Content uploaded by Peter B. Robinson
Author content
All content in this area was uploaded by Peter B. Robinson on Aug 31, 2016
Content may be subject to copyright.
Effect of Education and Experience on Self-Employment Success
THE EFFECT OF EDUCATION
AND EXPERIENCE
ON SELF-EMPLOYMENT
SUCCESS
Journal of Business Venturing
Volume 9, Issue 2, March 1994, Pages 141–156
PETER B. ROBINSON
The University of Calgary
EDWIN A. SEXTON
Virginia Military Institute
EXECUTIVE SUMMARY
The impact of education on the business success of an entrepreneur has been the subject
of much discussion and speculation in both the popular and academic press. The
literature is full of folklore focusing on the high-school drop out who made it big in the
business world armed with an education from the school of hard knocks. Until recently
this was part of the myth surrounding entrepreneurship. The myth takes shape in three
basic areas.
The first looks at the entrepreneur’s level of education relative to the general public. The
second area addresses the effect of education on people becoming entrepreneurs on a
macro level. Do people with higher levels of education start more businesses than people
with less education, does it increase the probability of becoming an entrepreneur? The
third area concerns the micro-economic effect on individual entrepreneurs. Does
education help an entrepreneur succeed?
Past research on education and entrepreneurship consists mostly of institutional studies
at universities with established programs. These offer good support for the outcome of
educational programs. However, these studies are poorly circulated and seldom
published because of the limited sample sizes (McMullan (1988) summarized several
such studies). In this study the literature is reviewed in three areas mentioned above and
new information on the relationships between education, experience, and self-
employment is provided.
The empirical part of this study examines the effect of education and experience using
U.S. census data. Self-employment is used as a surrogate for entrepreneurship and the
analysis controlled for farmers and professionals (medical doctors, lawyers, accountants,
etc.) so that it would more closely fit our conception of an entrepreneur. Earnings
potential was used as a measure of success. We recognize that success is a subjective
experience based on one’s expectations and actual outcomes: however, we believe that
1
Journal of Business Venturing, Volume 9, Number 2, March 1994
Effect of Education and Experience on Self-Employment Success
earnings provided a global indicator of success that is quantifiable relative to the sample
used. Four specific hypotheses were generated and tested using the data.
The first hypothesis (Self-employed have more years of formal education than those who
do not work for themselves.) was confirmed with the years of education for the self-
employed being 14.57years for all workers, 14.71 years for males, and 14.13 years for
female workers. Wage and salaried workers came in nearly one full year lower with:
13.58 years for all worked, 13.73 for male workers, and 13.40 for female workers.
Hypothesis two (The number of years of formal education will increase the probability of
becoming self-employed.) was supported with the probability of becoming self employed
increasing by 08% for each year of education providing a significant relationship (t =
32.11 for all workers, t = 21.9Sfor males ,and t = 20.76 for females, p < .000l for all
three).
Hypothesis three (The relationship between years of formal education and success of the
self-employed, as well as the general population will be positive and significant.) was
supported using the “Beta” coefficients in a “Probit” regression model, indicating that
self-employment and wage and salaried earnings increase significantly for each year of
education, Self-employment earnings increased $1207.63 a year for each year of
education ($1212.76for males and $414.81 for females). Wage and salaried workers
earnings increased $825.99a year for each year of education ($1023.33for males and
$369.37for females).
Hypothesis four (The relationship between experience and self-employment success will
be positive and significant, but weaker than the impact of education.) was supported. All
self-employed workers, both male and female, had over two years more experience than
their wage and salaried counterparts. There is a strong positive relationship between
self-employment and both experience and earnings with the exception of self-employed
females whose experience did not significantly impact their earnings.
In conclusion, a general education has a strong positive influence on entrepreneurship in
terms of becoming self-employed and success. Experience has a similar relationship
although not as strong. Future studies need to examine the impact of specific types of
education, such as business school or entrepreneurship classes, on the entrepreneurial
outcomes in the studies.
INTRODUCTION
A great deal of time, money, and effort has gone into the education of entrepreneurs.
Reports on the number of schools offering courses in small business and entrepreneurship
vary from 418 (Solomon and Femald 1991) to over 200 (Vesper 1985). Another study
found 176 schools offering undergraduate courses and 117 offering graduate courses in
major U.S. universities (Robinson and Haynes 1991), and 37 schools offering
undergraduate and 18 offering graduate courses in Canada (Robinson and Long 1991). In
terms of resources, Katz (1991) reported 102 endowed positions (both chairs and
professorships) in entrepreneurship and free enterprise.
With all of this activity, there is still the nagging question: are we doing anyone any
2
Effect of Education and Experience on Self-Employment Success
good? Does education have an affect on students becoming entrepreneurs, and are
educated entrepreneurs better able to succeed in today’s business environment? Again, a
great deal of time and effort have gone into research to calm our troubled minds and
ferret out the answers to these questions. Cooper and Cascon (1992) recently reviewed
the impact of “level of education” on entrepreneurship performance finding a generally
positive but mixed set of results based on 17 studies using a variety of measures of
performance and methods of analysis, and so the questions persist.
Three specific areas have been addressed in the past. The first derives from the old myth
of the highly successful entrepreneur who dropped out of school and by his own cunning
and hard work, founded and grew a large company. The basic question is, are
entrepreneurs as educated as the general public? The second area addresses the macro-
economic benefits to the surrounding education in terms of job creation and new firm
development. The question here is, are educated people (both general education, and
more specifically entrepreneurship programs) more likely to start and manage their own
businesses than less-educated people? The third area concerns the micro-economic effect
on individual entrepreneurs. Does education help an entrepreneur to succeed? This study
will review past literature in each of these areas and present new information on the
relationship between education and experience (a type of informal education) and self-
employment.
