6.1.1 Imprecision
of Mapping Auxiliary Data
Mapping auxiliary data has the advantage that it can use and modify concepts and indicators developed for vulnerability and food security, build on their efforts in data compilation, and cover a wide geographic area at relative low costs. Mapping auxiliary data has the disadvantage that it generally provides a resolution too coarse for understanding causes and effects of poverty, and the mapped indicator may not necessarily measure poverty, especially when a more narrow economic definition of poverty is needed. Mapping auxiliary data could be a feasible approach for a global poverty map that tries to raise awareness about the spatial distribution of poverty. Such a map could start with the descriptive study on the location of the poor by the International Fund for Agricultural Development (Jazairy et al., 1992) and then combine digital maps with expert opinion, similar to the approach that produced the maps for the Global Assessment of Soil Degradation (Oldeman, 1991).
6.1.2 Poor
Data Quality and Few Worked Examples for Measuring Access
A major constraint on measuring access to markets and services is data quality. Very few GIS databases in developing countries are developed with modeling applications in mind. As a result, considerable effort is required to update and correct poorly structured databases. Sophisticated algorithms have been developed to calculate accessibility but because of a lack of readily available and vertically integrated GIS databases in many developing countries, there are few examples where access measures have been applied at a scale and with a level of accuracy that makes them useful in an operational setting.
6.1.3 Limited
Data Availability of Geo-Referenced Survey Data
Very few developing countries routinely collect data that can be used to reliably map poverty. The only multi-national effort to map survey data in Africa is USAID's WASAP. As a result of this project, all DHS are now being geo-referenced using GPS at a cost of $10 per cluster (including equipment purchase and training). Geo-referencing the clusters post-survey with the help of maps or gazetteers is more expensive.
A related problem is the lack of a standard set of village names that allow survey data to be automatically joined to geo-referenced census data. The need to develop such "core" or "foundation" databases is a recurrent theme and one that the CGIAR is perhaps uniquely positioned to address, and benefit from. However, with a few notable exceptions (e.g., Corbett et al., 1996), the CGIAR has not produced publicly available, internationally-comparable GIS databases useful for poverty mapping.
6.1.4 Mapping
Modeled Results - High Costs and Institutional Barriers
The approach described for Burkina Faso requires intensive investments in digital data, vertical data integration, and modeling, which will be very expensive and difficult to implement for all developing countries. Such GIS work can be very time consuming with many potential setbacks during the location and identification of existing data sets, error checking and correction of data, and final integration of the assembled information which should permit spatial analyses, for example network analysis. In Burkina Faso, the GIS team encountered problems with data documentation, delays in obtaining some of the requested data sets, and unavailability of some data at disaggregated level.
The case study for Ecuador benefited from generous collaboration between national and international organizations and access to census data at the household level. The Burkina Faso example required collaboration and data from various government agencies. Poverty assessments and poverty maps with high spatial resolution have to overcome institutional rivalries and a natural reluctance of organizations to release data at disaggregated level. Reasons include legitimate concerns of data confidentiality, high access fees, and institutional inertia. High resolution poverty maps can become politically sensitive outputs, especially when they highlight the arbitrariness of previous decision making or become the basis for entitlements or social sector spending.
Detailed modeling of poverty estimates at the village or community level is most appropriate for narrow geographic targeting and for studying and understanding the complex relationships between land use, environment, and poverty. It will require close collaboration with national organizations and demands institutional and technical capacity within collaborating organizations to carry out complex quantitative analyses and modeling.
6.1.5 Correlation
versus Causation
Knowing where the poor live provides no information about why they are poor. Studies have shown that causes of poverty may differ from factors leading to its spatial concentration. The concentration of the poor generally results from a combination of structural and individual factors. The degree to which geographic (structural) or individual factors are causing poverty has implications for developing CGIAR's strategy of agricultural research. If geographic factors play an important role, then geographic targeting of agricultural research to the poor in marginal areas can become a useful tool to address poverty issues. If individual characteristics explain most of the local poverty, and individuals are free to migrate, then the mobility of people and capital can limit the success of targeting marginal areas geographically.
Implementing CGIAR's objective of poverty alleviation will require a critical examination of where and why poverty occurs. An international database of subnational poverty maps is not readily available and existing activities are incomplete geographically, too coarse to provide meaningful information, or measure concepts that are not of direct relevance to the objectives of the CGIAR centers. The question of how to build on these existing efforts, make additional investments, and develop a strategy for poverty mapping that benefits the CGIAR could be addressed in a workshop that brings together the CGIAR community, donor agencies, and other institutions interested in or working on poverty issues, both at the national and international level. At this workshop the following issues need to be resolved:
- What are the purposes
and applications of the poverty maps?
Possible applications include identifying areas of need, making decisions on regional priorities, defining agricultural research priorities, targeting interventions and resources, understanding the relationships between land use, environment, and poverty, and monitoring project and program impacts. The CGIAR-System as a whole will need poverty maps to set agricultural research priorities for specific commodities and regions in a way that benefits the poor and ultimately reduces the number of poor and the severity of poverty. Specific CGIAR-Centers will need poverty maps that help them design and implement programs that maximize benefits to the poor by increasing productivity of existing farming systems, introducing new crops and animals, and reforming fiscal, investment, and trade policies. The selected applications in turn will determine the appropriate scale or resolution of the mapping efforts.
- Which conceptual framework
and definition of well-being is the most appropriate for these
applications?
Based on the above a conceptual framework, definition of well-being, and comparable indicators can be selected.
- What is the most appropriate
level of effort?
