Identifying factors associated with business formation has been an important theme in entrepreneurship research for many years.* One useful analytic division of such factors has been the "pull" and "push" theses. The former, the larger of the two strains, focuses on factors which draw the individual into a business, e.g., market opportunity, mentors, social networks, need for achievement. It attributes entry to factors which pull the individual away from current economic activity toward business formation. "Pushes," in contrast, focus on factors which drive the individual from his current status into self-employment, e.g., a personal tragedy, mid-life crisis, a loss of employment. If "pulls" are the positive attractions, "pushes" are the negative drivers.

Though the authors are aware of no quantification, the loss of a job, i.e., unemployment or the prospects of unemployment, appears to be the most frequent "push." Unemployment is a proxy for economic stress. Few Americans can endure unemployment for even a few months without significant changes in their standards-of-living. Though unemployment compensation programs moderate the strains, replacement ratios and benefit duration are intended to provide incentives to return to work quickly as well as to cushion the financial impact of unemployment. In addition, a small number of working Americans are not covered by unemployment compensation programs. Any cushion, in these cases, is totally dependent on the individual's resources.

The loss of employment may also serve as a proxy for social and psychological stress. A social stigma is frequently attached to unemployment as illustrated by the truism, 'its easier to get a job from a job than get a job without one,' and increased social pathologies, e.g., family violence, during periods of unemployment have been documented. Thus, the relationship of unemployment to new business formation allows us to further explore the idea of "pushes" as an incentive for new business formation.

The relationship between unemployment and business formation impacts various research strands. Policy is obviously one. A decade-old stimulus for exploring the relationship between unemployment and business formations was a series of policy initiatives providing unemployed people with subsidies to form their own firms. The research issues implicit in these initiatives were 'what is the propensity of unemployed people to form businesses? How successful are those businesses? And, what incentives could increases the rate of entry and success?' The initiatives have become full-blown programs, particularly in France, Germany and the U.K. Experiments in the United States have met modest success, but 11 states have now institutionalized them (U.S. Department of Labor, Manpower and Training Administration, 1994). Though program experimentation and operation provides the definitive answer to the effectiveness of these initiatives, the basic research issues outside the direct program context reflect not only in potential revisions of the programs but also possible initiatives to focus on the low income and those with few saleable skills.

A different policy debate brings another reason to explore the relationship. The American economy and labor markets are undergoing major transitions, the scope and nature of which we are not completely clear. These transitions have brought stress to significant numbers of individuals and groups. This phenomenon has been accompanied by the rise in self-employment, and many are thought to have taken refuge in it, e.g., the consultant (U.S. Department of Labor, Bureau of Labor Statistics, 1995). Thus, the rise of self-employment is a negative phenomenon brought about by negative factors. When those negative factors recede, so will self-employment. One objective indicator of the degree self-employment is a voluntary phenomenon, i.e., a "pull" rather than a "push," is the employment status of the individual prior to formation (Dennis, 1996). If the individual is employed, entry is probably a "pull;" if the individual is unemployed, entry is probably a "push."

A third policy question involves those out of the labor force. The explored relationship traditionally attempts to tie unemployment and business formation. Segal's (1996) calculations, though ignored in his text, demonstrate that those outside the labor force play a more prominent role in formations than the unemployed. These "outsiders" include housewives, retired people, students, disabled and "discouraged workers." Since these individuals are not counted as part of the labor force (but often can slide in and out of it rapidly), they represent a phenomenon not usually considered when assessing the unemployment/formation relationship. Many individuals not in the labor force fit common definitions of disadvantaged; others are discouraged workers. Both hold great interest to policy-makers.

Finally, Reynolds (1994) argues that new business formation appears to be a necessary, but not sufficient condition for growth. What is the "sufficient" condition? Is it related to the nature of unemployment, i.e., who is unemployed? Is it related to the size of the group of people which transitions in and out of the labor force? The paper does not attempt to assess this set of issues. But they are issues clearly related to those being explored in the current research.

Implicit in each of these policy interests is the nature and substance of the businesses formed. 'What type of businesses are formed? How large are they? What are their chances for success?' If the first policy question is 'what is the employment status of those who form businesses?' The second question must be 'what type of businesses do they form?'



The unit of analysis for the term "unemployment" can focus either on the number of unemployed people (level of unemployment in society), or it can focus on the individual unemployed person. The literature contains both. Research of the first type focuses on the influence of unemployment on business formations in aggregate. It is concerned with populations, not individuals. This type usually associates the unemployment rate with subsequent increases in some measure of business formation. Highfield and Smiley (1987) exemplify this approach.

Research of the second type focuses on the unemployed person forming a new venture. Evans and Leighton (1989) exemplify this approach. Effectively they tie the employment status of individual a at time t to the employment status of individual a at time t+1.

These two units of analysis are quite different. More important, they may yield different results which are not necessarily contradictory. They may also yield similar results which not necessary complementary. Thus, the two strands must be carefully distinguished, and examined separately.

The Relationship of Business Formations to Unemployment

Prior research generally provides evidence of a positive relationship between new business formations and the level of unemployment. Higher unemployment tends to be associated with higher formations rates. However, the evidence is far from unanimous, and occasionally is even contradictory.

Birch rarely addresses unemployment directly, but his work challenges the general view. He argues throughout his published research that high formation rates characterize dynamic locations (1987, for example). Death and contraction are not particularly sensitive to place. With formations critical to employment growth, Birch's dynamic locations effectively are places with low unemployment rates (but considerable turbulence in labor markets). Hence, he infers that business births and low unemployment, or falling unemployment, are tied directly.