THE EDUCATED ENTREPRENEUR
The myth of the poorly educated entrepreneur began with the Horatio Alger stories and
other anecdotal evidence. It first found empirical support in a study of “light
manufacturers of hard goods” established in post-WWII Michigan (1945-1958) (Collins
and Moore 1964, p. 29). Being the first major study of its type, even with a very
restricted sample, the myth was supported and perpetuated. Since its publication, other
studies have addressed the issue of education and entrepreneurship with contradictory
findings. Douglass (1976) directly addressed the myth in a study that concluded that
while entrepreneurs may have been poorly educated in the past it is no longer the case.
Douglass cited research by Mayer and Goldstein (1961) and Collins and Moore (1964) to
demonstrate a clear trend showing that “the formal educational level of entrepreneurs has
been rising over the past fifteen years,” and while the proportion of people in the U.S.
population holding college degrees increased from 7.5% to 10.7% from 1960 to 1970,
college educated entrepreneurs increased from 9% to 37% from 1961 to 1975.
These findings and their conclusions have been supported by Thompson (1986) in the
Canadian portion of an international study. The Canadian entrepreneur had an average of
13 years of formal education with approximately 20% of the sample having 10 or fewer
years of education and over 33% reporting over 15 years of education. Cooper and
Dunkelberg (1987) also reported a U.S. sample of entrepreneurs with significantly higher
levels of education than the general population.
This study will also present data to support the fact that the self-employed as a group
have more education than wage and salaried workers, thus adding to the already
established weight of evidence refuting the negative stereotype of entrepreneurs being
relatively uneducated. Indeed, contrary to the Collins and Moore study, a much broader
empirical base was used here to demonstrate that new venture development is not just an
3
Journal of Business Venturing, Volume 9, Number 2, March 1994
Effect of Education and Experience on Self-Employment Success
option for the social and educational non-achievers (those than can work do and those
who can’t become self-employed), but is a viable career option for the more-educated
segment of society.
MACRO-ECONOMIC IMPACT
When considering past research, two distinct types of outcome have been measured. The
first type consists of student perceptions of the efficacy of the program. This may include
general feedback items as well as indications of intentions and expectations of future
business activity. Examples of this include studies from Boston University (Brush 1989),
Babson College (Homaday and Vesper 1981; Homaday and Vesper 1982; Homaday
1985), the University of Windsor (Kantor 1988), The University of Western Ontario
(Knight 1987), and The University of Calgary (McMullan and Boberg 1991). The results
typically indicate a positive disposition (ratings in the 3 to 5 range of a 5-point scale) with
a large proportion (usually over 30%) stating their intention to start business in the future
and/or their expectation to do so in the near future (about 5 years).
The second type of outcome takes a quantitative approach. Here the researcher is
concerned with the number of participants that go on to become business owners, the
number of jobs created, and/or the increase in performance as measured by sales,
profitability, or market share. Studies of this type started as early as 1978 when Coulton
(1978) reported research involving three innovation centers. Other studies along this line
include Clark, Davis, and Hamish (1984), Conner (1985), Homaday and Vesper (1982),
Robinson (1988), Stevenson (1983), and Watkins and Morris (1981).1
When considering the quantitative outcome of education on new venture start-ups,
employment and growth, several factors must be considered. First is the type or
specialization of the educational program. Studies vary on the type of education from a
general post-secondary education to business school alumni and on the most specific
level, participation and/or graduation from an entrepreneurship program.
Next to the specificity of the education experience is the intensity of the program.
Intensity is an indication of the amount of exposure to a particular topic or subject area
(in this case entrepreneurship) over time. A general business education would not provide
as intense an experience in new venture development as participating in an
entrepreneurship course, and one course would not be as intense as a concentration or
major. The extreme ends of a continuum of entrepreneurship education intensity would be
a general “liberal arts” program at one end and an “in residence” multi-sessional
entrepreneurship program at the other end.
The final factor to consider in program evaluation is the screening of participants. This
varies from self-education where enrollment is open, such as a university, to a highly
targeted program where applicants are screened, such as the Australian Enterprise
workshops (Wan 1988) or the University of Manchester (Watkins and Morris 1981).
Outcomes of entrepreneurship education programs vary widely according to these
factors.
McMullan (1988) summarized five studies that analyzed the business development by
1 The more recent studies have not been published primarily because they represent
institutional research that is no longer acceptable in the scholarly journals in this field
4
Effect of Education and Experience on Self-Employment Success
participants in entrepreneurship programs. The results indicate a broad range of outcomes
beginning with a single extracurricular course resulting in a 14.5% increase in the number
of new ventures (Clark, Davis, and Hamish 1984) and, at the high end, 76% (Watkins and
Morris 198 1) of the participants in the intensive, highly screened part-residential
program at the University of Manchester. In addition, McMullan cited research indicating
that each new venture created an additional 3 to 11 jobs. The factors affecting the
variation in job creation is not clear from the information provided. In a published study
not covered by McMullan, Wan (1988) found that 28% of the graduates of Australia’s
Enterprise Workshop became self-employed.
