Page 231 - Contributed Paper Session (CPS) - Volume 1

CPS1284 Rabeh M.
in that they entail a set of assumptions and computational issues (Fortin,
Lemieux, & Firpo, 2010). In this regard, the Recentered Influence Function (RIF)
regression approach recently suggested by Firpo, Fortin, and Lemieux (2009)
addresses these weaknesses and provides a straightforward regression-based
method for performing a detailed decomposition of some distributional
statistics such as quantiles, variance, and other statistics. The RIF is the key
concept of the unconditional quantile regression, the recently widely used
method of decomposition in the recent literature.
For this analysis, RIF (, ) is the function of explanatory variables:

(RIF(, )|X) = (3)

Where is the ℎ quantile and is the vector of parameters associated to

. Because RIF( , ) is unobserved in practice, we use the estimated

equation:
̂
( , ̂ ) = ̂ + −1( ≤ ̂ ) (4)

̂ ( ̂ )

Where is the estimated marginal density function of Y and I is an indicator
̂

function.
After estimating the model in equation (3) for the 10th(lowest percentile)
to 90th(highest percentile) quantiles of the population, we use the obtained
unconditional quantile regression estimates to decompose the different gaps
into a component attributable to differences in the distribution of
characteristics (composition effect) and a component due to differences in the
distribution of returns (wage structure) as follows:
̂
̂
̅̅̅̅̅
̅̅̅̅̅
̂ , − ̂ ̅ = ( , ̂ ) − ( ̅, ̂ ̅ ) = ( ̅ − ̅ ) ̅ + ̅ ( ̂ , − ̅ ) (5)
̅
̅
,

,
,

,
It is noteworthy that this RIF-based decomposition permits, after
computing both the composition effect and discrimination effect throughout
the wage distribution, to divide up the two effects into the contribution of
each explanatory variable. Moreover, the issue resulting from the use of
categorical predictors can also be straightforwardly resolved using the Yun's
method (2005) of normalization.
The empirical analysis is based on secondary data from the 2016
Palestinian Labor Force Survey (PLFS) that is prepared by the Palestinian
Central Bureau of Statistics (PCBS). PLFS is available on an annual basis for
each year from 1995 to 2016. The Palestinian Labour Force Survey Programme
conducts surveys quarterly. The survey provides basic information on the
relative size and structure of the Palestinian labour force, and the components
of employment, unemployment and time related underemployment.

3. Results
The results of Oaxaca–Blinder decomposition in Table 1 reveals that on
average, non-refugees earn 17% more wages than their refugees counterparts.
The composition effect explained by differences in productivity characteristics
presents 8.01% of the mean wage gap, while the discrimination effect explains

220 | I S I W S C 2 0 1 9