Census Dept. classifies GN divisions by poverty
Dr. Amara Satharasinghe
The Government of Sri Lanka places strong emphasis on poverty
alleviation as part of its overarching goal of national development.
Internationally, the Millennium Development Goals (MDGs) aim to reduce
extreme poverty by half by 2015.
In this worldwide concern to reduce poverty, the role of poverty
statistics cannot be overemphasized. A major problem at present is to
effectively reach the poor to address their right to access basic
services. This calls for poverty statistics for small areas as a basis
for prioritizing, planning and implementing location-efficient
development and poverty eradication projects.
In response, the Department of Census and Statistics (DCS) has
undertaken several location-specific poverty measurement efforts. The
Household Income and Expenditure Survey has been conducted once in four
years to estimate poverty indicators at national and district level.
In 2006, a special study was conducted by combining data from the
2002 Household Income and Expenditure Survey data with the 2001 Census
of Population and Housing data to estimate poverty levels at Divisional
Secretariat (DS) division level. In 2006, for the first time, official
poverty statistics at the DS division level were released and these were
useful to identify the poorest DS divisions for government's rural
development projects.
However, the DS level statistics may not be adequate to identify the
location and social characteristics of small population groups or
communities living in poverty to make effectively targeted
interventions. DCS therefore, conducted a study to develop a proxy
indicator to measure poverty at the smallest administrative level of
Grama Niladari (GN) divisions.
To provide a proxy measure of poverty at the level of GN divisions,
this study used nine indicators, which are correlated with poverty
measured by Headcount ratio i.e. percentage of population below poverty
line. These indicators reflect the type of energy sources used by
households for lighting and cooking, housing quality, employment status
of household heads, level of education of household heads and level of
education of household population over 20 years.
The indicators were selected for their high predictive power of
poverty at the DS Division level. The indicators selected for the study
were as follows:
a. Percentage of households using kerosene for lighting
b. Percentage of households using firewood for cooking.
c. Percentage of housing units not having permanent materials for
wall.
d. Percentage of housing units not having permanent materials for
floor.
e. Percentage of housing units not having permanent materials for
roof.
f. Percentage of household heads who have not passed G.C.E.A/L or
above.
g. Percentage of household heads who are not paid employees.
h. Percentage household members age 20 and above who have not passed
G.C.E. A/L and above examinations.
Statistically these nine indicators were combined into an index and
this index was labeled as Unsatisfied Basic Needs Index (UBNI). This
method avoids arbitrary indicator selection and the application of
arbitrary external weights, both common in many of the composite indices
currently used.
Statistically it was tested whether this index is related to the
headcount ratio, at DS division level and it was found that these two
indicators are strongly and positively correlated. An important feature
of this indicator is higher values of the UBNI indicates prevalence of
higher levels of poverty and vice versa.
As such, UBNI can be considered as a proxy measure of poverty. The
UBNI can be considered as an operational poverty assessment tool as it
is capable of measuring poverty quite well and it is easy to collect
data to compile this indicator.
The Unsatisfied Basic Needs Indicator, which can be considered as a
proxy measure of poverty, was then used to classify GN divisions into
five classes according to level of poverty within each district and the
resulting classification was presented in a set of maps.
Spatial distribution of other nine indicators selected for the study,
were also mapped to provide a visual comparison against the poverty
distribution. This classification was carried out at the GN division
level of each district separately so that within district variation of
proxy measure of poverty can be compared across districts.
To facilitate the comparison of spatial patterns, a standard
color-coding was used for all districts ranging from red for the highest
level of proxy measure of poverty to green for the lowest level. In the
same way maps were prepared for the nine indicators selected for the
study as well.
DCS recently released a report carrying these maps and data tables.
With the impetus provided by the Millennium Development Goals for
poverty reduction and the national focus on targeting interventions for
greater effectiveness in reaching the most marginalized, there is a
growing and strong demand for small area poverty data.
Lack of poverty data for small areas is a conspicuous and often
spoken about gap in our knowledge base.
This study provides an alternative measure of poverty levels from
census data, which is readily available for computing small area
statistics. The study successfully demonstrates that a set of
indicators, which capture basic manifestations of poverty, can be
statistically converted into a composite index (UBNI) which is highly
correlated with headcount ratio.
Therefore, UBNI can be used to identify small areas at high levels of
poverty. The data can be presented in maps for easy visual examination
of spatial distribution of poverty for small areas.
The data required for this index are available from Census of
Population and Housing and Household Income and Expenditure Survey and
can also be collected more easily and inexpensively.
A disadvantage of the method is that it does not provide information
on the direct level of poverty. However, in many cases, in the absence
of direct measures of poverty at small area levels, it is proxy measures
rather than direct poverty that is of concern for the interventions and
policies.
Poverty is an inherently relative concept, and the tool developed in
this paper is indeed aiming to provide a proxy indicator to measure
poverty across GN divisions. Therefore, this tool, allows evaluating at
low cost the poverty targeting efficiency of development projects.
(The writer is Deputy Director, Department of Census and Statistics).
|