1. GenderWageInequalityandwomen’s
self-employment
Magdalena Smyk Siri Terjesen Joanna Tyrowicz
FAME|GRAPE & Warsaw School of Economics Florida Atlantic University & NHH FAME|GRAPE & University of Warsaw & IZA
Motivation
According to a theory on entrepreneurship, labor
market inequality can lead to women being
more likely to become self-employed. How-
ever, it is difficult to verify this theory empiri-
cally due to various methodological challenges, and
the evidence on the relationship between gender
gaps in the labor market and the frequency of self-
employment is inconclusive.
We contribute to the existing literature by pre-
senting a framework that overcomes concep-
tual limitations of the theory. We then test this
framework empirically.
Analysis plan
We merge GWG estimates from labor market data (different for age groups and levels of education) with
the individual level GEM data.
1. We show the link between GWG and probability to enter self-employment among women and
men, separately.
2. We provide a counterfactual shares of self-employed women based on estimated parameters from
a regression for men.
3. We show differences in the link between counterfactual and factual shares of self-employed women.
Gender wage gaps and share of self-employed by gender
Estimates are adjusted for the effects of higher education, access to en-
trepreneurial network and capital, age, age squared, country and year effects.
Linear function coefficients are based on the OLS regression.
=⇒ positive link between
GWG and (early stage) self-
employment among men,
and among women
=⇒ necessity self-employment
versus wage-employed: steeper
slope for men than for women
=⇒ necessity self-employment
other self-employed: positive
slope for men, but negative for
women
Counterfactual and factual shares of self-employed women
Estimates are adjusted for the effects of higher education, access to en-
trepreneurial network and capital, age, age squared, country and year effects.
Linear function coefficients are based on the OLS regression. The counterfactual
averages are adjusted for the difference in constant term in regression for men
and for women (σm-σw) [levels of self-employment among women and men].
Comparison of slopes:
=⇒ steeper slope for actual
than counterfactual for early
stage self-employment
=⇒ fatter slope for actual
than counterfactual women for
necessity self-employment
=⇒ positive slope between
counterfactual share of necessity
self-employed women and GWG
Conclusions
• Women choose self-employment (SE) differently than men.
• GWG correlates positively with SE of women and men, but with necessity SE only for men.
• Gender-specific mechanisms of the SE choice among women reduce the effect of GWG.
Data
Gender Wage Gaps (GWG):
• individual-level labor market databases from
many sources;
• based on median wages of men and women;
• 36 countries (Argentina, France, Uruguay,
UK, among others);
• 10 years ( 2009-2019);
• five age groups;
• three education levels.
Global Entrepreneurship Monitor (GEM):
• over 580,000 individuals;
• four measures of entrepreneurial activity:
1. self-employed (versus wage-employed);
2. early stage self-employed (versus wage-
employed);
3. necessity self-employed (versus wage-
employed);
4. necessity self-employed (versus other
self-employed)
Empirical strategy
Step 1. Model of self-employment choice among
men in GEM:
SE(0/1)i = βm
0 + βm
i Xm
i + ϵi
Step 2. Prediction of the self-employment status
for women:
predict SE(0/1)i = ˆ
βm
0 + ˆ
βm
i Xf
i + ϵi
Step 3. Estimation of the set of equations:
SE(0/1)i = σm
+ σm
i Xm
i + γm
GWGi + ϵi
for every man in country c at time t.
SE(0/1)i = σf
+ σf
i Xf
i + γf
GWGi + ϵi
for every woman in country c at time t.
SE(0/1)i = σcf
+ σcf
i Xf
i + γcf
GWGi + ϵi
for every woman in country c at time t.
Acknowledgements
The funding received through Norwegian Financial Mecha-
nism 2014–2021 (grant 2019/34/H/HS4/00481, within the
GRIEG framework of National Science Center).