Poverty mapping can be a quick desktop exercise or a much larger investment in primary data collection, database construction, and development of statistical methods. The CGIAR community may decide to tackle just one issue, for example a global poverty map, or agree upon a multi-pronged approach. This could include producing different maps that help to set research priorities, study causes and effects, or serve as a baseline for monitoring project impacts.
- Which inputs are needed
to produce these poverty maps?
The exercise needs to balance cost versus resolution and international versus local datasets. Increased contribution of local expertise and institutions will be required to analyze and understand poverty issues in a specific country.
CGIAR's follow-up activities to such a workshop can then move into several directions:
- Lead the work on medium
resolution demographic databases.
This would continue efforts by NCGIA, CIAT, FEWS, and FAO's Regional Remote Sensing Project in Southern Africa. Selected CGIAR centers that have strong GIS capacity could take the lead on developing internationally comparable medium resolution demographic databases and become the custodians for regional datasets. An agreed upon regional set of subnational administrative boundaries will provide a common framework for analysis into which poverty related data can be integrated.
- Initiate and support the
geo-referencing of survey data like the multi-national WASAP effort.
Geo-referencing of survey data with a GPS should become a standard practice for all household surveys. Benefits of geo-referencing far outweigh the costs, which are minor compared to the overall data collection expenditures.
- Develop a coordinated
research plan for detailed case studies on poverty, environment,
and land use.
This could build on activities started by CIAT's Hillsides Program and the World Bank's Burkina Faso study. Case studies should expand on efforts of participatory poverty assessment and further examine the causes for spatial clustering of the poor.
- Initiate and support efforts
to produce consistent databases of poverty and human welfare for
large geographic regions.
Of all the variables used in the Nelson et al. study, rural poverty estimates were at the coarsest resolution. Thus producing a more detailed geo-referenced database of poverty and human welfare indicators would be the first step to improve on the study's estimates.
This could be accomplished with the help of a long-term project that compiles a comprehensive database of poverty and human welfare indicators at subnational reference units. Its first objective could be producing a database at first subnational level (states, provinces, etc.) for a continent (as has been considered by CIAT). Building such an international database would require a common agreed upon framework of administrative boundaries and involve different international, regional, and national collaborators. Eventually, this effort could be expanded to include data at second and third subnational levels.
Different scenarios can be envisioned to produce such subnational poverty maps which cover both economic and non-economic measures of poverty and interim and long-term products. Interim products will require fewer resources but will be incomplete and coarser in coverage. Interim products could be summarized in a world atlas of poverty:
Economic measures of poverty
The objective is to produce subnational poverty maps that show consumption or income-based poverty indicators. A close collaboration between the CGIAR-System and the World Bank should be encouraged, since the World Bank has made significant investments in data collection, poverty assessments, and methodological research. A first, interim product could be the compilation of all subnational poverty estimates and maps of corresponding boundaries from the World Bank poverty assessments. Ultimately, efforts to map economic measures of poverty, could produce the following two long-term products: (1) High resolution poverty maps based on integration of household survey and census data. This would be a continuation of efforts carried out by the World Bank in Ecuador and could start with countries where a recent LSMS survey has been completed and disaggregated census data can be made available. (2) High resolution poverty maps at the village or community level using a GIS and vertically integrated data from surveys, census, and maps. This would expand on the World Bank's Burkina Faso work and require close collaboration with national institutions and data custodians.
Non-economic measures of poverty
The objective is to produce subnational poverty maps of non-income-based poverty indicators along the lines of UNDP's Human Development Indexes (HDI) and Human Poverty Indexes (HPI). A close collaboration between the CGIAR-System and UNDP should be encouraged, since UNDP in collaboration with national agencies has produced subnational HDI and HPI estimates for selected countries. A first interim product could be the compilation of all existing subnational HDI and HPI estimates and corresponding maps of boundaries. Ultimately two long-term products can be envisioned: (1) A global map that shows subnational HDI and HPI for all countries. (2) HDI and HPI-type measures based on geo-referenced survey data. The latter assumes continuation of geo-referencing of DHS data at the cluster level and may even require post-survey geo-referencing for selected countries. These indicators could be displayed spatially at the cluster level or aggregated to new units of analysis.
- Support efforts for increased
geo-referencing of agricultural, population, and housing censuses.
This includes production and availability of digital maps showing boundaries of enumeration areas and a consistent coding system.
- Participate in efforts
to improve coordination in data collection and dissemination of
household surveys from different sectors.
An interagency working group between UNICEF, UNDP, and the Worldbank is currently trying to improve the relevance and cost-effectiveness of survey methods and encourage cooperation among agencies on data collection and analysis (Wanmali, 1997). The benefits of a spatial framework should be emphasized in such collaborative efforts which can help to reduce duplication in data collection efforts and produce more useful data products. For example, many surveys include a community module that elicits answers about community conditions from key informants. Community characteristics could be easily geo-referenced or even obtained from spatial analysis, thus making the repeated inquiry from different surveys less necessary. If geo-referencing of survey data becomes a common practice, sampling designs for national surveys could be slightly modified (without increasing cost of data collection) to make it easier to integrate surveys from different sectors or produce useful data for trend analyses. It will also lead to the development of databases with better spatial coverage over time.
- Encourage close collaboration
between any international efforts to produce poverty maps and
food security and vulnerability maps.
It becomes obvious from a review of existing food security and vulnerability maps that robust subnational poverty maps are a fundamental part of any food security assessment (at a minimum to capture long-term or chronic vulnerability). Current discussions by FAO to produce the Food Security and Vulnerability Information Mapping System (FIVIMS) should closely collaborate with any efforts to produce international poverty maps (Devereux, 1997).
[PREVIOUS] [CONTENTS] [NEXT]