         940
    1.59         4.78(*) 1982   3,353     4.03       2,389     4.94
915     1.86         3.08(*) 1984   3,196     5.92       2,276     7.47
  874     2.40         5.07(*) 1986   3,091     6.11       2,214     7.63
    829     1.93         5.70(*) 1987   2,994     4.78       2,163     5.92
      790     1.52         4.40(*) 1989   2,833     5.29       2,049     6.34
        749     2.14         4.20(*)

* Difference is statistically significant at the .01 level.

a Totals exceed sum of black plus white samples since the total sample
included other racial groups.

FINDINGS

The self-employment rates for the total sample over the study period are reported in table 1. These rates are comprised of the proportion of the sample who are entrants plus those who are stayers. The pattern of self-employment rates found exhibit a decline in the mid-1970s and a fairly consistent increase thereafter. This pattern is fairly consistent with that identified elsewhere (e.g., U.S. Department of Commerce 1969-1989). The rates range from a low of 3.69 in 1969 to a high of 6.11 in 1986.

Looking at the self-employment rates of white and black women, table 1 further indicates that the pattern of white self-employment rates reflected that of the total population over the study period, indicating a substantial decline in the 1970s and a fairly consistent increase thereafter. Their rates ranged from a low of 4.51 in 1969 to a high of 7.63 in 1986. For blacks, the pattern of self-employment rates was less discernable, exhibiting a more random fluctuation from year to year. Their rates ranged from a low of 1.41 in 1969 to a high of 2.66 in 1974. More importantly, in each of the survey years, the self-employment rates for blacks was substantially below that for whites. In each of these years the differences were statistically significant at or below the .01 level.

To begin to gauge the discrepancy between black and white self-employment rates for the different survey years, a consideration of the probabilities of entry into entrepreneurship and staying in entrepreneurship is relevant. As table 2 shows, both the probabilities of entry and staying in each of the different survey years is greater for whites relative to blacks. However, the black-white differential is considerably greater for the probability of entering than it is for staying.

As previously noted, while differences in rate schedules (entry and staying probabilities) might appear to explain variations in self-employment rates, by themselves they do not permit identification of the overall differences between self-employment rates attributable to each of them. Also relevant are differences in population distributions (pool of potential entries and stayers). Accordingly, table 2 shows the results of decomposing the difference in self-employment rates between blacks and whites.(2) For each of the survey years, both differences in rate schedules and differences in population distributions contribute substantially to the overall self-employment rate difference. Whereas differences in rate schedules on the average contribute roughly 60 percent, the difference in population distributions contributes roughly 40 percent.

The results of the rate schedule differences indicate that the differences in overall self-employment rates between blacks and whites, attributable to differences in the probability of entering in each of the survey years, account for an average of roughly 50 percent. In contrast, that which is attributable to the probability of staying is slightly less than 10 percent. With respect to the results of population distribution variations, the difference in overall self-employment rates between blacks and whites, attributable to differences in the pool of potential entries, is virtually nil, accounting for an average of roughly 2 percent. In contrast, that which is attributable to the pool of potential stayers is quite substantial, accounting for an average of roughly 40 percent.

DISCUSSION AND IMPLICATIONS

The present study employed a decomposition methodology in analyzing the difference in the longitudinal patterns of self-employment rates between black and white females in the United States. This methodology permitted an assessment of self-employment dynamics as suggested by a gross flow approach (i.e., considering the entry and stayer patterns). Accordingly, it enabled an identification of the extent to which overall self-employment rate differences over time were attributed to differences in entry and staying likelihoods. The methodology further enabled an identification of the extent to which such rate differences were attributable to differences in the pool of potential entries and stayers.

Overall, the decomposition results indicate that, together, differences in the probability of entry and differences in the pool of potential stayers account for a combined total of roughly 90 percent of the overall self-employment rate difference between black and white women over the survey years considered. These results reflect the notion that black women are far less likely to enter entrepreneurship than whites, but once they do, they are only marginally less likely to stay. Furthermore, the lower black entry probabilities directly translate into a smaller pool of potential black stayers. This, in turn, contributes to the black-white self-employment gap.

