CHILD POVERTY AND INEQUALITY: THE WAY FORWARD

coefficient of the distribution of income (standardized in terms of

Gini of household disposable income per capita)^55.

Table 10. Definition, description and data sources of the variables used in
regression analysis
Variable name Variable label Source Unit of Measure
Gini income Gini coefficient of the current
distribution of disposable
household income per capita

SEDLAC

complemented

by WIID

Percentage points

Gini income Gini coefficient of the
distribution of disposable
household income per capita
in 1990

SEDLAC

complemented

by WIID

Percentage points

1990

GDP/c gr Per capita average annual
growth rates GDP in constant
prices

ECLAC Percentage based
on US dollar
figures at constant
2000 prices
Gini education Gini index of the distribution
of years of education among
the working population (25- 64
years old)

SEDLAC Percentage points

Tot- fob International terms of trade,
fob

ECLAC Index, 2000=100

Remittances Workers' remittances / GDP UNCTAD Percentage of
GDP
FDI Net Stock of Foreign Direct
Investment/GDP

UNCTAD Percentage of

GDP

(^55) Of the 324 cells on current income inequality, 175 are filled with SEDLAC
data, 11 from WIDER’s WIID2c (of these 1 is taken from Szekely (2003), 3 from
Gasparini (2003), 3 from (SEDLAC 2006), 1 from Deininger and Squire (2004), 2
from Szekely and Hillgert (2002), 3 from Badeinso-Eclac (2008), 13 from WDI
(2007), 1 (Argentina 2007) from national sources. 98 data-points were
interpolated by filling gaps of 1-2 years part of stable time series. In 3 cases the
interpolation filled gaps of years, and in 3 cases of 4 years, especially for the early
1990s. 23 cells (for Ecuador, Guatemala, Nicaragua, and Paraguay in the early
1990s) are blank. In most cases, data refer to disposable household income per
capita. A successful check was carried out to ensure that the trend of the data
filled in by interpolation replicated the trend of other income concepts. While in
most cases it was possible to ascertain that the data referred to disposable
income, lack of information in survey questionnaires did not allow identification
of the income concept used. This might introduce a measurement error in the
dependent variable. However, in view of the strong co-variance of the Gini’s for
all income concepts, it is likely that including data referring to an unknown
income concept may bias the country intercepts in the fixed effect estimation,
without affecting the parameters of the explanatory variables.