|
| Map
7a. |
Population
below the poverty line (%): Technical Advisory Committee (TAC )
definition |
| Map 7b. |
Population below the poverty line (%): World Bank rural poverty rates |
| Map 7c. |
Population
below the poverty line (%): less than US$ 1 day-1 |
| Map 7d. |
Population
below the poverty line (%): less than US$ 2 day-1 |
| Table 4. |
Comparison of
absolute and relative numbers of poor by region, based on different poverty line threshold definitions |
| Table
5. |
Poverty rates by region/country: different poverty line threshold
definitions |
| Table
6a. |
Number of poor by region/country and production system: TAC-defined poverty
threshold |
| Table
6b. |
Number of poor by region/country and production system: World Bank rural
poverty threshold |
| Table 6c. |
Number of poor by region/country and production
system: less than US$ 1 day-1
poverty threshold |
| Table
6d. |
Number of poor by region/country and production
system: less than US$ 2
day-1 poverty threshold |
It
is now generally agreed that human well-being has many dimensions, and that
poverty can be defined as a pronounced deprivation in well-being. It means
lacking food, shelter and clothing; being sick and having very limited or no
access to health services; being illiterate and having few or no educational
opportunities; having little security, and being very vulnerable to outside
events such as natural disasters and economic crises; being excluded from
power and political access; and, most of all, having no hope for the future.
There is no single indicator to measure all these dimensions
of poverty simultaneously. Efforts to measure human well-being have
thus concentrated on collecting data separately for some of these
dimensions—for example, with the help of data on income or
consumption parameters to measure material deprivation, and with
indicators of health, nutrition and education to capture low levels
of achievement in health and education.
Producing a map that shows the location of the poor in the world must
rely on the results from
these national and international data-collection efforts. However, current
investments in data collection and methodology development for statistical
estimation and mapping
techniques have not been sufficient to produce a global map at a
resolution that goes significantly below the national average.
International data collection that captures the income/consumption,
demographic/health and nutrition dimensions of human well-being have
probably received a larger share of investment and international
coordination than other dimensions (see WHO, 2001, for example).
Even these areas, which have received significant attention by
international and national agencies in the past two decades, have
produced results with severe limitations for our ability to show
where the poor are located on a global level. Some of these
limitations stem from a lack of international comparability of
country surveys. For example, with regard to household
consumption/expenditure measurements we need to overcome differences
in survey design and questions asked, such as different recall
periods for capturing spending on food, or how to make adjustments
for household size, different poverty lines and measurement errors.
Other limitations are a result of differences in coverage. For
example, 15% of the world’s population had only one household income
or expenditure survey over the past decade, and thus no trend
analysis is possible (World Bank, 2001). The most important
limitation, however, lies in the resolution of the data. The typical
sample size in these surveys is designed to produce representative
national statistics, with a breakdown in a handful of units of
analysis, such as estimates for urban and rural areas in three to
five major regions.
On a more optimistic note, various efforts are under way, both from
the demand and the supply side, that are advancing the development
of poverty maps, which could make a global, high-resolution poverty
map a reality within a few years. International and national
development agencies are increasingly interested in focusing
development efforts on the poor. Thus, an increasing demand for
appropriate maps could help to shape prioritisation efforts that go
beyond country rankings, improve geographic targeting, and
illuminate the cause-and-effect relationships between poverty and
other dimensions of development, such as environmental and health
outcomes.
On the supply side, three major developments are driving the
process.
-
Increased availability of geo-referenced, especially socio-economic, data. More spatial
data are becoming available because of falling costs of digital mapping
software and remote-sensing products and, most importantly, because of the
available power and convenience of data integration once a geographic
location has been assigned.
Over
the past 5 years, international efforts have improved the
availability of digital census data by administrativeunits
(CIESIN, 2000), and of map layers that are relevant for delineating malaria
risk (MARA, 2001), for example.
-
Efforts
to distribute survey data with assigned geographic locations.
An
example of this was a regional pilot project that assigned latitudes
andlongitudes to more than 2000 enumeration areas
('clusters') for 12 different demographic and health surveys carried out in the 1990s and that now allows the calculation of
reliable estimates for new units of analysis such as agro-ecological
zones.3
Geo-referencing clusters with the help of GPS units is now a standard
practice for most demographic and health surveys. Another example is the
efforts byvarious UN agencies that are compiling and distributing
past and future nutrition surveys over the Internet. The
current online version (WHO, 2001) lists all surveys for a country
falling within specific quality criteria, with
corresponding sample size and general location information, which could
easily be linked to a gazetteer and then presented in map
format.
-
Efforts
to develop and refine statistical techniques that combine census and
survey data to produce maps that go beyond the resolution
permitted by the original sample size of the survey. Efforts by
the research department in the World Bank and experts in
universities have greatly improved modelling techniquesfor small-area
estimation, that has led to higher-resolution poverty maps for South
Africa, Ecuador, Nicaragua and Panama (e.g., Statistics South
Africa, 2000;Elbers and Lanjouw, 2001, see also Appendix
4).
While work in the these areas is increasing the supply of data, methods and
maps, those efforts are stilldriven mostly by individual research interests,
the entrepreneurial spirit of task managers and ad hoc data compilation and
integration efforts. There is a tremendous opportunity to accelerate these
activities andmove them beyond their research and pilot status to a
mainstream effort. However, this will requireincreased financial support and
a more coordinated strategy between development agencies,
internationaliinstitutions focusing on survey, mapping and analysis and
institutions responsible for national censuses, statistical
services and mapping.