The low black self-employment rate patterns found in this study and others is, in part, surprising. Theories pertaining to entrepreneurs as disadvantaged workers would predict a relatively TABULAR DATA OMITTED higher self-employment rate pattern for blacks relative to whites (Light 1972). Because a disproportionate number of blacks have limited marketable wage skills, they might be more likely to opt for self-employment. Likewise, theories pertaining to entrepreneurs as outsiders would also predict higher black self-employment rates (Min 1984). The reason entrepreneurs enter self-employment is because they believe that they are viewed (and accordingly treated) as outsiders by the mainstream wage-employment sector.

One possible explanation for the paradoxically low black self-employment rate is that the limited assets blacks possess hinder them from entering into entrepreneurship (Evans and Leighton 1987). Another possible explanation hinges on the discrimination black entrepreneurs encounter from consumers (Borjas and Bronars 1987). Light, whose disadvantaged worker theory would predict a relatively high rate of self-employment among blacks, notes that historically blacks lacked rotating credit associations, like Asian Americans, and hence, access to assets (Light 1972). Furthermore, black women's access to assets is arguably further handicapped by family structure. Roughly 65 percent of white females were married over the 1970 to 1989 period, whereas this figure was roughly 50 percent for black females. To the extent that married-couple households have greater incomes and assets than other household types (i.e., single-parent households) and to the extent that the proportion of single-parent households is greater for blacks than whites, then black women on the whole have less access to assets (U.S. Department of Commerce 1991).

Our findings can be viewed as initial evidence for addressing the paradoxically low black self-employment rate, at least in regard to black versus white women. Our findings indicated that the relatively low rates found were primarily attributable to very limited likelihoods of entry into entrepreneurship (which in turn gives rise to a limited pool of potential stayers). Therefore, to the extent to which limited access to assets may greatly diminish their otherwise presumed propensity to enter into self-employment, our findings explain the surprisingly low black female self-employment rates. To the extent black females were found to be almost as likely as whites to stay self-employed once they entered self-employment, our findings suggest that the explanation hinging on the discrimination of white consumers against black businesses may be less relevant.

POLICY IMPLICATIONS

If the objective is to increase the much limited black self-employment rates, the results indicate that the key lies in increasing their probability of entering into entrepreneurship. Once this is achieved, the pool of potential black stayers will increase and, to the extent to which blacks have almost comparable probabilities of staying as whites, the self-employment gaps found here will be greatly diminished.

One solution for increasing the probability of blacks entering entrepreneurship would be for the government to increase the opportunities for black women by further underwriting commercial banks, credit unions, state and community micro-loan funds, and other lending institutions and mechanisms that make start-up capital available. Furthermore, government can support such community-based institutions as the National Center for Neighborhood Enterprise which encourages black entrepreneurial endeavors (Woodson 1987).

FURTHER RESEARCH

While the study's methodological approach has been helpful in explaining the seeming paradox of relatively low black female self-employment rates, applicability of its results are somewhat limited. For example, while our findings do not support the discrimination explanation, but rather appear to support the asset explanation, a direct test of the impact of assets in determining entry into entrepreneurship is necessary. What role do changes in family structure, including the presence or absence of spouse and children, as they affect assets, play in determining entry into entrepreneurship for black and white females? What role do cultural values and informal support systems play? Obviously, further avenues of research must be considered.

1 To appreciate this point, assume the following example with respect to the self-employment rates of two populations A and B. Population A's probability of entry is equal to 1. whereas B's is equal to .2. Population A's probability of staying is equal to .4, whereas B's is equal to .8. On initial inspection it might appear that population B has a higher self-employment rate than that of A; given B's higher entry and staying probabilities. Assume however, that B's pool of potential entries constitutes 95 percent of its population and that of potential stayers is 5 percent, while the opposite population distribution characterizes population A. Accordingly, population A does have a higher self-employment rate than population B. Population A's self-employment rate = (.1 x .05) + (.4 x .95)= .385, whereas population B's self-employment rate = (.2 x .95) + (.8 x .05) = 230. Hence, differences in the population distributions can certainly effect overall rate differences.