Stevenson (1983) considered general business school alumni and found that between 11%
and 35% of the graduates from the Harvard School of Business were self-employed. The
proportion varied depending on the number of years from graduation the former students
were at the time the survey was administered. Robinson (1988) considered the alumni at
The Wichita State University, Barton School of Business and found 15.36% were full-
time business owners while an additional 6.06% owned a part-time business.
MICRO-ECONOMIC IMPACT
Many of the studies mentioned earlier have attempted to shed light on the impact
entrepreneurship has for the individual. The case has been made for the importance of
education for the individual in terms of fostering the creative spirit (McMullan and Long
1990), increasing awareness of the process and options (Hills 1988), and contribution to
the special qualities of the entrepreneur (McMullan, Long and Wilson 1985; Ronstadt
1987). These have primarily been theoretical, conceptual or anecdotal. Price (1992) went
beyond postulating about the effect of education on the self-employed and provided
empirical data on the increased employment and increased sales following an intensive
outreach program aimed at the women and minority owner-managers of small firms.
PURPOSE
The current study examines the relationship between formal education, experience and
self-employment, as well as other moderating variables related to an individual’s life
experience. In addition, it provides a base line for comparison with other outcome studies
from the general U.S. population in terms of the probability or proportion of self-
employed, their levels of education, and their relative earnings. The study also sheds
some light on the effect education (both formal and experience) has on the success of the
self-employed by examining several specific hypotheses using both linear multiple
regression models and a nonlinear regression technique called “Probit.” Where regression
was used in the analysis a correlational relationship was demonstrated, which, although it
does not demonstrate causality, may suggest causation-based data structure and
supporting analysis. In the remainder of this study we will define the terms and variables
used in the analysis, describe the research methodology used, provide detailed results,
and discuss the implications of those results.
DEFINITIONS
We defined a “self-employed person” as one whose primary job was in the self-
employment sector (Borjas 1986). While other definitions exist, alternative ways of
defining this variable did not lead to significantly different results from what are reported
5
Journal of Business Venturing, Volume 9, Number 2, March 1994
Effect of Education and Experience on Self-Employment Success
here. We also realize that while not all self-employed workers are necessarily
entrepreneurs, this variable will eliminate most of those who are not entrepreneurs,
leaving a composite of entrepreneurs and small business owners. Unfortunately the
census data are not specific enough for further differentiation using more rigorous
definitions.
The exact proportion of those who are entrepreneurs and those who are small business
owners would be subject to definitional dispute. When does one stop being an
entrepreneur and become a small business owner, or vice versa? We were saved this long
argument by the nature of the data. Although it is not pure entrepreneurial sample, the
result can be considered a conservative representation of that population on some
variables and can give some general parameters for other variables.
Success in this study was defined as earning potential for both self-employed and wage
and salaried workers. We recognize that success is a subjective experience based on the
congruence of one’s expectations and the actual outcomes; however, we believe that
earnings provides a global indicator of success that is quantifiable relative to the sample
used. If, as has often been reported, money is a way of keeping score for entrepreneurs,
we can use their scorecard to indicate relative levels of success. In addition, more
subjective measures were not available through the census data. While this is a useful
measure of success relative to other variables, we would not wish to imply that because
the self-employed have greater annual earnings that they are more successful than the
salaried sector. Again, the subjective evaluation of success must be taken into account.
Education in this study represents the years of formal education a person had
obtained at the time the data was gathered. Because of the limitations of the data
set, the type of education cannot be specified beyond the number of years
completed. Generalizing the results of this empirical study examining general
education to outcomes specific to entrepreneurship education might be stretching
the findings. However, when examined in the context of the literature cited earlier,
general education can be considered an anchor on one end of a continuum of
educational specificity, with the other end being an in-depth residency program in
entrepreneurship, extended over a considerable period of time (see page 145).
Experience was included as a variable because of its close tie to education. It is
defined as the number of years an individual has been able to work after
completing his or her education. Two assumptions upon which this variable must
depend are (1) most people will have spent the time since completing their
education in productive pursuits, either employed or self-employed, and (2) people
tend to learn through those productive pursuits. There is no assumption of the type
of experience gained inherent in the study, thus, the experience may or may not be
related to the current occupation of either the self-employed or wage and salaried
workers. Because of the possible variation in experience and an operational
definition based on time after education available (see Table 1) we expect the
effect of experience on earnings and the probability of self-employment to be
similar to education but with less intensity.
6
Effect of Education and Experience on Self-Employment Success
Being a time-based variable, experience, as defined here, would likely correlate
highly with age and maturity. These could also be used as surrogates for
experience based on the assumptions listed above. The definition incorporated into
the study was selected because of use in past economic research; it was closer to
job-related experience than age, and it was measurable with the data available
where maturity was not.
TABLE 1: DEFINITON OF VARIABLES
EDUCATION = Years of education completed
EXPERIENCE = Age – Education - 5
MARRIED = 1 if individual is currently married and 0 otherwise
OTHER MARITAL = 1 if divorced or widowed and 0 otherwise
SINGLE = 1 if never been married and 0 otherwise
CHILDREN = 1 if own children am present and 0 otherwise (males)
CHILDREN = 1 if own children under the age of six are present and 0
otherwise (females)
HOURS WORKED = The typical number of hours worked per week
WEEKS WORKED = The number of weeks worked in the survey year
OTHER INCOME = Total family income - individual’s income measured in dollars
OCCUPATION = 11 occupational dummy variables controlling for Professionals,
Managers, Technicians, Sales, Clerical, Service, Mechanical,
Construction, Agriculture Related, Production, and Laborers.
