Towards a Food Atlas for Sub-Saharan Africa: food availability, deficits and aid deliveries
The 'problems' of food aid
Even though food aid cannot offer the solution of the problems of Sub-Saharan Africa, it definitely has been contributing significantly to the alleviation of its most acute food shortages. Yet, as is well known (see Barrett and Maxwell, 2005; Clay and Stokke 2000), inappropriate use of the instrument can be detrimental as well. Oversupply of imported food aid may discourage local production and, through pressure on domestic prices, can make it even more difficult for local farmers to survive adverse weather conditions and the aftermath of calamities. Excessive domestic procurement may foster speculation by traders and cause price hikes that may be particularly pronounced in some regions in view of the limited capacity of trade and transport infrastructure to smoothen out local deficits. Furthermore, the quality standards and the administrative requirement on tracking and tracing of shipments may amplify this volatility by making it harder for local producers to compete with traders that are able to supply high quality in large quantities. To this the question is added whether food aid should be given in kind, in the form of the major commodity consumed by intended recipients or of some valuable commodity such as vegetable oil, or in cash.
The need for decision support tools
It becomes very clear that, combined with the urgency of recipient needs, the provision of food aid is a very subtle issue that calls for dedicated decision support tools. Ideally, such tools should, on the basis of the latest information about the current availability of staples on the fields and in stocks as well as about the nutritional status and the purchasing power of the recipient population, enable the agencies in charge to coordinate their actions along the chains of delivery and to anticipate the impact of the operations on local markets.
A model for the optimal delivery of food aid
Remarkably, despite the critical importance of adequate management of these flows for the livelihood of many, the empirical information collected has so far not been incorporated in such decision support tools. This has motivated research to develop a spatially explicit model for the optimal delivery of food that could, in principle, account for detailed high-resolution information on the prevailing food situation across Sub-Saharan Africa, and assist the agencies involved, from the donor down to the actual operations, in identifying pockets of food insecurity, and in suggesting sources for local procurement as opposed to foreign imports, while taking into consideration the trade and transport network available. The model resolution will be 5 arcminutes (about 10x10km), and distinguishes between locally produced and consumed food, commercially traded food and food aid. In every cell on the grid, consumers and producers decide whether to buy or sell food, depending on market clearing prices and purchasing power. Demand for commercial food and food aid leads to flows of these goods from grid to grid, with transportation costs determining the optimal sourcing and route to be followed.
Use of the model
This decision support tool is not meant as a substitute for ground knowledge on the logistics of actual deliveries or for the operational experience of the agencies in emergency operations. Rather, it intends to provide a bigger picture of the prevailing food situation, by accommodating detailed information on consumer needs, agro-ecological conditions, international and local trade and the capacities of the various delivery channels. Hence, one of the major strengths of the tool is the comprehensive integration of various data sources into a single framework, while the possibility of graphically reproducing the data by means of maps facilitates easy communication with policy makers and experts in the field.
The development of the model
In 2005, much efforts were spent on the development of the underlying database. It consists of data for each cell grid covering the whole of Sub-Saharan Africa, is compiled for the year 2000, on the basis of information from a vast array of sources, with possibilities for ready update to later years. These include maps (i.e. data at grid level), but also data at district, province, and country level. Dedicated software was developed to conduct these task that can manage the data from the original source down to the geographic maps. Here, we only highlight two elements of the database: the representation of infrastructure and transport costs in Sub-Saharan Africa, the estimation of consumption per capita and by cell.
Infrastructure and the delivery of food aid
Lack of adequate infrastructure is often mentioned as one of the primary problems of Africa. It impedes commercial trade in (food) crops, but also presents a serious problem when emergency aid has to be transported from import harbors to the people in need.