Given the existing data constraints, we decided to base the following global
poverty maps on national-level poverty rates. Similarly, the tables with
summary statistics on the number of poor by country and production system
use the same national-level rates.
The
maps depict these national rates within the extent of the livestock
production system map developed in the previous section. Users should keep
in mind that no spatial differentiation has been made in poverty rates
between production systems within a country. Case studies and more detailed
country data show a higher incidence of poverty in sparsely populated and
remote areas (measured by the headcount, the percentage of poor living below
a poverty line), and sometimes in low-potential, marginal agricultural
areas. These spatial patterns, however, do not appear in other locations,
and not enough quantitative data yet exist to generalise over regions, or to
identify other general patterns. As these global maps refer to the poor
rather than to poor livestock keepers, we decided to explore further spatial
disaggregation by production system for the global map in Section 2.5, and
we carried out a higher-resolution analysis for East Africa based on more
detailed data from household surveys in Section 3.3.
The tables giving summary statistics are probably most robust for the
aggregation of the number of poor people at sub-continental level. While
these poverty rates probably underestimate the number of poor in
grassland/rangeland-based systems, the absolute number of poor should be
closer to reality for the mixed systems given the higher headcounts in these
systems.
Even this restriction on national-level data and our decision to use a
poverty measure based on household income and expenditure surveys (which is
the most common approach for measuring poverty and the most widely available
at international level) leaves significant room for variation in measures of
the relative and absolute number of poor. One major reason for these
differences in the number of poor is the choice of poverty-line level. The
poverty line is the threshold in income or consumption below which a
household is classified as poor.
Internationally comparable lines, such as the widely cited less than US$ 1 day-1
(actually, the most recent line is equal to less than US$ 1.08 day-1
using 1993 purchasing power parity [PPP] estimates) are useful for producing
continental and global totals. Data based on an international poverty line
thus show the number of people who cannot purchase a roughly similar basket
of commodities (World Bank, 2001). National poverty lines
are needed to capture the inter-country differences in economic and social
status and to assess progress at a national scale. Poverty lines differ
between countries (e.g. see the comparison of different poverty lines in
East Africa, Section 3.3) and even within countries, to reflect, for
example, differences in the cost of living between urban and rural areas.
The following maps and tables based on these rates do not have a common
reference point, so international comparisons should be made with caution.
We
used four different data sets and poverty lines, two international lines
(less than US$ 1 day-1 and less than US$ 2 day-1)
and two national lines, one from the ILRI priority-setting exercise based on
Technical Advisory Committee (TAC) of the Consultative Group on
International Agricultural Research (CGIAR) data (Gryseels et al., 1997) and
one for the rural population living below the rural poverty line, to compare
differences in the numbers of poor. Table 4 clearly indicates that an
international line of less than US$ 1 day-1 underestimates the
number of poor in North Africa and Central and South America, which
typically have set their national poverty lines closer to the less than US$
2 day-1 figure. The less than US$ 1 day-1 is closer to
the national poverty lines in low-income countries of SSA and SA. A level of
less than US$ 2 day-1 is closer to national poverty lines in
middle-income countries.
Calculations are based on data from the World Bank (2001) and the CGIAR
(1996) as adapted and summarised in Thornton et al. (2000). Poverty rates
(headcount) are based on the latest household survey, typically (but not
always) in the past 5–10 years. No adjustments were made to standardise to
a common base year, for example by applying estimated growth rates of per
capita private consumption from national accounts. Survey data did not exist
for all countries within each region. For countries where such data were not
available, a regional popu-lation-weighted average was estimated for each of
the four regions (Asia, CSA, SSA and WANA) and then applied to the countries
with no data. Numbers may not add up to the total because of rounding.
Table 5 shows the various poverty rates by region and country. For
Map
7a, Map 7b, Map
7c and Map 7d and Table 6a,
Table 6b, Table
6c and Table 6d, we used the TAC poverty rates (CGIAR, 1996) and
the rural poverty thresholds (as defined by each country) and less than US$
1 day-1 and less than US$ 2 day-1 rates from the World Bank
(2001). Total numbers of poor people by system, by country and by region are
shown in Table 6a, Table 6b,
Table
6c and Table 6d, and Map
7a, Map 7b, Map
7c and Map 7d show the percentage of the
population below the four poverty lines. As noted above, these maps depict
national rates within the extent of livestock production systems, and no
spatial differentiation has been made in poverty rates between production
systems within a country. Thus for each map, there is only one poverty rate
associated with each country. The blank areas seen in some countries
correspond to the ‘other’ legend category in Map 3 (i.e. areas defined
as not having livestock according to our production system classification).
The TAC and rural
country poverty rate maps (Map 7a
and Map 7b) are quite similar. The less than
US$1 day-1 map (Map 7c) shows that much of CSA, most of WANA and
parts of Asia are in the lowest category (0–15% of the population), while
high poverty rates are shown in much of SA and SSA. The less than US$2 day-1
map (Map 7d), by comparison, shows a general increases in all regions in the
percentage of the population below this poverty line, which is not
surprising given the fact that well over 60% of the population in Asia,
Africa and CSA as a whole fall in this category (and compare with the
figures in Table 4).
|