INDUSTRY = 10 industry dummy variables controlling for Agriculture,
Manufacturing, Transportation and Communication, Wholesale
Trade, Retail Trade, Finance, Business Services, Persona1
Services, Construction, and Professional
Et = Annual wage and salary earnings measured in dollars
Es = Annual self-employment earnings measured in dollars
SE = 1 if the individual is self-employed and 0 of the individual is
employed in the wage and salary sector
Table 1 provides a more-detailed list of variables along with their operational definitions
and coding. Many of these codes are taken directly from census data while others are a
product of our data manipulation. These will be of greater interest when the discussion
leads to the issue of testing one variable while holding all others constant in order to
control for their effects.
7
Journal of Business Venturing, Volume 9, Number 2, March 1994
Effect of Education and Experience on Self-Employment Success
METHODS
Sample
We employ the Public-use B sample of the 1980 U.S. Census of Population. This sample
contains individual data from the largest Standard Metropolitan Statistical Areas
(SMSAs) in the U.S. and has the full range of economic and demographic information on
each observation. We limit the sample by including only 23- to 64-year-old non-student
workers who either worked for wages and salary in 1979 or who were self-employed in
that year. A proportional or stratified random sample was used in the analysis with sample
sizes listed in each table.
Empirical Model.
Our analysis consists of a wage and salary earnings function (El), a self-employment
earnings function (E2), and a self-employment decision function (SE). The specifications
of these functions are as follows:
El=XlBl +ul (1)
El = X2B2 + u2 (2)
SE=ZY +e (3)
where the vectors Xl, X2, and Z represent independent variables available in our sample
that affect wage and salary earnings, self-employment earnings, and the decision to
choose self- employment, respectively. Bl , B2, and Y are coefficient vectors that must be
estimated and u1, u2, and e represent random errors. The two earnings functions,
equations (1) and (2), will be estimated using a multiple regression technique. Equation
(3), the self-employed decision function, will be estimated using a nonlinear regression
technique called “Probit.” It is a nonlinear function because the dependent variable (SE)
is a binary variable equal to 1 if the worker is self-employed and equal to 0 if he or she is
employed for wages or salary. The definitions of the variables used in the estimation of
the above equations are found in Table 1 and the means of selected variables are
presented in Table 2 for both self-employed and wage and salary workers.
A summary of some of the significant characteristics of this sample is shown in Table 3.
We present the probability of being self-employed, and the average annual earnings in
both the self-employed and wage and salary sectors for (1) all workers, (2) males, and (3)
females. It is immediately apparent that both the probability of being self-employed and
the compensation for self-employment vary widely as do earnings in the wage and
salaried sector. This leads us to explore some of the factors that influence the probability
of one’s becoming self-employed and one’s success at self-employment.
TABLE 2 Mean of Selected Variables for Self-Employed and Wage and Salaried Workers
Self-Employed Wage and Salary Workers
8
Effect of Education and Experience on Self-Employment Success
Variable All Male Female All Male Female
Education 14.57 14.71 14.13 13.58 13.73 13.40
AGE 44.13 44.31 43.57 40.95 41.16 40.68
EXPERIENC
E24.56 24.60 24.44 22.37 22.43 22.28
MARRIED 0.81 0.82 0.77 0.71 0.76 0.64
CHILDREN* 0.48 0.48 0.16 0.47 0.48 0.15
CC-RES 0.40 0.40 0.40 0.43 0.41 0.44
CC-WORK 0.34 0.36 0.29 0.40 0.41 0.38
HOURS
WORKED 41.01 44.50 30.37 39.27 42.34 35.31
WEEKS
WORKED 43.68 45.79 37.28 45.62 47.80 42.81
OTHER
INCOME 11,734 8,659 21,078 11,137 7,481 15,840
n21,352 16,065 5,287 159,804 89,917 69,887
* The CHILDREN variable measure the proportion of moles who have children of any age at home and
the proportion of females who hove children under the age of six at home
Hypothesis
Based on past research in education and entrepreneurship several hypotheses were
developed and tested. In addition, as the research progressed several interesting
relationships emerged that will be discussed later in the paper. The initial hypotheses that
were generated and tested are as follows:
Hl: Self-employed individuals will report more years of formal education than those who
do not work for themselves.
H2: The number of years of formal education will increase the probability of
becoming self-employed.
H3: The relationship between years of formal education and success of the self-
employed, as well as the general population, will be positive and significant.
H4: The relationship between experience and self- employment success will be
positive and significant, but weaker than the impact of education.
RESULTS
9
Journal of Business Venturing, Volume 9, Number 2, March 1994
Effect of Education and Experience on Self-Employment Success
The Self-Employment Decision.
Equation (3) above represents the self-employment decision function. Estimating this
function using a nonlinear probit model we can determine how certain variables affect the
self-employment decision. Table 4 reports the estimated partial derivatives of the
probability of being self- employed with respect to each of the independent variables
(holding the other variables constant). The coefficient represents the “beta weight” in the
nonlinear regression equation and as such it is analogous to the slope of the variable’s
vector. The t ratio tests whether the coefficient is significantly different from zero (a slope
of zero or no effect).
For example, the coefficient on EDUCATION in each of the three probit regressions
reported would be interpreted as saying that each additional year of education increases
the probability of being self-employed by approximately one percent (0.8%). The t ratio
indicates that this is highly significant (p < .000l) relative to a coefficient of zero.
