We examine the question of the economic returns to immigrants engaged
in self-employment. Contradictory, reports in the literature concernin
g the superior or inferior earnings of the self-employed relative to w
age/salaried workers are related to the choice of functional form of t
he earnings equation. Based on samples of four entrepreneurial immigra
nts plus control samples of Blacks and Whites from the 1980 Public Use
Microdata Sample, we find large differences in the net effect of self
-employment, depending on the use of a linear (absolute dollar values)
or loglinear (relative returns) form. We examine various explanations
for the discrepancy and identify the role of outliers as significant.
The loglinear form fits the data better bur at the cost of obliterati
ng substantively important information, namely the preponderance of th
e self-employed among positive outliers. Effects of excluding the latt
er from the linear form and the theoretical and policy implications of
alternative specifications of the earnings equation are examined.