Mapping infrastructure
Hence, one of the major challenges within the project was the compilation of a single consistent map of infrastructure in Africa, including primary, secondary and tertiary roads, coastal sea routes, inland waterways and railways. Although road maps were available for the African continent in digital format, different definitions of road quality seemed to have been applied for different parts of Africa, leading to large differences in road densities that could not be explained by economic or historical reasons. Hence, use of other (hard-copy) maps was made to correct the initial digital maps for these definition differences. For railways, additional data were gathered on the actual operation of lines; for example in Angola, only a part of the available Benguela railway could be used because of destroyed bridges and landmines until its very recent reconstruction. Such information obviously is important, since in this case, simple representation of available physical infrastructure would lead to a serious overestimation of the possibilities for transport. The same holds for points where roads are crossing rivers or ferries have to be used for transport over lakes. Here, an assessment of the capacity of bridges and ferries had to be done, again to avoid overestimation of the transport possibilities.
Costs of transportation
Secondly, differing transport costs are associated to different types of infrastructure, and a second major task therefore was to translate infrastructure availability to cell-by-cell costs. Estimation of transport costs uses the infrastructure maps with different categories as point of departure, and corrects for the presence of landmines (ICBL, 2001), for general safety in a country as measured in the Aggregate Governance Indicators dataset (World Bank, 2006), and per capita GDP. Prohibitive costs at border crossings for countries at war are based on the Uppsala conflict database (Uppsala University, 2006). Maps 1 and 2 show the main corridors connecting Africa and the secondary roads on the continent.
Map 1. Primary roads in Africa                         
Map 2. Primary and secondary roads in Africa
Both maps highlight a number of characteristics of the African continent. First, far from being one integrated whole, different more or less integrated regions can be recognized, in particular West Africa, and South-East Africa, while the regions only have sparse connections with each other. Secondly, the density of corridors and secondary roads is very low in the heart of Africa (including the Democratic Republic of the Congo, Central African Republic, Angola) and equally low in the Sudan and Somalia. Many of these regions have been or are still largely dependent on food aid deliveries, and the lack of a well-developed infrastructure implies that costs of reaching the people in need are high. We return to this point in the box on refugees in Africa.
Estimation of consumption
The estimation of consumption is based on recorded weights of women and children in the Demographic and Health Surveys (DHS) carried out for USAID in the majority of the African Counties. To go from recorded weights to calorie intake, conversion formulas from weights to calorie from FAO (2004) were used, while updating of these results to the base year 2000 and imputation of missing countries was done using data on the Human Development Index ranking (UNDP, 2001). Map 3 shows the estimation of per capita calorie consumption for Sub-Saharan Africa thus calculated. The resulting estimates for consumption differ substantially from consumption estimates by FAO that are based on calorie availability. This result is consistent with the observation that the direct measurement of undernutrition through the weighing of people (as in the DHS surveys) leads to a substantially lower percentage of undernutrition than the official estimates by FAO (see Figure 6). We note that for children, the differences between the FAO/UNICEF estimates and DHS-based estimates of undernutrition are much less pronounced since, for children, FAO/UNICEF estimates are also based on direct weighing. In 2006, a scientific paper will appear in which the procedure for estimating the consumption is explained and defended.
Map 3. Consumption per capita in Kcal
Figure 6. Number of undernourished. Comparing FAO with DHS-based estimates
In 2006, an atlas with maps for Sub-Saharan Africa will be compiled to be distributed on CD-rom, consisting of ‘zoomable’ maps (i.e. zooming in on a particular spot until grid values become visible) supplemented with tables, and explanatory text. The Atlas is an intermediate milestone towards the completion of the full decision support tool and comprises three parts, namely the database, where the main steps in the process of database construction are highlighted and results shown; the calibrated model, which adds to the database the flows among sites and the price relationships induced by these flows; and indicators, where special indicators of interest to WFP will be elaborated upon. Follow-up applications could focus on a particular region or country groupings, making it possible to incorporate more detailed and reliable information. Scientific outputs will include background papers that describe the methodology of the consumption estimates used in the project.