2 As can be seen in table 2, the absolute difference in white/black self-employment rates for each of the survey years is only slightly larger than the actual difference. This stems from the fact that in each of the years the pool of potential black entries is marginally larger than that of whites, as indicated by the negative values in the |P.sub.e~ column. To the extent to which there is only a slight difference between the actual and absolute difference, we opted to use the actual difference as the denominator in calculating the proportional contribution of the population distribution and rate schedule components. Had we alternatively used the absolute difference as the denominator, the contribution of each of the components would have changed ever so slightly.

REFERENCES

Borjas, G., and S. Bronars (1987), "Self-Employment and Consumer Discrimination," University of California at Santa Barbara, working paper.

Cuba, R., D. Dezenzo, and A. Anish (1983), "Management Practices of Successful Female Entrepreneurs," American Journal of Small Business 7 (2), 40-46.

Dadzie, K.Q., and Y. Cho (1989), "Determinants of Minority Business Formation and Survival: An Empirical Assessment," Journal of Small Business Management 27 (July), 56-61.

DeCarlo, J.F., and P.R. Lyons (1979), "A Comparison of Selected Personal Characteristics of Minority and Non-Minority Female Entrepreneurs," Proceedings of the 39th Annual Meeting of the Academy of Management, 369-373.

Evans, D.S., and L.S. Leighton (1987), "Self-Employment Selection and Earnings over the Life Cycle," Washington, D.C.: U.S. Government Printing Office.

Gomolka, E. (1977), "Characteristics of Minority International Small Business Enterprises," American Journal of Small Business (July), 178-184.

Hisrich, R., and C. Brush (1986), "Characteristics of the Minority Entrepreneur," Journal of Small Business Management 24 (October), 1-8.

Kitagawa, E.M. (1955), "Components of a Difference Between Two Rates," Journal of the American Statistical Association 50, 1168-1194.

Light, I. (1972), Ethnic Enterprise in America: Business and Welfare Among Chinese, Japanese, and Blacks. Berkeley, Calif.: University of California Press.

Longstreth, M., K. Stafford, and T. Mauldin (1987), "Self-Employed Women and Their Families: Time Use and Socioeconomic Characteristics," Journal of Small Business Management (July), 30-37.

Mescon, T., and G.E. Stevens (1982), "Women as Entrepreneurs: A Preliminary Study of Female Realtors in Arizona," Arizona Business (7), 9-13.

Min, P.G. (1984), "From White-Collar Occupations to Small Business: Korean Immigrants Occupational Adjustment," Sociological Quarterly 25 (Summer), 333-352.

Neider, L. (1987), "A Preliminary Investigation of Female Entrepreneurs in Florida," Journal of Small Business Management 25 (3), 22-29.

Pellegrino, E.T., and B.L. Reece (1982), "Perceived Formative and Operational Problems Encountered by Female Entrepreneurs in Retail and Service Firms," Journal of Small Business Management 20 (2), 15-24.

Scott, W.L. (1983), "Financial Performance of Minority- versus Nonminority-Owned Businesses," Journal of Small Business Management 22 (1), 42-48.

Sexton, D.L., and C.A. Kent (1981), "Female Executives and Entrepreneurs: A Preliminary Comparison," in Frontiers of Entrepreneurship Research, ed. J.A. Hornaday, J.A. Timmons, and K.H. Vesper, Wellesley, Mass.: Babson College, Center for Entrepreneurial Studies, 40-55.

U.S. Department of Commerce, Bureau of the Census (1987), Current population reports, special studies series, P23 No. 146, Women in the American Economy. Washington, D.C.: U.S. Government Printing Office.

@ @