Economists propose a series of theories to explicate the ground that favoritism in labour market produced and supplied several empirical methodological analysis to analyze favoritism job. This sort of research can be divided into two types: 1 is to prove the hypothesis of favoritism theory ; the other is to gauge the degree and determiners of favoritism.

3.1 Becker favoritism coefficient

Becker ( 1957 ) is the first to make the quantitative research on favoritism. He defined the favoritism coefficient as the difference between pay rate under favoritism and pay rate under non-discrimination.

D = Wn/Wd- ( Wn/Wd ) 0

Where N, 500 stand for different groups which are gain different intervention. n refer to the nepotism shown group while vitamin D refer to the favoritism shown group. ( Wn/Wd ) 0 is the pay rate between groups under non-discrimination. It equal to the ratio of fringy productiveness of this two group: MPn/MPd. If Wd & lt ; MPd, there is favoritism against 500 group. If Wn & gt ; MPn, there is nepotism shown to n group. So the entire favoritism coefficient consists of two parts: one is favoritism against 500 group ; the other is nepotism to n group. ( Xie Sisheng, 2005 ) .

Oaxaca ( 1973 ) extend Becker ‘s favoritism coefficient in term of per centum.

D = { Wn/Wd- ( Wn/Wd ) 0 } / ( Wn/Wd ) 0 ( 3.1 )

Where D is the coefficient of favoritism. Rewrite ( 3.1 ) after the common logarithm.

Ln ( D+1 ) =Ln ( Wn/Wd ) -Ln ( Wn/Wd ) 0

Harmonizing to OLS arrested development

and ( 3.2 )

Where N and make bold the mean value of productiveness feature of two group. I?n and I?d are the coefficients of ordinary least squares arrested development. Since the ( Wn/Wd ) 0 is the pay rate under non-discrimination, it can non be observed. So if the return rate to productiveness feature of two groups is equal, we regard this state of affairs as non-discrimination.

In this manner, we can acquire ln ( D+1 ) =n I”I? where I”I? = I?n – I?d

3.2 Oaxaca & A ; Blinder decomposition

Through the pay distribution of work forces and adult females, we can cipher the mean rewards of work forces and adult females, both of this two norm rewards are by and large non the same. Oaxaca ( 1973 ) & A ; Blinder ( 1973 ) usage Mincer ( 1970 ) pay equation arrested development on male and female sample to the factors which affect the mean pay differential between work forces and adult females.

( 3.3 )


=wage for single in group I

=control variable of productiveness features

=coefficient of the arrested development line

=error term which follow a normal distribution with discrepancy, one = m, degree Fahrenheit.

To take the control variables of pay equation, the first rule is to avoid the endogenous variables that contain favoritism. In other words, the control variables that step productiveness features can non reflect favoritism. These variable normally include old ages of instruction, work experience, current occupation term of office, matrimonial position and geographical location.

Harmonizing to pay equation, gender pay derived function can be expressed

( 3.4 )

Take a arrested development on the mean of pay equation and the last term equal to 0.

Plus and subtractions. We can acquire the equation relation to males.

( 3.5 )

The first term is differences in mean gifts comparative to males which is besides the explained portion of pay derived function. The 2nd term is portion of pay derived function caused by the difference in estimated coefficients. It is by and large regarded as unexplained portion of pay derived function which attribute to pay favoritism.

Neumark ( 1988 ) proposed that employers ‘ penchant for different type of worker is different and that if employers ‘ penchant for female and male labour force is homogenous of grade nothing, a discrimination-free pay construction can be estimated.

( 3.6 )

refer to non-discriminatory pay coefficient. The first term on the right side is portion of pay derived function which can be explained by discernible difference in features. The 2nd term is how much male is overestimated. Consequently, the 3rd term is how much female is underestimated. The 2nd and 3rd term can be interpreted as the pay differential attributed to favoritism.

As can non be observed, Neumark ( 1988 ) offered that take the leaden norm which combine male arrested development coefficient with female arrested development coefficient as index.

( 3.7 )

( 3.8 )

Where Ten is the observation matrix for the pooled sample and Xm is the observation

matrix for the male sample.

3.3 Sampling Method

The information we use in this thesis is from 2002 Chinese Household Income Project which was organized by Inter-University Consortium For Political And Social Research ( ICPSR ) and used to deduce Chinese single income distribution and the determiners that affect the income distribution in urban and rural countries of the People ‘s Republic of China. Taking into history regional difference among Chinese eastern, cardinal and western countries, the undertaking cover 12 states which are of different geographics and economic characteristics. The sample of probe is selected from National Bureau of Statistics Household Survey. And the questionnaire study is besides conducted by National Bureau of Statistics. This agreement guaranteed the continuity and scientificity of sample choice. Using the secondary sampling in a big sample, 20632 samples were collected through a series of questionnaire-biased interviews. This study concept family income harmonizing to international criterions. Using China ‘s official income estimation method, income and subsidy in sort, the value of household self-consumption, and the lease are non included into disposable income. So the definition of income construction harmonizing to international criterion is more close to the world.

3.3 Variable Choice

The dataset of urban single income contain 151 variables and 20632 instances. But for the intent of this thesis, we do the following procedure with the variable.

Wage: the pay we analyse in this article is the income from occupation. After excepting monetary value subsidy, local subsidy, life disbursals for the laid-off and subsidy for minimal living criterion from entire income, we can acquire the pay which worker truly derive from occupation.

Education: the instruction variable we chose in this article is the figure of old ages the respondents get educated. In the questionnaire, we investigated the figure of old ages that the participants get educated until 2002.

Work experience: from the information of study, we can cognize the first twelvemonth when the respondents started the first occupation. The twelvemonth of work experience is derived from the expression: work experience=2002-the first twelvemonth single Begin to work.

Occupation: in this study business was divided into 10 kindsi?sOwner ( director ) of private house ; Self-employed ; Professional ; Director of authorities agent, establishment and endeavor ; Department manager of authorities agent, establishment and endeavor ; Clerical/office staff ; Skilled worker ; Unskilled worker ; Salesclerk or Service worker ; Other




Owner ( director ) of private house






Department manager of authorities agent, establishment and endeavor


Director of authorities agent, establishment and endeavor


Clerical/office staff


Skilled worker


Unskilled worker


Salesclerk or Service worker