Race and gender differences in business ownership and business turnover: an empirical study using a new, unique data series

Business Economics, Oct, 2002 by Richard J. Boden, Brian Headd

Empirical analyses of business turnover have heretofore been confined mostly to the manufacturing sector, largely due to the lack of suitable data for other industries. Moreover, empirical inquiries into business turnover by demographic category of ownership have been few and of questionable quality; again owing to inadequate data. This study uses a novel longitudinal Bureau of the Census employer data series to examine the survival prospects of new employer businesses (business started in 1992) for four different, mutually exclusive classifications of business ownership--namely, White non-Hispanics; White Hispanics; Blacks; and Asian and other minorities. Regression-like proportional hazard models are spec fled and estimated to produce insight into the impact of certain business, industry; and geographic characteristics on businesses' risk of dissolution over the 1992-1996 time period. The results show that there are differences across race and gender of ownership in business dissolution rates and factors re lated to dissolution.

**********

In this study, we look at how business longevity and determinants thereof vary with the race and gender of business owners. We do so with a novel, longitudinal microdata series on employer businesses that permits classification of businesses according to their owners' demographic status. Empirical examination of race and gender of business ownership and business survival can be traced back to Bates (1990).

Background

The theoretical literature on business turnover is somewhat thin, with Jovanovic (1982) being perhaps the most widely cited contribution. He develops a general equilibrium model of business formation and dissolution. In Jovanovic's model, individuals who enter business ownership are assumed to be unaware of their latent entrepreneurial ability, but acquire this knowledge ex post; and those with inferior entrepreneurial ability subsequently select out of business ownership.

There have been a number of empirical studies on business survival/closure, however. Two studies that are most similar to ours in terms of approach and quantitative methods used are worth noting. Audretsch and Mahmood (1995) use longitudinally linked data from the Dun and Bradstreet Corporation on U.S. manufacturing establishments started in 1976. These researchers estimated proportional hazard models for all manufacturing establishments, including a small proportion of establishments that were parts of multi-establishment firms. They find, among other things, that an establishment's risk of dissolution is negatively related to its initial employment size, other things equal. Mata and Portugal estimate hazard models of firm dissolution using data on Portuguese firms that were started in 1983. (Their data span the 1983-1988 time period.) They, too, find that a firm's risk of dissolution is negatively related to its initial employment size, other factors held constant. In separate, descriptive tabulations, Mata and Portugal find that estimated survival rates for establishments that were parts of multi-establishment firms systematically exceeded the estimated survival rates of single, independent establishment firms.

Dissolution vs. Failure

The data used in our study permit identification of business dissolutions or closures, but contain no information pertaining to the reason for a business' dissolution. This is an important caveat, because business dissolution and "business failure" are not synonymous; rather, the latter is a subset of the former. Though failure does not have a precise technical definition, in the past the Dun and Bradstreet Corporation used to refer to a business closure resulting in a loss to the business' owner(s) and/or creditor(s) as a business failure. More generally, one might think of a business failure as an involuntary business closure, a subset of which includes bankruptcy.

Headd (2000) exploited a unique data source--the 1992 Characteristics of Business Owners (CBO) data series--to separate successful from unsuccessful business closures and the factors underlying the two outcomes. The 1992 CBO data unite administrative records data on business characteristics with owner-provided survey data on owner and additional business characteristics. (1) In his research sample (comprised of businesses started between 1989 and 1992), Headd finds that 29.1 percent of owners of businesses that closed between 1992 and 1996 considered their businesses to be successful at the time of closure. He further presents figures on n-year survival rates that are strikingly consistent with Phillips and Kirchoff (1989) and considerably more optimistic than the dismal, uncorroborated "statistics" on new firm mortality so often mentioned in the popular media. Finally, Headd also estimated logit models to determine the factors that distinguish firms that successfully ceased operation from those that represen t de facto business failures. (2)

Using a New Data Series to Analyze Business Survival

Using a unique, matched data series, the analysis in this study focuses on the determinants of the dissolution of employer businesses started in 1992--over the 1992-1996 time period--by businesses' demographic category of ownership. This data series consists of the 1989-1996 Business Information Tracking Series (BITS) matched with the 1992 Survey of Minority-Owned Business Enterprises (SMOBE) and 1992 Survey of Women-Owned Businesses (SWOBE) data series. (3) (The BITS is a micro-data file linking all U.S. employer firms annually, begin-fling with 1989. The SMOBE and SWOBE data are cross-sectional data and are only available for economic census years--i.e., years ending in "2" or "7." These datasets, and consequently the matched data series, contain limited information on business characteristics, including sales, payroll, employment, legal form of organization, industry, and geographic location.) These data series were matched and re-weighted by Dr. Alicia Robb while she was employed by the Office of Advocacy of the U.S. Small Business Administration (SBA). This matched file enables longitudinal analyses of businesses by gender and demographic category of ownership. (4) Robb's own business survival analyses using these data can be found in Robb (2000).