TABLE 3: Summary Statistics on Probability of Self-Employment and Average Salaries
by Sector
Groups Self-Employment
Probability Annual Earnings
S-E Sector Salaried Sector
All Workers 0.118 $19,471.51 $14,568.45
MALE 0.152 $23,343.98 $19,170.03
FEMALE 0.070 $7,704.68 $8,648.04
In the first regression, using the entire sample, married males are the reference group.
Several findings are of interest. We see that the following variables have a significant
positive influence on an individual’s probability of being self-employed: level of
education, potential work experience, having children, and having sources of income
within the family other than your own income. In addition, the following variables have a
negative impact on the probability of being self-employed: being single, widowed, or
divorced as compared to being married.
Separate “Probit” regressions were also performed for males and females. On each of the
significant variables the signs of the coefficients were the same, although the magnitudes
varied somewhat. This indicates that the factors that influence men and women to
become self-employed move in the same direction. This is not to say, however, that they
move in that direction for the same reasons. For example, the presence of children has a
positive influence on the self-employment decision of both men and women. However, it
is very likely that this positive influence is for different reasons. For example, Birley
(1989) cited research that found that parenthood was more important than marriage as a
factor affecting a woman’s decision for self- employment because of the flexibility it
allows in scheduling work, whereas children add to a man’s stability.
Self-Employment Earnings.
10
Effect of Education and Experience on Self-Employment Success
The estimated regression coefficients for Equations (2) and (3) are reported in Table 5.
Again, the coefficients can be interpreted as partial derivatives of self-employment
earnings with respect to each of the independent variables. For example, in the case of the
EDUCATION variable we find that, holding all else constant, an additional year of
education is expected to result in an increase in self-employment earnings of $1,207.63 in
the ALL regression, $1,212.76 in the MALE regression, and $414.81 in the FEMALE
regression. In addition it will result in an increase in wage and salaried income of
$825.99 in the ALL regression, $1023.33 in the MALE regression, and $369.00 in the
FEMALE regression. In general, those variables with a positive sign are contributing to
the success of the self-employed worker (through higher earnings) and those variables
with negative coefficients are detracting from his or her success. Each of the coefficients
is of the expected sign and all but a few are statistically significant.
In the separate regressions performed for males and females we find several interesting
results. First, as noted above, the returns on education (in terms of increased earning
potential) are nearly three times higher for men than for women. Also of note is the fact
TABLE 4: Probit Regressions Indicating the Impact of Selected Variables on the
Probability of Self-Employment
Variable All Workers Males Females
EDUCATION .008 (32.11) .008 (21.95) .008 (20.76)
EXPERIENCE .002 (38.77) .003 (33.43) .002 (20.78)
OCCUPATION aaa
INDUSTRY bbb
OTHER
MARITAL
-.021 (- 6.39) -.008 (- 2.49) -.014 (- 6.25)
SINGLE -.023 (-10.38) -.021 (- 6.12) -.028 (- 9.09)
CHILDREN .014 (10.40) .019 (9.19) .014 (6.23)
OTHER
INCOME
.00054(10.84) .00044 (4.84) .00053 (9.73)
n-SALARIED 159,804 89,917 69,887
n-SELF-
EMPLOYED
21,352 16,065 5,287
* The dependent variable = 1 if self-employed and 0 otherwise. The asymptotic t ratios
we in parentheses.
a Equation includes dummy variables for 11 occupations.
b Equation includes dummy variables for 10 industries.
that while widowed, divorced and single men have significantly lower self-employment
earnings than married men, these same groups for women have either significantly higher
or at least no lower self-employment earnings than married women. Finally, the presence
of children has a significantly positive influence on the self-employment earnings of men
11
Journal of Business Venturing, Volume 9, Number 2, March 1994
Effect of Education and Experience on Self-Employment Success
and a negative but insignificant impact on female self-employment earnings (see also
Birley 1989).
Hypotheses
Specific results related to the hypotheses to be tested confirmed the suspected
relationships between education and entrepreneurship. The first hypothesis (HI: Self-
employed have higher levels of education than those who do not work for themselves.) is
supported in Table 2 which shows the mean level of education for self-employed as:
14.57 years for all workers, 14.71 years for males, and 14.13 years for female workers.
Wage and salaried workers came in nearly one full year lower with: 13.58 years for all
workers, 13.73 for male workers, and 13.40 for female workers.
Hypothesis two (H2: The general level of education will increase the probability of
becoming self-employed.) was tested and supported in Table 4. The probability increased
by 0.8% for each year of education providing a significant relationship (t = 32.11 for all
workers, t = 21.95 for males, and t = 20.76 for females, p c .0001 for all three).
Hypothesis three (H3: The general level of education will increase the success of the self-
employed, just as it increases success in the general population.) was tested and
supported in Table 5. The “Beta” coefficients indicate that self-employment and wage and
salaried earnings increase significantly with each year of education. Self-employment
earnings increased $1207.63 a year for each year of education ($1212.76 for males and
$414.81 for females). Wage and salaried workers earnings increased $825.99 a year for
each year of education ($1023.33 for males and $369.37 for females). The beta
coefficients represent the slope of a vector representing the relationship between
education and earnings, in each case the slope represents a significant positive
relationship as tested using a t test between the slope and a slope of 0 (All self-employed
t = 23.67, male self-employed t = 19.14, female self-employed t = 6.48, all wage and
salaried t = 90.22, male wage and salaried t = 75.2, and female wage and salaried t =
39.07, p < .000l for all groups).
Hypothesis four (H4: Experience will have a similar relationship to self-employment as
does education with a weaker overall impact) is supported in Tables 3, 4, and 5. All self-
employed workers, both male and female, had over two more years of experience than
their wage and salaried counterparts. Table 4 showed a strong positive relationship
between self-employment and experience with t values exceeding 20.78 and p < .0001.
