Page 169 - Contributed Paper Session (CPS) - Volume 1

CPS1254 Yu Jin et al.
departments at all levels, heads of enterprises and institutions, and heads of
scientific, educational, cultural and health institutions. The middle class
occupations include service personnel, production and processing personnel,
maintenance equipment personnel, planting and feeding personnel, informal
employees and other employment personnel difficult to classify; the lower
class occupations are farmers. For the level of parental education, in the
process of data collation, we found that most of the parents of the surveyed
individuals have a low level of education, and very few have received high
school or higher education. Based on this, the education level of parents is re-
classified. Firstly, the information-deficient individuals are deleted, and then
the education level is divided into three categories: high, middle and low by
analogy with the classification of occupation. The upper classifications include
secondary and higher education, the middle classifications are primary
education, and the lower classifications are non-formal education and
illiteracy. In addition to household registration, parental occupation and
parental education, gender isalso included as an important "environmental"
factor in the analysis.

a. Measurement methods
The tools used in the existing research on measure inequality of
opportunity may be different, but the ultimate aim is to work on the inequality
index of the Computer Association (Bourguigonon, 2003; Ferreria and
Gignoux, 2008; Checchi and Peragine, 2010), thus comparing the inequality of
opportunities in different regions or different exogenic conditions. According
to different tools, measurement methods can be divided into parameter
method and no-parameter method. However, both parametric and no-
parametric methods first need to separate the functions of "environment" and
"effort" by constructing a virtual counterfactual distribution, and then calculate
the inequality index of opportunity.
Bourguigonon (2007),Ferreira and Gignous (2011) put forward the
parameter method. They suggest that the logarithm of income be set in linear
form and the least squares method be used to estimate (1) and (2). The
construction of is as follows:

1
= {exp ( )1 } … … , {exp ( )1 } （1）
1

Where is the fitted value of the parameter, represents the number of

individuals in the -th class “environment”, and 1 represents the unit row

vector of the dimension. This method of constructing the counterfactual

distribution is called the parameter method.
Checchi and Peragine（2010）proposed a nonparametric method for
estimating :

1
= {( )1 , … , ( )1 } (2)
1
158 | I S I W S C 2 0 1 9