Storey (1991) reviewed six studies from the U.K. and two from the United States attempting to relate unemployment and firm formations. These studies differed substantially in methodology and data sets employed, and the sundry variables incorporated into their analyses other than unemployment. However, all used some variant of the first unit of analysis. Summarizing their results, Storey wrote,

The broad consensus is that time series analyses point to unemployment being, ceteris paribus, positively associated with indices of new firm formation, whereas cross sectional, or pooled cross sectional studies appear to indicate the reverse (p. 177).

A cross-national comparison of six purposefully similar studies suggested that unemployment and new venture formation may be related on a regional (sub-national) basis (Reynolds, et. al., 1994, a). While the relationship occurred more often than not, the results were inconsistent across the countries examined. In the U.S., the formation rate was associated with the unemployment level, but not associated with the change in unemployment level (Reynolds, 1994).


The Relationship of Business Formations to Unemployed Individuals

Reynolds, et. al. (1994,b) matched those actively in the process of starting a business with current employment status. He reported that six percent of the nascent entrepreneurs, i.e., people in the active process of starting a firm, interviewed in 1993 were unemployed. Nascent entrepreneurs represented 7.5 percent of the unemployed population, somewhat more than than would be expected if occurring by chance. He also reported on categories of people out-of-the-labor force.* Except for students, these categories appeared less frequently among nascent entrepreneurs than their share in the population. However, those non-employed were preparing twice as many businesses as those unemployed. The proportions were 82 percent employed, 6 percent unemployed, and 12 percent non-employed. Reynolds' samples were comparatively small implying large errors from the mean, but the samples were large enough to produce suggestive results.

Evans and Leighton (1989) examined individuals and their subsequent business formation activity. The authors used approximately 300,00 labor market participants interviewed in the Current Population Survey (CPS) between 1968 and 1987 for whom there were two consecutive years of data. This time series analysis concluded that unemployed workers are about twice as likely to start unincorporated businesses as are employed workers. (Data limitations forced the authors to define owners of incorporated firms as wage and salary workers.) Movement into self-employment from out-of-the-labor force was not explored.

Segal (1996) also examined the individual, and also used the Current Population Survey as his data set. He determined that, proportionally, self-employed people were less likely to have come from unemployment than wage and salary workers. Table 1 provides Segal's data. It shows that 92.5 percent of the people who were self-employed in the years 1984 through 1993 had been employed the prior year. Just 1.7 percent of the self-employed had been unemployed and 5.8 percent had been non-employed. "All workers," i.e., all employed persons, proved less stable than did self-employed persons. Relatively more came into their jobs from unemployment (3.0 percent) and non-employment (6.3 percent), and relatively fewer from employment (90.7 percent).

Eight-three and nine-tenths (83.9) percent of self-employed workers had been employed in their own business the prior year. That means 8.6 percentage points (92.5 minus 83.9) moved from working for someone else to self-employment. The proportion of newly self-employed people -- omitting those moving from self-employment to self-employment -- therefore, is: 8.6 percentage points from employment (53 percent of those shifting status), 1.7 percentage points from unemployment (11 percent of those shifting status), and 5.8 percentage points from outside the labor force (36 percent of those shifting status).

Segal's data offer two salient points: unemployed persons provided a disproportionately large share of people forming new businesses. The proportion is similar to that found by Evans and Leighton. Second, the number of non-employed entering is over three times the number entering from unemployment.

Data from the Current Population Survey as used by Segal and Evans and Leighton has a number of liabilities for present purposes. Two should be noted. The data are collected annually and refer to the longest "job" held. Thus, they do not capture every change in employment status. In year one a respondent may have been out-of-the-labor force for nine months and self-employed for three; in year two that individual could be self-employed for the first five months and out-of-the-labor force for the last seven. His status for statistical purposes is unchanged -- out-of-the-labor force. The effect of this data collection procedure is to minimize status changes. This could be particularly significant in households where secondary wage earners foray into the labor force. Further, people moving from self-employment in one business to self-employment in a second business cannot be determined; statistically, there is no change. Self-employed people are notorious recividists. The figures are, therefore, likely to minimize employment shifts.

The survey also tells us virtually nothing about the business. We don't know if the business is new or mature, large or small, purchased or founded. Data are available on the number of hours worked in the individual's primary activity. These numbers provide some feel for the substance of the venture undertaken. However, the individual could be initiating a business with partners, meaning that his hours may not necessarily reflect the substance of the business or his contribution to its formation, or he could be initiating a solo operation or one employing 50 people while working the same number of hours.

Employement status in preceding and current year for self-employed
people and all workers- current population survey, 1983 - 1993 averages






All Self-Em. All Workers





Out of Labor Force

















Source: Segal's calculations from Current Population Survey. All self-employed calculations by author from Segal.

Finally, owners of incorporated businesses were classified by government statistical agencies as wage and salary workers prior to the mid-1970s. Those using data from the CPS prior to that time could not include incorporated owners in their analyses, e.g., Evans and Leighton. Those using the data set covering later years, e.g., Segal, found incorporated owners classified appropriately.

Previous Page | Main Menu | Next Page



1997 Babson College All Rights Reserved
Last Updated 1/15/97 by Geoff Goldman & Dennis Valencia

To sign-up for the Center for Entrepreneurial Studies' publication lists,
please register with the
Entrepreneurship WebTeam.