Table 5 shows a similar positive relationship with earnings with the exception of self-
employed females whose experience did not significantly impact their earnings (t = .95, p
> .10). All other relationships were significant at p c .0001.
TABLE 5: Earnings Functions for Self-Employed and Wage and Salaried Workers Self-
12
Effect of Education and Experience on Self-Employment Success
Employed Workers
Self-Employed Workers* Wage and Salaried Workers**
Variable All Male Female All Male Female
EDUCATION 1207.63
(23.67)
1212.76
(19.14)
414.81
(6.48)
825.99
(90.22)
1023.33
(75.20)
369.37
(39.07)
EXPERIENCE 571.19
(12.21)
696.03
(11.96)
56.76
(.95)
389.27
(52.56)
601.08
(52.10)
69.00
(9.57)
OCCUPATION a a a a a a
INDUSTRY b b b b b b
HOURS
WORKED
107.37
(12.77)
89.80
(8.21)
115.51
(12.68)
138.19
(66.60)
127.34
(36.78)
131.58
(70.22)
WEEKS
WORKED
239.91
(24.29)
323.05
(23.05)
118.12
(12.77)
219.13
(124.29)
294.24
(86.67)
168.15
(117.35)
OTHER
MARITAL
-956.20
(-2.40)
-948.74
(-3.56)
1603.48
(4.13)
-726.50
(-12.70)
-1347.27
(-12.62)
532.41
(11.31)
SINGLE -3536.72
(-7.02)
-3788.18
(-6.13)
170.89
(.27)
-1573.35
(-22.71)
-2523.80
(-22.70)
1009.42
(15.92)
CHILDREN 2575.18
(8.68)
2781.32
(7.47)
-369.43
(-.80)
781.46
(16.30)
1112.72
(14.56)
-155.60
(-2.60)
R2 .33 .28 .21 .49 .37 .41
F 318.77 205.78 45.69 4655.12 1703.52 1572.11
n 21,352 16,065 5,287 159,804 89,917 69,887
* The dependent variable is annual self-employment income.
** The dependent variable is annual wage and salary income. The subjects t-ratios testing
the difference between the betas coefficient and 0 is in parentheses
a Equation includes dummy variables for 11 occupations.
b Equation includes dummy variables for 10 industries
DISCUSSION
Comparing Self-Employed and Wage and Salary Earnings.
Because most of the findings for our Wage and Salary earnings function are similar to
what we found in the Self-Employed case we will not review them in any detailed way
(see Table 5). Rather, it might be useful to note some interesting comparisons between the
13
Journal of Business Venturing, Volume 9, Number 2, March 1994
Effect of Education and Experience on Self-Employment Success
two types of earnings.
First, it is interesting to note the effect of education and experience on the earnings
function of self-employed and wage and salaried workers. It would have been useful to
have a statistical test of the difference between the coefficients; however, since these are
not measures of central tendency or variance, no test could be found to test the difference
between Bl and 82 in a linear regression. It should be remembered that these represent the
slopes of earnings vectors as defined in formula (1) and (2). As X, and X, increase so do
E, and E, respectively. Therefore, the difference in earnings increases with the years of
education and/or experience (see Figure 1). The net result is that although education is
important for wage and salaried workers, it is even more important to the success of the
self-employed as indicated by their earnings or more precisely the slope of the earnings
vector.
A second point we wish to discuss is that the average earnings for self-employed males
($23,343.98) are higher than average earnings for wage and salary males ($19,170.03),
yet for females the opposite is true with average self-employment earnings at $7,704.68
and average wage and salary earnings at $8,648.04. The difference between male and
female earnings was consistent with the literature (Hisrich 1981) and is usually accounted
for by a variable dealing with the type and scope of the business. Perhaps of greater
14
Effect of Education and Experience on Self-Employment Success
importance in this study is identifying other situational factors in the social environment
that help account for the differences.
Several factors stand out in importance. First, both the mean level of education and the
returns to education are higher for self-employed workers. This is true for both males and
females. A number of explanations exist. First, education tends to increase one’s sense of
efficacy and self-esteem, which in turn increases one’s ability to perceive opportunities
and pursue them. In addition, highly educated workers are likely to possess more
information about self-employment opportunities and are probably better able to assess
their chances at success in this sector. This might account for the higher mean level of
education in the self-employed sector. The higher returns to education among the self-
employed might be due to their ability to perceive where they are able to receive the
highest return on their human capital investment and choosing that sector. We might
expect those individuals who do not feel that the wage and salary sector is giving them an
adequate return on their investment in education to self-select into self-employment.
Another significant factor is the influence of children on the self-employed. In standard
human capital models, the argument is that anything that makes a worker more
productive will increase his or her earnings. For men, the presence of children is
supposed to be a stabilizing influence in their lives. A more stable worker is a more
productive worker and, therefore, we would expect that, all else held constant, men with
children would earn more than men without children. Children have traditionally had the
opposite effect on women. Children in general, and young children in particular,
represented interruptions in the women’s labor force participation (Birley 1989; Bowen
and Hisrich 1986). During these interruptions her human capital was allowed to
depreciate, her productivity declined, and so did her earnings. While this standard
argument holds true in both the wage and salary case and the self- employment case, we
observe a curious phenomenon. The “returns” to children are over twice as high for self-
employed males and the “penalty” for children is over twice as large for self-employed
females.
