Click here to return to the report index
Mapping Indicators of Poverty in West Africa

1. Background

GRID-Arendal was approached by TAC for assistance in the production of reliable statistical and cartographic products on poverty and land use potential. These activities are integrated into an on-going collaborative project between UNEP and the CGIAR, 'Use of GIS in Agricultural Research'.

The poverty mapping activities have been conducted in a relatively short period (March - May 1997), starting with a small workshop discussing scope, data sources and outputs of the study. Invaluable inputs to these preliminary activities were given by Jake Brunner of the World Resources Institute and Uwe Deichmann from the UN Statistical Division. GIS analysis, interpretation and presentation activities were carried out at GRID-Arendal by the authors of the study.


2. Objectives

Generation of reliable statistical and cartographic products to communicate the relationship between rural poverty and land use potential in West Africa, in order to provide information to ensure optimal use of research investment.

Furthermore, the project served as a pilot study to investigate an appropriate approach to identify the location of poor people on a global basis with a reasonable investment of time and resources.


3. Approach

Measuring poverty requires taking into account a variety of factors. A standard indicator frequently used is the pure economic value 'GNP per capita'. Using this attribute makes sense if the study is to be comparable at a global level. However, GNP data is not available universally at the desired accuracy. We therefore chose to represent poverty - following the example of a WRI study in progress entitled 'Human Development versus Aridity in West Africa' - using the indicator variables of the Human Development Index (HDI) (UNDP, 1996)1. The actual data comes from the Health and Demographic Survey (HDS)1.

Because none of the HDI indicators are explicitly captured in that survey, surrogate variables are used from the DHS data to infer poverty levels2. The four surrogate variables used in this study are: Child Mortality, Adult Female Literacy, Primary School Enrolment, and Children with Stunted Growth. These data exist for 2263 sample points within West Africa3. In a second step, we excluded all urban samples to get a better idea of the situation of the rural poor (1113 samples).

As with 'poverty', the term 'marginal land', in terms of potential for crop production, has to be approximated in a most reasonable way, since there is no obvious, general and applied definition of it. To detect possible influences of spatial factors on the degree of human development/poverty, the data was combined with 4 different approximations of 'Marginal Land'4. Those can be grouped in two categories:

- Biophysical:

- Agroclimatic Zones
- Land Degradation

- Socio-economic:

- Population Density
- Accessibility to Infrastructure and Roads


The first three factors were analysed for all West-African countries, whereas the accessibility data was only available for Burkina Faso and Mali. The accessibility data was developed under the project. The definitions of the above factors can be found in the appendix.

When representing the data in maps and graphics, we chose to display each surrogate variable for human development separately rather than combine them in the suggested HDI, to keep the process more transparent.

The correlation of the HDI surrogates with the marginal condition factors was carried out using a GIS. GIS and statistical processing included point and polygon overlay analysis, using Arc/Info and Arc/View software. For each HDI sample point, a geographically referenced value was extracted from each thematic layer. An average and standard error was then calculated for each surrogate variable by thematic classes.


4. Results

The presentation of results consists of:

  • a map for each indicator for human development (representing an approximation of poverty) by thematic class,
  • the corresponding graphs displaying the HDI indicators in correlation with the background data, and
  • a general interpretation of the results.

Notes:

  1. See appendix 2 and 3 for detailed definition

  2. Although poverty levels were not measured directly, the terms 'Human Development' and 'Poverty' are used interchangeably in the text.

  3. The Human Development Survey was conducted in the following countries: Burkina Faso, Cameroon, Central African Republic, Cote d'Ivoire, Ghana, Liberia, Libya, Mali, Niger, Nigeria, Togo, Zaïre.

  4. We explicitly avoid giving absolute indications of what is more marginal. This can vary from factor to factor and is therefore left for interpretation by the reader.