We recall from a previous section that the presence of children increases the probability
of self-employment for men and women. These same children have the effect of
increasing their father’s earnings and decreasing their mother’s earnings. Women with
children often choose self-employment for the flexibility it allows (Birley 1989). This
same flexibility can easily be translated into more interrupted careers and, via the human
capital explanation, lower earnings. Stoner, Hartman and Arora (1990) further document
the dynamics of the situational variables for women self-employment in terms of work-
home role conflicts and their effect on job and life satisfaction.
Referring back to Table 3, we see that for males the self- employed tend to be older, to be
more experienced, more likely to be married, and they tend to work more hours per week
than the wage and salaried. These things would all tend to produce higher earnings in the
self-employed. For women, the self-employed tend to be more likely to be married, more
likely to have young children, work fewer hours per week, and work fewer weeks per
year. These things would all lower the self-employment earnings of women relative to the
female wage and salary earnings. It is therefore reasonable that we find higher self-
15
Journal of Business Venturing, Volume 9, Number 2, March 1994
Effect of Education and Experience on Self-Employment Success
employment earnings for men and the opposite for women. This finding is also supported
in the literature (Hisrich and Brush 1984; Hisrich 1986).
Finally, from Table 3 under the probability of being self-employed we find that the
probability of being self-employed for “all workers” is .118, or in other words, 11.8% of
the workers are self-employed. This figure combined with the probit regression
coefficient for “education” from Table 5, makes it possible to establish a base line for the
comparison in educational studies. For example, for each year beyond the 13.58 years of
basic education (Table 1, education level of wage and salary workers) the probability of
self-employment increases by .8%. Thus, graduates of four-year post-secondary programs
should have 13.74% probability of self-employment [(16 years-l 3.58 years) X .008].
This is very close to the range found by Robinson at The Wichita State University of
15.36% for business school alumni and Stevenson (1983) for the 1977 class of Harvard
business school graduates at 11 .O%, but lower than Harvard business school alumni
from other years.
CONCLUSIONS
Education does indeed have a close relationship to entrepreneurship, in that entrepreneurs
(self-employed) do have a higher level of education than those in the wage and salaried
sector.
In addition, higher levels of education increase both the probability of becoming self-
employed and the success of individuals in that sector in terms of the earnings.
Experience is similar to education in its relationship with self-employment, the difference
being in the intensity of the effect on both probability and success. One conclusion is,
therefore, to get an education and then experience.
One major shortcoming of the study was our inability to study the effect of specific types
of education or educational programs such as business school or entrepreneurship
programs as opposed to general levels of education. We have pieced together enough of
the literature that it appears that as the specificity of the educational experience increases
so too does its effect on increasing the probability of self-employment. We cannot
generalize this conclusion to the success or earnings of those programs. That is a question
for further research. However, this does add to the body of knowledge refuting the
stereotype of the uneducated entrepreneur and showing entrepreneurs as being highly
educated with a significant positive relationship between education and the
entrepreneurial outcomes.
There are distinct gender differences in both the probability of becoming self-employed
and the earnings. A good deal of the differences can be accounted for by the situational
factors such as education, experience, the actual time working (both hours worked per
week and weeks worked per month) and role demands within the familial context. In
addition, the results presented here shed some light on the often misquoted belief that
there are more women entrepreneurs than male entrepreneurs or that more women are
16
Effect of Education and Experience on Self-Employment Success
starting businesses than men.2 At the time the data was collected (1979), the probability
of males being self-employed was over twice that of females and the overall numbers
support the proportions. If current trends persist, over time, the probability of males and
females being self-employed should converge.
Finally, back to education, it seems that education is the key to bettering one’s position in
life. The earnings function is positive and significant for all the groups tested. On the
whole, the effect is much stronger and more positive than any of the other variables
examined in this study with the exception of children, which is positive with a high
magnitude for males and negative with a high magnitude for females. The effect of
education and experience increase over the span of one’s career as the levels of
experience increase. It is, therefore, critical that one carefully plan the specifics of the
education in terms of adding experience and selecting programs based on the intensity of
the program and the pedagogy involved.
REFERENCES
Birley, S. 1989. Female entrepreneurs: are they really any different? Journal of Small
Business Management 27( 1):23-37.
Borjas, G.J. 1986. The self-employment experience of immigrants. Journal of Human
Resources, 8(4)&X-506.
Bowen, D.D., and Hisrich, R.D. 1986. The female entrepreneur: A career development
perspective. Academy of Management Review, 11(2):393-407.
Brush, C.G. 1989. Entrepreneurship Courses at Boston University: Perceptions of
Students Graduating Between 1985-1988. Working Paper, Boston University.
Clark, B.W., Davis, C.H., and Harnish, V.C. April 1984. Do courses in Entrepreneurship
aid in new venture creation? Journal of Small Business Managemenr: 263 1.
Collins, O., and Moore, D. 1964. The Enterprising Man. Bureau of Business and
Economic Research, Michigan State University, East Lansing, MI.
Conner, P. C. 1985. The facilitation and stimulation of entrepreneurship of young
persons in Ireland through the youth enterprise program. In J.A. Homaday, E.B.
Shils, J.A. Timmons, and K.H. Vesper, eds., Frontiers of Entrepreneurship
Research. Wellesley, MA: Babson College, pp. 588-608.
2 The “rate of increase” of women starting their own business is greater than that of men.
In other words, the number of women starting businesses is increasing faster than the
increase in tbe number of men starting businesses. This is due to the large differential
between the number of male and female business owners. A slight increase in a small
pool will represent a larger proportional change than a large increase much larger pool.
17
Journal of Business Venturing, Volume 9, Number 2, March 1994
Effect of Education and Experience on Self-Employment Success
Cooper, A. C., and Cascon, F.J.G. 1992. Entrepreneurs, processes of founding, and new-
firm performance. In D.L. Sexton and J.D. Kasarda, eds., The State of the Art of
Entrepreneurship. Boston, MA: PWS-Kent Publishing Company, p. 301-340.
Cooper, A.C., and Dunkelberg, W.C. Winter 1987. Entrepreneurial research: Old
questions, new answers and methodological issues. American Journal of Small
Business: 1l-21.
Colton, R.M. November 1978. Technological innovation through entrepreneurship.
Engineering Education, 69(2): 193-196.
Douglass, M.E. December 1976. Relating education to entrepreneurial success. Business
Horizons: 40-44.
Hills, G. 1988. Variations in university entrepreneurship education: An empirical study
of an evolving field. Journal of Business Venturing 3: 109-122.
Hisrich, R.D. 1986. The woman entrepreneur: Characteristics, skills, problems, and
prescriptions for success. In D.L. Sexton and R.W. Smilor, eds., The Art and
Science of Entrepreneurship. Cambridge, MA: Ballinger Publishing Company,
pp. 61-81.
Hisrich, R. D., and Brush, C. January 1984. The woman entrepreneur: Management skills
and business problems. Journal of Small Business Management: 30-37..
Homaday, J. A. 1985. Alumni opinions of the value of the curriculum in general business
and in entrepreneurship courses. J.A. Homaday, E.B. Shils, J.A. Timmons, and
K.H. Vesper, eds., Frontiers of Entrepreneurship Research. Wellesley, MA:
Babson College, pp. 56&565.
Homaday, J. A., and Vesper, K.H. 1981. Alumni perceptions of entrepreneurship courses
after six to ten years. Entrepreneurship Education, Waco, TX: Baylor University.
Homaday, J.A., and Vesper, K.H. 1982. Entrepreneurial education and job satisfaction. In
J.A. Homaday, J.A. Timmons, and K.H. Vesper, eds., Frontiers of
Entrepreneurship Research. Wellesley, MA: Babson College, pp. 526-539.
Kantor, J. Spring 1988. Can entrepreneurship be taught ?: A Canadian experiment.
Journal of Small Business and Entrepreneurship: 12-l 9.
Katz, J.A. 1991. Endowed positions: Entrepreneurship and related fields.
Entrepreneurship: Theory and Practice 15(3):53-68.
Knight, R. M. Summer 1987. Can Business Schools Produce Entrepreneurs? An
Empirical Study. Journal of Small Business and Entrepreneurship: 17-26.
Mayer, K., and Goldstein, S. 1961. The First Two Years: Problems of Small Firm Growth
and Survival. Washington, DC: Small Business Administration.
18
Effect of Education and Experience on Self-Employment Success
McMullan, C.A., and Boberg, A. L. 1991. The relative effectiveness of projects in
teaching entrepreneurship. Journal of Small Business and Entrepreneurship 9(1):
14-24.
McMullan, WE. Fall 1988. The economics of entrepreneurship education. Journal of
Small Business and Entrepreneurship, 6(1): 8-l 8.
McMullan, WE., and Long, WA. 1990. Developing New Ventures: The Entrepreneurial
Option. New York, NY: Harcourt Brace Jovanovich, Publishers.
McMullan, W. E., Long, W. A., and Wilson A. Summer 1985. MBA Concentration in
Entrepreneurship. Journal of Small Business Management: 18-22.
Robinson, P.B. 1988. The Alumni of the Barton School of Business. Research Report,
Wichita, KS: The Center for Entrepreneurship, Wichita State University.
Robinson, P.B., and Haynes, M. 1991. Entrepreneurship education in America’s major
universities. Entrepreneurship: Theory and Practice 15(3): 41-52
Robinson, P. B., and Long, W. A. 1992. Entrepreneurship education in Canadian
universities. Journal of Small Business and Entrepreneurship 10(1):6 l-69.
Ronstadt, R. Spring 1987. The educated entrepreneur: A new era of entrepreneurial
education is beginning. American Journal of Small Business: 37-53.
Solomon, G.T., and Femald, L.W 199 1. Trends in small business management and
entrepreneurship education in the United States. Entrepreneurship: Theory and
Practice 15(3): 25-40.
Stevenson, H.H. 1983. Who are the Harvard self-employed. In J.A. Homaday, J.A.
Timmons, and K.H. Vesper, eds., Frontiers of Entrepreneurship Research.
Wellesley, MA: Babson College, pp. 233-254.
Stoner, C.R., Hartman, R.L., and Arora, R. 1990. Work-home role conflict in female
owners of small business: An exploratory study. Journal of Small Business
Management 28(1):30-38.
Thompson, P. Winter 1986. Characteristics of the small business entrepreneur in Canada.
Journal of Small Business and Entrepreneurship: 5-l 1.
Vesper, K.H. 1985. Entrepreneurship Education 1985. Wellesley, MA: Center for
Entrepreneurial Studies.
Watkins, D., and Morris, J.F. 1981. UK training and the development of entrepreneurs. In
K. Vesper, eds., Frontiers of Entrepreneurship Research, Welleley, MA: Babsone
College, pp. 413 – 427.
19
Journal of Business Venturing, Volume 9, Number 2, March 1994
